article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the distance between the trajectories of an asteroid ( orbiting around the sun ) and our planet gives a first indication in the search for possible earth impactors .
we call it orbit distance and denote it by @xmath3 . a necessary condition to have a very close approach or an impact with the earth is that @xmath3 is small .
provided close approaches with the planets are avoided , the perturbations caused by the earth make the asteroid trajectory change slowly with time .
moreover , the perturbations of the other planets produce small changes in both trajectories .
the value of the semimajor axis of both is kept constant up to the first order in the small parameters ( the ratio of the mass of each perturbing planet to the mass of the sun ) .
all these effects are responsible of a variation of @xmath3
. we can study the evolution of the asteroid in the framework of the restricted 3-body problem : sun , planet , asteroid .
then it is easy to include more than one perturbing planet in the model , in fact the potential energy can be written as sum of terms each depending on one planet only .
if the asteroid has a close encounter with some planet , the perturbation of the latter generically produces a change in the semimajor axis of the asteroid .
this can be estimated , and depends on the mass of the planet , the unperturbed planetocentric velocity of the small body and the impact parameter , see @xcite .
the orbits of near - earth asteroids ( neas , i.e. with perihelion distance @xmath4 au ) 149,597,870 km ] are chaotic , with short lyapounov times ( see @xcite ) , at most a few decades .
after that period has elapsed , an orbit computed by numerical integration and the true orbit of the asteroid are practically unrelated and we can not make reliable predictions on the position of the asteroid .
for this reason the averaging principle is applied to the equations of motion : it gives the average of the possible evolutions , which is useful in a statistical sense .
however , the dynamical evolution often forces the trajectory of a nea to cross that of the earth .
this produces a singularity in the averaged equations , where we take into account every possible position on the trajectories , including the collision configurations .
the problem of _ averaging on planet crossing orbits _ has been studied in @xcite for planets on circular coplanar orbits and then generalized in @xcite including nonzero eccentricities and inclinations of the planets .
the work in @xcite has been used to define proper elements for neas , that are integrals of an approximated problem , see @xcite . in this paper
we compute the main singular term by developing the distance between two points , one on the orbit of the earth and the other on that of the asteroid , at its minimum points .
this choice improves the results in @xcite , @xcite , where a development at the mutual nodes was used , because it avoids the artificial singularity occurring for vanishing mutual inclination of the two orbits .
moreover , we show that the averaged vector field admits two lipschitz - continuous extensions from both sides of the orbit crossing set ( see theorem [ der_jumps ] ) , which is useful for the numerical computation of the solutions .
the orbit distance @xmath3 is a singular function of the ( osculating ) orbital elements when the trajectories of the earth and the asteroid intersect .
however , by suitably choosing a sign for @xmath3 we obtain a map , denoted by @xmath5 , which is analytic in a neighborhood of most crossing configurations ( see @xcite ) . here
we prove that , near to crossing configurations , the averaged evolution of @xmath5 is more regular than the averaged evolution of the elements , which are piecewise differentiable functions of time .
the paper is organized as follows .
section [ s : orbdist ] contains some preliminary results on the orbit distance . in sections [ s :
avereq ] , [ s : extract ] , [ s : gensol ] we introduce the averaged equations , present the results on the singularity extraction method and give the definition of the generalized solutions , which go beyond crossing singularities . in section [ s : dminevol ]
we prove the regularity of the secular evolution of the orbit distance .
section [ s : numexp ] is devoted to numerical experiments : we describe the algorithm for the computation of the generalized solutions and compare the averaged evolution with the solutions of the full equations of motion .
we also show how this theory can be applied to estimate earth crossing times for neas .
let @xmath6 , @xmath7 be two sets of orbital elements of two celestial bodies on confocal keplerian orbits . here
@xmath8 describes the trajectory of the orbit and @xmath9 is a parameter along the trajectory , e.g. the true anomaly .
we denote by @xmath10 the two - orbit configuration , moreover we set @xmath11 . in this paper
we consider bounded trajectories only .
choose a reference frame , with origin in the common focus , and write @xmath12 , @xmath7 for the cartesian coordinates of the two bodies .
for a given two - orbit configuration @xmath13 , we introduce the _ keplerian distance function _
@xmath14 , defined by @xmath15where @xmath16 is the two - dimensional torus and @xmath17 is the euclidean norm .
the local minimum points of @xmath14 can be found by computing all the critical points of @xmath18 .
for this purpose in @xcite , @xcite , @xcite , @xcite the authors have used methods of computational algebra , such as resultants and grbner s bases , which allow us to compute efficiently all the solutions . apart from the case of two concentric coplanar circles , or two overlapping ellipses , the function @xmath18 has finitely many stationary points .
there exist configurations attaining 4 local minima of @xmath18 : this is thought to be the maximum possible , but a proof is not known yet .
a simple computation shows that , for non - overlapping trajectories , the number of crossing points is at most two , see @xcite .
let @xmath19 be a local minimum point of @xmath20 .
we consider the maps @xmath21for each choice of the two - orbit configuration @xmath13 , @xmath22 gives the orbit distance .
the maps @xmath23 and @xmath3 are singular at crossing configurations , and their derivatives do not exist . we can deal with this singularity and obtain analytic maps in a neighborhood of a crossing configuration @xmath24 by properly choosing a sign for these maps . we note that @xmath23 , @xmath3 can present singularities without orbit crossings .
the maps @xmath23 can have bifurcation singularities , so that the number of minimum points of @xmath14 may change .
therefore the maps @xmath23 , @xmath3 are defined only locally .
we say that a configuration @xmath13 is non - degenerate if all the critical points of the keplerian distance function are non - degenerate .
if @xmath13 is non - degenerate , there exists a neighborhood @xmath25 of @xmath26 such that the maps @xmath23 , restricted to @xmath27 , do not have bifurcations . on the other hand , the map @xmath3 can lose regularity when two local minima exchange their role as absolute minimum .
there are no additional singularities apart from those mentioned above .
the behavior of the maps @xmath23 , @xmath3 has been investigated in @xcite . however , a detailed analysis of the occurrence of bifurcations of stationary points and exchange of minima is still lacking . here
we summarize the procedure to deal with the crossing singularity of @xmath23 ; the procedure for @xmath3 is the same .
we consider the points on the two orbits corresponding to the local minimum points @xmath28 of @xmath18 : @xmath29 we introduce the vectors tangent to the trajectories @xmath30 at these points @xmath31 and their cross product @xmath32 we also define @xmath33 the vector @xmath34 joins the points attaining a local minimum of @xmath18 and @xmath35 . from the definition of critical points of @xmath18 both the vectors @xmath36 , @xmath37
are orthogonal to @xmath34 , so that @xmath38 and @xmath34 are parallel , see figure [ regrule ] .
denoting by @xmath39 , @xmath40 the corresponding unit vectors and by a dot the euclidean scalar product , the distance with sign @xmath41 is an analytic function in a neighborhood of most crossing configurations .
indeed , this smoothing procedure fails at crossing configurations such that @xmath36 , @xmath37 are parallel .
a detailed proof can be found in @xcite .
note that , to obtain regularity in a neighborhood of a crossing configuration , we lose continuity at the configurations with @xmath42 and @xmath43 .
the derivatives of @xmath44 with respect to each component @xmath45 , @xmath46 of @xmath13 are given by @xmath47 we shall call ( signed ) orbit distance the map @xmath5 .
let us consider a restricted 3-body problem with the sun , the earth and an asteroid .
the motion of the 2-body system sun - earth is a known function of time .
we denote by @xmath48 the heliocentric position of the asteroid and the planet respectively .
the equations of motion for the asteroid are @xmath49\ , , \label{restr3b}\ ] ] where @xmath50 is gauss constant and @xmath51 is a small parameter representing the ratio of the earth mass to the mass of the sun .
we study the motion using delaunay s elements @xmath52 , defined by @xmath53 where @xmath54 are keplerian elements , @xmath55 is the mean motion and @xmath56 is the time of passage at perihelion .
delaunay s elements of the earth are denoted by @xmath57 .
we write @xmath58 for the two - orbit configuration , where @xmath59 are delaunay s elements of the asteroid and the earth respectively . using the canonical variables @xmath60
, equations can be written in hamiltonian form as @xmath61 where we use @xmath62\ , , \hskip 1 cm n\in\n\,,\ ] ] for the symplectic identity matrix of order @xmath63 .
the hamiltonian @xmath64 is the difference of the two - body ( asteroid , sun ) hamiltonian @xmath65 and the perturbing function @xmath66 with @xmath67 considered as functions of @xmath68 .
the function @xmath69 is the sum of two terms : the first is the direct part of the perturbation , due to the attraction of the earth and singular at collisions with it .
the second is called indirect perturbation , and is due to the attraction of the sun on the earth .
we can reduce the number of degrees of freedom of by averaging over the fast angular variables @xmath70 , which are the mean anomalies of the asteroid and the earth . as a consequence
, @xmath71 becomes a cyclic variable , so that the semimajor axis @xmath72 is constant in this simplified dynamics . for a full account on averaging methods in celestial mechanics
see @xcite .
the averaged equations of motion for the asteroid are given by @xmath73 where @xmath74 , @xmath75 are some of delaunay s elements , and @xmath76 with @xmath77 , is the vector of the averaged partial derivatives of the perturbing function @xmath69 .
equation corresponds to the scalar equations @xmath78 we can easily include more planets in the model . in this case the perturbing function is sum of terms @xmath79 , each depending on the coordinates of the asteroid and the planet @xmath80 only , with a small parameter @xmath81 , representing the ratio of the mass of planet @xmath80 to the mass of the sun .
note that , if there are mean motion resonances of low order with the planets , then the solutions of the averaged equations may be not representative of the behavior of the corresponding components in the solutions of .
moreover , when the planets are assumed to move on circular coplanar orbits we obtain an integrable problem .
in fact the semimajor axis @xmath72 , the component @xmath82 of the angular momentum orthogonal to the invariable plane and the averaged hamiltonian @xmath83 are first integrals generically independent and in involution ( i.e. with vanishing poisson s brackets ) .
taking into account the eccentricity and the inclination of the planets the problem is not integrable any more .
in @xcite the secular evolution of high eccentricity and inclination asteroids is studied in a restricted 3-body problem , with jupiter on a circular orbit .
nevertheless , crossings with the perturbing planet are excluded in that work . in @xcite
there is a similar secular theory for a satellite of the earth .
the dynamical behavior described in @xcite , @xcite is usually called _ lidov - kozai mechanism _ in the literature and an explicit solution to the related equations is given in @xcite . if no orbit crossing occurs , by the theorem of differentiation under the integral sign the averaged equations of motion ( [ averrhs ] ) are equal to hamilton s equations @xmath84 where @xmath85 is the averaged perturbing function .
the average of the indirect term of @xmath69 is zero . when the orbit of the asteroid crosses that of the earth a singularity appears in , corresponding to the existence of a collision for particular values of the mean anomalies .
we study this singularity to define generalized solutions of which go beyond planet crossings .
since the semimajor axis of the asteroid is constant in the averaged dynamics , we expect that the generalized solutions can be reliable only if there are no close approaches with the planet in the dynamics of equations .
in the following we denote by @xmath24 a non - degenerate crossing configuration with only one crossing point , and we choose the minimum point index @xmath86 such that @xmath87 . for each @xmath13 in a neighborhood of @xmath88
we consider taylor s development of @xmath89 , @xmath90 , in a neighborhood of the local minimum point @xmath19 : @xmath91 where @xmath92 is the hessian matrix of @xmath18 in @xmath93 , and @xmath94 is taylor s remainder in the integral form
. is a multi - index , hence @xmath95 for a vector @xmath11 and a smooth function @xmath96 . ]
we introduce the approximated distance @xmath97 where @xmath98,\ ] ] and @xmath99 if the matrix @xmath100 is non - degenerate , then it is positive definite because @xmath101 is a minimum point , and this property holds in a suitably chosen neighborhood @xmath25 of @xmath24 . at a crossing configuration @xmath102
the matrix @xmath100 is degenerate if and only if the tangent vectors @xmath103 , @xmath104 are parallel ( see @xcite ) : @xmath105 first we estimate the remainder function @xmath106 . to this aim
we need the following : there exist positive constants @xmath107 , @xmath108 and a neighborhood @xmath109 of @xmath110 such that @xmath111 holds for @xmath112 in @xmath109 .
moreover , there exist positive constants @xmath113 , @xmath114 and a neighborhood @xmath25 of @xmath24 such that @xmath115 holds for @xmath13 in @xmath25 and for every @xmath116 . [ inequalities ] from , we obtain the existence of a neighborhood @xmath109 of @xmath117 and a constant @xmath118 such that @xmath119 we select @xmath109 so that no bifurcations of stationary points of @xmath18 occur and there exists a constant @xmath120 with @xmath121 for each @xmath122 . relation together with , yield for some @xmath123 . moreover , we can find a neighborhood @xmath25 of @xmath24 such that there are no bifurcations of stationary points of @xmath18 , and the inequalities hold for some @xmath124 : in fact @xmath125 depends continuously on @xmath13 and @xmath126 is positive definite .
there exist @xmath127 and a neighborhood @xmath25 of @xmath24 such that @xmath128 where @xmath129 with @xmath130 . [ diff_bounded ] by lemma [ inequalities ] we can choose two neighborhoods @xmath25 , @xmath131 of @xmath24 and @xmath132 respectively such that both and hold in @xmath133 .
we restrict @xmath25 , if necessary , so that there exists @xmath134 with @xmath135 for each @xmath136 ( see figure [ choosew ] ) . in @xmath137 we have @xmath138 } \leq \frac{1}{\sqrt{c_1}[1+\sqrt{c_1}]}\frac{\vert \delta_h^2 - d^2\vert}{\delta_h^3 } \leq c\ ] ] for a constant @xmath127 . using the boundedness of @xmath139 , @xmath140 in @xmath141 we conclude the proof .
now we estimate the derivatives of the remainder function @xmath142 .
there exist @xmath127 and a neighborhood @xmath25 of @xmath24 such that , if @xmath143 is a component of delaunay s elements @xmath144 , the estimate @xmath145 holds for each @xmath146 , with @xmath147 as in proposition [ diff_bounded ] .
therefore the map @xmath148 where @xmath149 , can be extended continuously to the whole set @xmath25 .
[ extend_rem ] in the following we denote by @xmath150 some positive constants .
we write @xmath151 and give an estimate for the two terms at the right hand side .
we choose a neighborhood @xmath152 of @xmath117 as in proposition [ diff_bounded ] so that , using , and the boundedness of the remainder function , we have @xmath153 in @xmath154 .
moreover in @xmath155 we have @xmath156 since @xmath157 and the derivatives @xmath158^{-1}\frac{\partial}{\partial y_k}\nabla_v d^2(\cale , v_h(\cale))\ ] ] are uniformly bounded for @xmath159 since bifurcations do not occur .
hence the relation @xmath160 holds in @xmath155 , with @xmath161 .
we also have @xmath162 for @xmath163 , in fact @xmath164 for each @xmath165 with @xmath166 . using again
, we obtain @xmath167 from , we obtain and the assert of the proposition follows using the boundedness of @xmath168 in @xmath141 . from in proposition [ extend_rem ] the average over @xmath16 of the derivatives of @xmath106 in
is finite for each @xmath13 in @xmath25 , and can be computed by exchanging the integral and differential operators : therefore the average of the remainder function is continuously differentiable in @xmath25 . on the other hand , the average over @xmath16 of the derivatives with respect to @xmath143 of @xmath140 are non - convergent integrals for @xmath169 : for this reason the averaged vector field in is not defined at orbit crossings .
next we show , exchanging again the integral and differential operators , that the average of these derivatives admit two analytic extensions to the whole @xmath25 from both sides of the singular set @xmath170 . for this purpose , given a neighborhood @xmath25 of @xmath24 , we set @xmath171 with @xmath44 given by .
there exists a neighborhood @xmath25 of @xmath24 such that the maps @xmath172 where @xmath143 is a component of delaunay s elements @xmath144 , can be extended to two different analytic maps @xmath173 , defined in @xmath25 .
[ extend_approx ] we choose @xmath25 as in proposition [ extend_rem ] and , if necessary , we restrict this neighborhood by requiring that @xmath174 in @xmath25 , so that @xmath175 is analytic . to investigate the behavior close to the singularity , for each @xmath159 , we can restrict the integrals to the domain @xmath176 for a suitable @xmath177 . by using the coordinate change @xmath178 and
then polar coordinates @xmath179 , defined by @xmath180 , we have rcl _ ddd & = & _ b d + & = & _ 0^rd= ( - d_h ) , with @xmath181 .
the term @xmath182 is not differentiable at @xmath183 .
we set @xmath184 with @xmath44 as in , and define on @xmath25 the two analytic maps @xmath185 we observe that @xmath186 ( resp .
@xmath187 ) corresponds to the derivative of @xmath188 with respect to @xmath143 on @xmath189 ( resp .
@xmath190 ) .
now we state the main result .
the averages over @xmath16 of the derivatives of @xmath69 with respect to delaunay s elements @xmath143 can be extended to two lipschitz
continuous maps @xmath191 on a neighborhood @xmath25 of @xmath24 .
these maps , restricted to @xmath189 , @xmath190 respectively , correspond to @xmath192 . moreover the following relations hold : @xmath193\ , , \label{diff}\end{aligned}\ ] ] with the derivatives of @xmath44 given by .
[ der_jumps ] using the results of propositions [ extend_rem ] , [ extend_approx ] we define the extended maps by @xmath194\,,\ ] ] with @xmath195 given by .
we show that the maps @xmath196 are lipschitz continuous extensions to @xmath25 of @xmath192 .
the maps @xmath195 are analytic in @xmath25 , thus we only have to study the integrals @xmath197 . from proposition [ extend_rem ] we know that these maps are continuous .
we only need to investigate the behavior close to the singularity , therefore we restrict these integrals to the domain @xmath198 introduced in .
we prove that the maps @xmath199 with @xmath200 , are bounded .
first observe that the derivatives @xmath201 are bounded in @xmath202 , otherwise we could not find the analytic extensions @xmath173 introduced in proposition [ extend_approx ] . ]
where @xmath203 are bounded in @xmath204 , and @xmath205 is unbounded but cancels out in the difference . ]
then we show that the maps @xmath206 are bounded in @xmath207 . using , we write the integrand function in the right hand side of as the sum of @xmath208 and of terms of the following kind : @xmath209\ , , \hskip 0.5 cm -\frac{1}{2}\frac{{\cal p}_3^{(h)}}{\delta_h^3 } \frac{\partial^2
{ \cal r}_3^{(h)}}{\partial y_j\partial y_k}\ , , \label{easy_bound}\ ] ] @xmath210 @xmath211 the integrals over @xmath198 of the terms in are not bounded in @xmath207 , but their sum is bounded and corresponds to . in the following we denote by @xmath212 some positive constants .
moreover we use the relation @xmath213 and the developments @xmath214 \hskip 1 cm ( s=3,5)\ ] ] with @xmath215 @xmath216(\cale , v_h+t(v - v_h))\,dt\ .
\label{pbetas}\ ] ] by developing we obtain @xmath217(\cale , v_h+t(v - v_h))\,dt \\ & = & -\frac{s}{2}\int_0 ^ 1 d^\beta\biggl(\frac{{\cal r}_3^{(h)}}{\delta_h^2}\biggr)(\cale , v_h+t(v - v_h))\,dt + \mathfrak{r}_s^{(h)}(\cale , v)\,,\end{aligned}\ ] ] with @xmath218 . moreover , we have @xmath219 we can estimate the terms in as follows : @xmath220\ ] ] where @xmath221 so that @xmath222 moreover @xmath223 so that @xmath224 we conclude that @xmath225 and we obtain the estimate @xmath226 for @xmath163 . using , , and the estimate @xmath227 that follows from the boundedness of @xmath228 we can bound both terms in by @xmath229 , which has finite integral over @xmath198 . on @xmath25 follows by differentiating with respect to @xmath230 the relation @xmath231 ] to estimate the integrals of the terms in we observe that @xmath232 with @xmath233 for @xmath234 , we have @xmath235\,dv\right| + c_{26}\int_{\cal d}\frac{|v - v_h|^4}{\delta_h^5}\,dv\nonumber\end{aligned}\ ] ] and @xmath236 passing to polar coordinates @xmath179 , defined by @xmath237 , we find that @xmath238for each @xmath239 and @xmath240 with @xmath166 , in fact @xmath241 moreover , passing to polar coordinates @xmath179 , we have @xmath242\,dv = \nonumber \\ & = & \int_0^r \frac{\rho^4}{(d_h^2+\rho^2)^{5/2}}\,d\rho \sum_{|\gamma|=3}c_\gamma \int_0^{2\pi}(\cos\theta)^{\gamma_1}(\sin\theta)^{\gamma_2}\,d\theta = 0 \label{sndbound}\end{aligned}\ ] ] for some functions @xmath243 , @xmath244 . thus the integrals in are uniformly bounded in @xmath207 . in
we have used @xmath245 for each @xmath246 , with odd @xmath247 . finally , using again , we obtain @xmath248 for some functions @xmath249 .
hence also the integrals in are uniformly bounded in @xmath207 .
to estimate the integrals of the terms in we make the following decomposition : @xmath250 with @xmath251(\cale , v_h+t(v - v_h))\,dt\end{aligned}\ ] ] and @xmath252 . for the first term in
we obtain @xmath253\left[\frac{\partial v_h}{\partial y_k}\cdot{\cal a}_h(v - v_h)\right]\,dv\biggr|\nonumber \\ \nonumber & + & c_{33}\nonumber\end{aligned}\ ] ] where we have used polar coordinates and the inequalities .
the two integrals at the right hand side of vanish : in fact using fubini - tonelli s theorem and passing to polar coordinates @xmath179 , by relations we obtain @xmath254 for some functions @xmath255 .
the computation for the other integral is analogous .
the second term in is estimated in a similar way : @xmath256 and the integral at the right hand side vanishes as well .
we conclude the proof observing that , using and the theorem of differentiation under the integral sign , the derivatives @xmath257 , @xmath258 , restricted to @xmath259 , @xmath190 respectively , correspond to @xmath260 , and their difference in @xmath25 is given by .
if @xmath24 is an orbit configuration with two crossings , assuming that @xmath87 for @xmath261 , we can extract the singularity by considering the approximated distances @xmath262 and the remainder function @xmath263 .
we show that generically we can uniquely extend the solutions of beyond the crossing singularity @xmath264 .
this is obtained by patching together classical solutions defined in the domains @xmath189 with solutions defined in @xmath190 , or _
vice versa_. let @xmath265 be a value for the semimajor axis of the asteroid and @xmath266 be a continuous function defined in an open interval @xmath267 , representing a possible evolution of the asteroid orbital elements @xmath268 .
we introduce @xmath269 with @xmath270 where @xmath50 is gauss constant and @xmath271 is a known function of time representing the evolution of the earth . is constant and represents the trajectory of a solution of the 2-body problem .
if we consider more than one perturbing planet then @xmath272 changes with time due to the planetary perturbations . ]
let @xmath273 be the set of times @xmath274 such that @xmath275 , and assume that each @xmath276 is isolated , so that we can represent the set @xmath277 as disjoint union of open intervals @xmath278 , with @xmath279 a countable ( possibly finite ) set .
we say that @xmath280 is a generalized solution of if its restriction to each @xmath281 is a classical solution of and , for each @xmath282 , there exist finite values of @xmath283 choose @xmath284 and a time @xmath56 such that @xmath285 , with @xmath286 , @xmath287 , @xmath288 .
we show how we can construct a generalized solution of the cauchy problem @xmath289 let @xmath290 be the maximal classical solution of , defined in the maximal interval @xmath291 .
assume that @xmath292 , and @xmath293 , with @xmath24 a non - degenerate crossing configuration such that @xmath294 for some @xmath86 .
let @xmath25 , @xmath295 be chosen as in theorem [ der_jumps ] .
suppose that there exists @xmath296 such that @xmath297 for @xmath298 .
let @xmath299 .
by theorem [ der_jumps ] there exists @xmath300 such that @xmath301 in fact relation is fulfilled by the solution of the cauchy problem is the vector with components @xmath302 , @xmath303 introduced in theorem [ der_jumps ] . ]
@xmath304 which corresponds to the solution of in the interval @xmath305 and is defined also at the crossing time @xmath276 .
let us denote by @xmath306 its value for @xmath307 .
using again theorem [ der_jumps ] we can extend @xmath290 beyond the crossing singularity by considering the new problem @xmath308 the solution of fulfils @xmath309 the vector field in corresponds to @xmath310 on @xmath190 , thus we can continue the solution outside @xmath25 and this procedure can be repeated at almost every crossing singularities . indeed , the generalized solution is unique provided the evolution @xmath311 is not tangent to the orbit crossing set @xmath170 .
moreover , if @xmath312 the extraction of the singularity , described in section [ s : extract ] , can not be performed . in case @xmath313 for @xmath298
the previous discussion still holds if we exchange @xmath314 with @xmath315 . in this case
we prove that the secular evolution of @xmath5 is more regular than that of the orbital elements in a neighborhood of a planet crossing time .
we introduce the secular evolution of the distances @xmath44 and of the orbit distance @xmath5 : @xmath317 assume these maps are defined in an open interval containing a crossing time @xmath276 , and suppose @xmath318 is a non - degenerate crossing configuration at time @xmath276 , as in section [ s : extract ] . in the following
we shall discuss only the case of @xmath44 .
the same result holds for @xmath5 , taking care of the possible exchange of role of two local minima @xmath319 as absolute minimum .
let @xmath290 be a generalized solution of and @xmath320 as in , .
assume @xmath321 is a crossing time and @xmath322 is a non - degenerate crossing configuration with only one crossing point .
then there exists an interval @xmath323 , @xmath324 such that @xmath325 .
[ regevol ] let the interval @xmath323 be such that @xmath326 , where @xmath25 is chosen as in theorem [ der_jumps ] .
we can assume that @xmath327 for @xmath328 , @xmath313 for @xmath329 ( the proof for the opposite case is similar ) .
for @xmath330 the time derivative of @xmath331 is @xmath332 here @xmath333 , @xmath334 , @xmath335 denote the vectors of partial derivatives with respect to @xmath336 respectively .
the derivative @xmath337 exists also for @xmath307 . on the other hand , by theorem [ der_jumps ] ,
the restrictions of @xmath338 to @xmath339 and @xmath340 admit two different continuous extensions to @xmath276 . by , since @xmath341 , we have @xmath342 where @xmath343 is the poisson bracket with respect to @xmath144
. thus the time derivative of @xmath331 exists and is continuous also in @xmath307 .
using a model with 5 planets , from venus to saturn , we compute a planetary ephemerides database for a time span of @xmath344 yrs starting from epoch 0 mjd ( november 17 , 1858 ) with a time step of @xmath345 yrs .
the computation is performed using the fortran program orbit9 , included in the orbfit free software . from this database
we can obtain , by linear interpolation , the evolution of the planetary trajectories at any time in the specified time interval .
we describe the algorithm to compute the solutions of the averaged equations beyond the singularity , where @xmath69 is now the sum of the perturbing functions @xmath79 , @xmath346 , each related to a different planet .
we use a runge - kutta - gauss ( rkg ) method to perform the integration : it evaluates the averaged vector field only at intermediate points of the integration time interval . when the asteroid trajectory is close enough to an orbit crossing , then the time step is decreased to reach the crossing condition exactly . from theorem [ der_jumps ] we can find two lipschitz - continuous extensions of the averaged vector field from both sides of the singular set @xmath170 .
to compute the solution beyond the singularity we use the explicit formula giving the difference between the two extended vector fields , either of which corresponds to the averaged vector field on different sides of @xmath170 .
we compute the intermediate values of the extended vector field just after crossing , then we correct these values by and use them as approximations of the averaged vector field in at the intermediate points of the solutions , see figure [ fig : cross_step ] .
this rkg algorithm avoids the computation of the extended vector field at the singular points , which may be affected by numerical instability .
a difficulty in the application of this scheme is to estimate the size of a suitable neighborhood @xmath25 of the crossing configuration @xmath24 fulfilling the conditions given in section [ s : extract ] .
we performed some tests to compare the solutions of the averaged equations with the corresponding components of the solutions of the full equations . here
we show two tests with the asteroids @xmath347 xb and 1620 ( geographos ) .
we considered the system composed by an asteroid and 5 planets , from venus to saturn .
we selected the 8 values @xmath348 , with @xmath349 , for the initial mean anomaly of the asteroid and the same for the planets . using the program orbit9
, we performed the integration of the system with these 64 different initial conditions ( i.e. we selected equal initial phases for all the planets ) .
then we considered the arithmetic mean of the four equinoctial ] the equinoctial orbital elements have been introduced in @xcite . ]
orbital elements @xmath350 of the asteroid over these evolutions , and compared them with the results of the secular evolution . in figures [ fig : evolcomp_79xb ] , [ fig : evolcomp_1620 ] , we show the results : the crosses indicate the secular evolution , the continuous curve is the mean of full numerical one and the gray region represents the standard deviation from the mean .
the correspondence between the solutions is good . during the evolution
the distance between the asteroid and the earth for some initial conditions attains values of the order of @xmath351 au for @xmath352 ( geographos ) , and @xmath353 au for @xmath347 xb . in figure
[ fig : evolcomp_1620 ] the earth crossing singularity is particularly evident near the epoch 3000 ad .
some numerical tests of the theory introduced in @xcite , with the planets on circular coplanar orbits , can be found in @xcite .
the results of section [ s : dminevol ] can be used to estimate the epoch in which the orbit of a near - earth asteroid will cross that of the earth .
we are interested in particular to study the behavior of those asteroids whose orbits will cross the earth in the next few centuries , so that they must have a small value of @xmath3 already at the present epoch .
we can consider , for example , the set of potentially hazardous asteroids ( phas ) , which have @xmath354 au and absolute magnitude @xmath355 , i.e. they are also large . in figure
[ fig : dmin ] we show 3 different evolutions of the signed orbit distance @xmath5 for the pha @xmath347 xb . here
we draw the full numerical ( solid line ) , secular ( dashed ) and secular linearized ( dotted ) evolution of @xmath356 . by proposition [ regevol ]
the linearization of the secular evolution @xmath357 can give a good approximation also in a neighborhood of a crossing time .
we propose a method to compute an interval @xmath291 of possible crossing times .
we sample the _ line of variation _ ( lov ) , introduced in , which is a sort of ` spine ' of the confidence region ( see also @xcite ) , and compute the signed orbit distance @xmath5 for each virtual asteroid ( va ) of the sample .
then we compute the time derivative of @xmath358 for each va and extrapolate the crossing times by a linear approximation of the evolution .
we set @xmath359 $ ] , with @xmath360 the minimum and maximum crossing times obtained ( see figure [ lov_linear ] ) . in the computation of @xmath291
we take into account a band centered at the earth crossing line @xmath264 : in this test the width of the considered band is @xmath361 au . we describe a method to assign a probability of occurrence of crossings in a given time interval , which is related to the algorithm described above . for each value of the lov parameter @xmath362
we have a va at a time @xmath363 , so that we can compute @xmath364 .
thus , using the scheme of figure [ lov_linear ] we can define a map @xmath365 from the lov parameter line to the time line .
the map @xmath365 gives the crossing times , using the linearized secular dynamics , for the vas on the lov that correspond to the selected values of the parameter @xmath362 .
moreover , we have a probability density function @xmath366 on the lov .
therefore , given an interval @xmath367 in the time line , we can consider the set @xmath368 and define the probability of having a crossing in the time interval @xmath367 as @xmath369 finally , in figure [ confronto ] we show the corresponding interval @xmath370 obtained by computing the secular evolution ( without linearization ) of the orbit distance for each va of @xmath347 xb .
the sizes of @xmath291 and @xmath370 are almost equal , but the left extremum of @xmath370 is @xmath371 years before .
we have studied the double averaged restricted 3-body problem in case of orbit crossing singularities , improving and completing the results in @xcite , @xcite .
this problem is of interest to study the dynamics of near - earth asteroids from a statistical point of view , going beyond the lyapounov times of their orbits . we have also proved that generically , in a neighborhood of a crossing time , the secular evolution of the ( signed ) orbit distance is more regular than the averaged evolution of the orbital elements .
the solutions of this averaged problem have been computed by a numerical method and then compared with the solutions of the full equations in a few test cases .
the results were good enough ; however , we expect that the averaging technique fails in case of mean motion resonances or close encounters with a planet .
we plan to perform numerical experiments with a large sample of near - earth asteroids showing different behaviors : this will be useful to understand the applicability of the averaging technique to the whole set of neas .
we would like to thank an anonymous referee for his interesting comments , that allowed us to improve the paper .
e. bowell and k. muinonen , _ earth - crossing asteroids and comets : groundbased search strategies _ , in hazards due to comets & asteroids " ( ed .
t. gehrels ) , tucson : the university of arizona press , ( 1994 ) , 149197 .
( mr2186811 ) [ 10.1007/s10569 - 005 - 1623 - 5 ] g. f. gronchi , _ an algebraic method to compute the critical points of the distance function between two keplerian orbits _ , cel .
* 93 * ( 2005 ) , 297332 . g. f. gronchi and p. michel , _ secular orbital evolution , proper elements and proper frequencies for near - earth asteroids : a comparison between semianalytic theory and numerical integrations _ , icarus , * 152 * ( 2001 ) , 4857 . ( mr2291871 ) [ 10.3934/dcdsb.2007.7.755 ] g. f. gronchi and g. tommei , _ on the uncertainty of the minimal distance between two confocal keplerian orbits _ , discrete contin .
dyn . syst .
b , * 7 * ( 2007 ) , 755778 . | we study the long term evolution of the distance between two keplerian confocal trajectories in the framework of the averaged restricted 3-body problem .
the bodies may represent the sun , a solar system planet and an asteroid .
the secular evolution of the orbital elements of the asteroid is computed by averaging the equations of motion over the mean anomalies of the asteroid and the planet . when an orbit crossing with the planet occurs the averaged equations become singular .
however , it is possible to define piecewise differentiable solutions by extending the averaged vector field beyond the singularity from both sides of the orbit crossing set @xcite , @xcite . in this paper
we improve the previous results , concerning in particular the singularity extraction technique , and show that the extended vector fields are lipschitz - continuous .
moreover , we consider the distance between the keplerian trajectories of the small body and of the planet . apart from exceptional cases , we can select a sign for this distance so that it becomes an analytic map of the orbital elements near to crossing configurations @xcite .
we prove that the evolution of the ` signed ' distance along the averaged vector field is more regular than that of the elements in a neighborhood of crossing times .
a comparison between averaged and non - averaged evolutions and an application of these results are shown using orbits of near - earth asteroids .
the evolution of the orbit distance in the double averaged restricted 3-body problem + with crossing singularities * giovanni f. gronchi@xmath0 , chiara tardioli@xmath1 * @xmath0dipartimento di matematica , universit di pisa ( italy ) @xmath2dpartement de mathmatique ( naxys ) , fundp ( belgium ) ` [email protected] ` ` [email protected] ` |
in his pioneering work on abelian categories , gabriel investigated a categorial abstraction of the classical notion of noetherian module ( * ? ? ? * chapitre ii.4 ) .
this was one source of inspiration that later led gabriel and ulmer to define the concepts of finitely presentable and finitely generated object in any locally small category @xcite .
also the english summary in @xcite . ) gabriel and ulmer showed that , in an algebraic context , the two concepts agree with the standard notions of finitely presented and finitely generated algebra , respectively ; see lemma [ lem : var ] below .
the purpose of this note is to show that the gabriel - ulmer generalisation affords two widely applicable isomorphism criteria for directed colimits of sequences of objects and morphisms in an arbitrary category . in several concrete situations ,
analogous isomorphism criteria are typically obtained by _
ad hoc _ arguments .
the results given here can thus play the rle of discerning the general from the specific in situations of actual interest . to corroborate this statement , in section [ s : app ]
we shall give two applications of the isomorphism criteria ; we defer further remarks until then .
now we turn to the statement of our main result , for which we recall a number of definitions . throughout , we let @xmath3 , and we let @xmath4 denote the category of sets and functions .
also , we always let @xmath5 denote a locally small category , i.e. a category such that the class @xmath6 of morphisms between any two given objects @xmath7 of @xmath5 is a set .
we recall the standard notion of directed colimit ( = filtered colimit over a directed index set ) in a category ; see e.g. ( * ? ? ?
* section 2.5 ) , where , following a long - standing algebraic tradition , ` directed colimits ' are called ` direct limits ' . a set @xmath8
partially ordered by @xmath9 is ( _ upward _ ) _ directed _ if for any @xmath10 there exists @xmath11 with @xmath12 .
a _ directed diagram _ in @xmath13 is a pair @xmath14 , where @xmath10 and @xmath15 , such that @xmath16 is a @xmath13-object for each @xmath17 , and @xmath18 is a @xmath13-arrow for each @xmath15 .
we call the @xmath19 s the _ transition morphisms_. a _
cocone _ for the diagram @xmath20 is a @xmath13-object @xmath21 equipped with @xmath13-arrows @xmath22 , one for each @xmath17 , that satisfy the commutativity relations @xmath23 , for each @xmath24 with @xmath15 .
a _ colimit _ in @xmath13 of the diagram @xmath25 is a _
universal cocone _
@xmath26 : one such that , for any other cocone @xmath27 , there is a unique @xmath13-arrow @xmath28 satisfying @xmath29 for each @xmath17 .
the @xmath13-arrows @xmath30 of the universal cocone are called the _ colimit morphisms _ , and @xmath21 is the _
colimit object_. the universal property of course entails that directed colimits are unique to within a unique @xmath13-isomorphism .
following @xcite , we say that a @xmath13-object @xmath31 is _ finitely presentable _ if the covariant hom - functor @xmath32@xmath33 @xmath34 preserves directed colimits .
explicitly , this means that if @xmath26 is the colimit in @xmath13 of the directed diagram @xmath25 , then for every @xmath13-arrow @xmath35 the following two conditions are satisfied .
( see fig .
[ fig : fp ] . )
( m)[matrix of math nodes , row sep=4em , column sep=4em , text height=1.5ex , text depth=0.25ex , ampersand replacement= & ] a & b + & b_j + & b_i + ; ( m-2 - 2 ) edge node[descr ] @xmath19 ( m-3 - 2 ) ; ( m-1 - 1 ) edge node[above ] @xmath36 ( m-1 - 2 ) ; ( m-1 - 1 ) edge [ bend right=50 ] node[left ] @xmath37 ( m-3 - 2 ) ; ( m-1 - 1 ) edge [ bend right=30 ] node[right ] @xmath38 ( m-3 - 2 ) ; ( m-1 - 1 ) edge [ bend right=20 ] node[descr ] @xmath39 ( m-2 - 2 ) ; ( m-2 - 2 ) edge [ bend right=60 ] node[right ] @xmath40 ( m-1 - 2 ) ; ( m-3 - 2 ) edge [ bend right=90 ] node[right ] @xmath41 ( m-1 - 2 ) ; @xmath42 here , ( [ eq : factorization ] ) is known as the _ factorisation property _ , and ( [ eq : essential - unique ] ) as the _ essential uniqueness _ property . again after @xcite , we say that @xmath31 is _ finitely generated _
if @xmath43 preserves directed colimits of diagrams @xmath20 all of whose transition morphisms @xmath19 are monomorphisms in @xmath13 .
[ rem : e]in the literature , the following condition is often used in place of ( [ eq : essential - unique ] ) : @xmath44 where @xmath45 is the arrow whose existence is granted by .
. e.g. @xcite .
but it is easy to check that @xmath35 satisfies ( [ eq : factorization ] ) and ( [ eq : essential - unique ] ) if , and only if , it satisfies ( [ eq : factorization ] ) and ( [ eq : essential - uniquebis ] ) .
it will transpire from section [ s : proof ] that ( [ eq : essential - unique ] ) is technically expedient in this note , insofar as it does not depend on .
we next specialise directed diagrams to sequences . by a _ sequence of objects and morphisms _ in @xmath13 we mean a pair @xmath46 with @xmath47 a @xmath5-object , and @xmath48 a @xmath5-arrow . for any two integers @xmath49 we define the @xmath5-arrow @xmath50 as @xmath51 and we also set @xmath52 equal to the identity arrow on @xmath47 . by the _ colimit _ in @xmath13 of such a sequence we mean the colimit of the directed diagram @xmath53 , with @xmath54 . in the central definition that follows we consider a second sequence @xmath55 with
associated directed diagram @xmath56 , @xmath57 .
[ d : confluence]we say that two sequences @xmath46 and @xmath55 in the category @xmath13 are _ confluent _ if there exist integers @xmath58 , @xmath59 , together with @xmath13-arrows @xmath60 and @xmath61 for each @xmath62 , such that the commutativity relations @xmath63 hold .
( see fig .
[ f : confluence ] . )
( m)[matrix of math nodes , row sep=4em , column sep=3em , ampersand replacement= & ] & a_i_n & a_i & a_i_n+1 & + & b_k & b_k_n & b_k_n+1 & + ; ( m-1 - 2 ) edge node[descr ] @xmath64 ( m-2 - 3 ) ; ( m-2 - 3 ) edge node[descr ] @xmath65 ( m-1 - 4 ) ; ( m-1 - 4 ) edge node[descr ] @xmath66 ( m-2 - 4 ) ; ( m-1 - 2 ) edge node[above ] @xmath67 ( m-1 - 3 ) ; ( m-1 - 3 ) edge node[above ] @xmath68 ( m-1 - 4 ) ; ( m-1 - 2 ) edge [ bend left=40 ] node[descr ] @xmath69 ( m-1 - 4 ) ; ( m-2 - 2 ) edge node[below ] @xmath70 ( m-2 - 3 ) ; ( m-2 - 3 ) edge node[below ] @xmath71 ( m-2 - 4 ) ; we shall see in theorem [ t : confluence - implies - iso ] below that , in full generality , confluence of sequences implies isomorphism of the colimit objects .
the converse , however , fails even in the best behaved categories . in @xmath4 , write @xmath72 as a colimit of the sequence @xmath73 . further write @xmath72 as the colimit of the constant sequence @xmath74 , where each arrow is the identity function . then it is clear that the sequences are not confluent .
the second sequence does not consist of finitely presentable or finitely generated objects .
indeed , one checks that finitely presentable objects in @xmath4 coincide with the finitely generated ones , and they are precisely the finite sets .
the main result of this note is that confluence is equivalent to isomorphism of the colimit objects under appropriate finiteness assumptions : [ t : confluence = iso]let @xmath13 be any locally small category .
suppose two sequences @xmath46 and @xmath55 in @xmath13 admit colimit objects @xmath31 and @xmath21 in @xmath13 , respectively . 1 .
[ t : confluence = iso : item1 ] suppose that each @xmath47 and each @xmath75 is finitely presentable .
then @xmath31 and @xmath21 are isomorphic if , and only if , the two sequences are confluent .
[ t : confluence = iso : item2 ] suppose that each @xmath47 and each @xmath75 is finitely generated , and that each @xmath68 and each @xmath70 is a monomorphism .
then @xmath31 and @xmath21 are isomorphic if , and only if , the two sequences are confluent . before turning to the proofs
, we remark that extensions of theorem [ t : confluence = iso ] to more general directed diagrams are possible . indeed , the case of all directed diagrams
is closely related to grothendieck s completion of a category under inductive systems ( `` ind - completion '' ) , for which see ( * ? ? ?
* vi and references therein ) ; cf .
also remark [ rem : ind ] below .
such extensions , however , call for generalised notions of confluence that are , as far as we can see , harder to apply . since the case of sequences appears to be most common in concrete situations see the examples in section [ s : app]we only deal with it in this note .
[ t : confluence - implies - iso ] let @xmath76 be two sequences in @xmath13 with colimits @xmath77 and @xmath78 , respectively . if @xmath76 are confluent then @xmath31 and @xmath21 are isomorphic .
let @xmath58 and @xmath59 be indices such that @xmath79 and @xmath80 , @xmath62 , are a confluence between the given sequences .
consider the subsequence @xmath81 .
a proof of the next claim may be obtained applying standard technical results on final functors , see e.g. @xcite or ( * ? ? ? * vi.1.5 ) .
we prefer to give a direct proof here .
the colimit of the diagram associated with @xmath82 is @xmath83 .
clearly @xmath83 is a cocone for @xmath82 .
let @xmath84 be an arbitrary cocone for @xmath82 .
let us set @xmath85 and @xmath86 .
the poset @xmath8 is cofinal in @xmath72 , i.e. for any @xmath87 there is @xmath88 such that @xmath89 .
let @xmath90 be the function that associates with each @xmath87 the minimum @xmath88 with @xmath89 .
it is elementary to verify that @xmath91 is a cocone for @xmath46 . by the universal property of the colimit @xmath77
, there exists a unique @xmath5-arrow @xmath92 that factors all arrows in @xmath93 through @xmath94 .
hence , _ a fortiori _ , @xmath95 factors all @xmath96 through @xmath97 .
it remains to show that @xmath95 is the unique @xmath5-arrow that has the latter factorisation property . if @xmath98 is another arrow such that @xmath99 for every @xmath62 , then @xmath100 for @xmath87 . upon noting that any @xmath101 is either of the form @xmath102 for some @xmath103 , or else
can be written in the form @xmath104 for some @xmath103 , we infer that @xmath105 must be equal to @xmath95 .
note now that @xmath106 is a cocone for the diagram associated with the sequence @xmath82 . for this
it is sufficient to verify the commutativity of the diagram below .
( m)[matrix of math nodes , row sep=3em , column sep=2.4em , ampersand replacement= & ] a_i_n & & a_i_n+1 & & + b_k_n & & b_k_n+1 & & b + ; ( m-1 - 1 ) edge node[left ] @xmath64 ( m-2 - 1 ) ; ( m-2 - 1 ) edge node[descr ] @xmath65 ( m-1 - 3 ) ; ( m-1 - 3 ) edge node[right ] @xmath66 ( m-2 - 3 ) ; ( m-1 - 1 ) edge node[above ] @xmath107 ( m-1 - 3 ) ; ( m-2 - 1 ) edge node[below ] @xmath108 ( m-2 - 3 ) ; ( m-2 - 1 ) edge [ bend right=30 ] node[below ] @xmath109 ( m-2 - 5 ) ; ( m-2 - 3 ) edge [ bend right=20 ] node[below , xshift=-.9cm , yshift=.2 cm ] @xmath110 ( m-2 - 5 ) ; we compute : @xmath111 since , by the claim above , @xmath31 is the universal cocone for the diagram associated with @xmath82 , there must exist a unique @xmath112 that factors the family of arrows @xmath113 , i.e.@xmath114 arguing symmetrically , we obtain a unique @xmath115 such that @xmath116 now , to prove that the composition @xmath117 is the identity on @xmath31 , we compute for every @xmath62 : @xmath118 so @xmath117 factors all colimit morphisms of the colimit @xmath83 of @xmath82 .
but , by the universal property of colimits , there exists a unique morphism with this property , and it is the identity hence @xmath119 . the proof that @xmath120 is symmetrical .
therefore @xmath36 and @xmath121 are isomorphisms , and the proof is complete .
the next technical lemma establishes a key finiteness property of directed colimits of diagrams all of whose objects satisfy condition with respect to the colimit .
it asserts that any arrow @xmath122 that equalises two colimit arrows must already equalise two transition maps .
we will use this in the proof of lemma [ l : local ] below .
[ l : weak - implies - eventual ] let @xmath123 be a directed diagram in @xmath13 , and let @xmath124 be its colimit .
suppose that every @xmath13-arrow @xmath125 satisfies w.r.t .
@xmath123 . consider any @xmath126-arrow @xmath122 with @xmath15 .
if @xmath127 , then there exists @xmath128 such that @xmath129 . the colimit morphism @xmath130 factors through each transition morphism @xmath131 with @xmath15 , in symbols , @xmath132 by hypothesis : @xmath133 applying to ( [ eq : one][eq : two ] ) with @xmath134 , @xmath135 , and @xmath136
, we infer that there must be @xmath128 such that @xmath137 .
[ l : local ] let @xmath76 be two sequences in @xmath13 with colimit objects @xmath138 and @xmath139 , respectively .
suppose that @xmath31 and @xmath21 are isomorphic .
suppose further that the following hold for each @xmath140 . 1 .
[ l : local : item1 ] every @xmath13-arrow @xmath141 satisfies w.r.t .
[ l : local : item2 ] every @xmath13-arrow @xmath142 satisfies w.r.t . @xmath46 .
[ l : local : item3 ] every @xmath13-arrow @xmath143 satisfies w.r.t .
[ l : local : item4 ] every @xmath13-arrow @xmath144 satisfies w.r.t .
@xmath55 . then @xmath46 and @xmath55 are confluent .
let @xmath112 be an isomorphism .
by induction on @xmath145 , we define sequences of integers @xmath146 and @xmath147 , and morphisms @xmath148 such that the following equalities hold : @xmath149 for @xmath150 , we set @xmath151 . by ( h[l : local : item1 ] ) , there exists an arrow @xmath152 , for some @xmath153 , that factors @xmath154 .
let us set @xmath155 ; hence @xmath156 , and is satisfied .
next , by ( h[l : local : item3 ] ) there exists an arrow @xmath157 , for some @xmath158 , that factors @xmath159 .
set @xmath160 ; then @xmath161 , and holds , too .
for the inductive step , assume that such arrows @xmath162 have been defined for @xmath163 .
since , by ( h[l : local : item1 ] ) , @xmath164 satisfies with respect to @xmath165 , there exist @xmath166 and @xmath167 such that @xmath168 we may safely assume that @xmath169 . for , in the contrary case , it suffices to replace @xmath170 with its composition with a transition arrow @xmath171 for some @xmath172 , and to rewrite the right - hand side of as @xmath173 . by the induction hypothesis
, @xmath174 satisfies the following equality : @xmath175 composing both sides of on the right with @xmath176 we obtain @xmath177 . using , the latter equation yields @xmath178 , and hence @xmath179 in light of ( h[l : local : item2 ] )
, we may apply lemma [ l : weak - implies - eventual ] to ( with @xmath180 ) to deduce that there exists @xmath181 , @xmath182 , such that @xmath183 upon setting @xmath184 and @xmath185 , reads @xmath186 so that in definition [ d : confluence ] is satisfied . to conclude this part of the inductive step ,
we show that @xmath64 satisfies : @xmath187 the inductive construction of the arrow @xmath80 , and the verification that in definition [ d : confluence ] holds , are symmetric to the above .
the proof is complete .
the right - to - left implications in parts 1 and 2 of the statement are given by theorem [ t : confluence - implies - iso ] .
for the left - to - right implications , it is enough to check that all hypotheses in lemma [ l : local ] are satisfied .
concerning part 1 , by assumption each @xmath47 is finitely presentable , @xmath21 is the colimit of @xmath55 , and @xmath31 is the colimit of @xmath46 ; hence every @xmath13-arrow @xmath141 satisfies w.r.t .
@xmath55 , and every @xmath13-arrow @xmath142 satisfies w.r.t .
conditions ( h[l : local : item3]h[l : local : item4 ] ) in lemma [ l : local ] hold by a symmetric argument .
the fact that lemma [ l : local ] applies under the assumptions of part 2 is equally straightforward .
the theorem is proved .
for background on the universal - algebraic notions used in this subsection we refer to ( * ? ? ?
* chapter 2 ) . by a _ variety of algebras _
* section 2.8 ) we mean , as usual , the class of all algebraic structures of the same type that satisfy some set of equations ; equivalently , by birkhoff s variety theorem ( * ? ? ?
* theorem 2.14 ) , a class that is closed under homomorphic images , subalgebras , and direct products .
we regard varieties as categories in the obvious manner , with the homomorphisms as morphisms .
if @xmath188 is any variety , there is a forgetful functor @xmath189 that carries algebras to underlying sets , and homomorphisms to functions .
then @xmath190 has a left adjoint @xmath191 called the _ free functor _ , and the algebra @xmath192 is called the _ free algebra _ ( in @xmath188 ) _ generated by the set @xmath82_. free algebras are characterised to within isomorphism by the usual universal mapping property ; see ( * ? ? ?
* theorem 2.9 ) .
let us write @xmath193 for @xmath192 , @xmath82 of finite cardinality @xmath194 .
congruence _ on an algebra @xmath31 in @xmath188 is an equivalence relation on @xmath31 that is also a subalgebra of @xmath195 . by the first isomorphism theorem @xcite ,
congruences on @xmath31 are in bijection with kernels of onto homomorphisms with domain @xmath31 .
an algebra in @xmath188 is _ finitely generated _ if it is isomorphic to a quotient of @xmath196 , for @xmath194 an integer .
a congruence on @xmath31 is _ finitely generated _ if it is the intersection of all congruences containing some finite subset of @xmath195 .
adapting a standard notion from group theory , an algebra in @xmath188 is _ finitely presented _ if it is ( isomorphic to ) a quotient of @xmath196 modulo some finitely generated congruence , for some integer @xmath194 .
it is a standard fact that varieties of algebras are closed under directed colimits ( * ? ? ?
* theorem 2.8 ) .
[ lem : var]in any variety of algebras , an algebra is finitely presentable respectively , finitely generated in the sense of gabriel and ulmer if , and only if , it is finitely presented respectively , finitely generated in the usual algebraic sense .
it is well known that in any variety of algebras each algebra is representable either as a directed colimit of finitely presentable algebras ( see e.g. ( * ? ? ?
* lemma vi.2.2 ) ) , or as a directed colimit of its finitely generated subalgebras ( see e.g. ( * ? ? ? * theorem 2.7 ) ) with inclusion maps as transition homomorphisms .
we omit the proof of the fact that one can specialise these results as follows .
a. [ eq : app1 ] each finitely generated algebra in a variety is the directed colimit of a sequence of finitely presented algebras and surjective transition homomorphisms .
b. [ eq : app2 ] each countable algebra in a variety is the directed colimit of a sequence of finitely generated subalgebras and injective transition homomorphisms . 1 .
[ cor : var : item1 ] suppose that each @xmath47 and each @xmath75 is a finitely presented algebra .
then @xmath31 and @xmath21 are isomorphic if , and only if , the two sequences are confluent .
[ cor : var : item2 ] suppose that each @xmath47 and each @xmath75 is a finitely generated algebra , and that each @xmath68 and each @xmath70 is an injective homomorphisms .
then @xmath31 and @xmath21 are isomorphic if , and only if , the two sequences are confluent
. apply theorem [ t : confluence = iso ] and lemma [ lem : var ] , together with the easy observation that a homomorphism between algebras is a monomorphism if , and only if , it is an injective function ( * ? ? ?
* proposition 2.3 ) . in a number of classical papers , derry , kurosh , and
malcev developed a technique to construct complete isomorphisms invariants for countable torsion - free abelian groups of finite rank in terms of equivalence classes of families of matrices .
the hypothesis of finite rank was later removed by fuchs ; we refer to his textbook account @xcite for details and references .
we sketch here a proof of the fact that the existence of invariants of this sort essentially follows from corollary [ cor : var ] .
indeed , we can represent each countable torsion - free abelian group as the directed colimit of some sequence @xmath197 with each @xmath198 an injective group homomorphism .
the approximation is a special case of , up to the fundamental theorem of abelian groups that a finitely generated torsion - free abelian group is free of finite rank .
corollary [ cor : var ] now tells us that two sequences as in ( [ eq : groups ] ) have isomorphic groups as colimit objects if , and only if , they are confluent . choosing @xmath199-module bases , representing each group homomorphism featuring in a confluence diagram as a matrix with integer entries , and writing down the commutativity conditions ( [ eq : comm1][eq : comm2 ] ) in terms of products of such matrices , one obtains an equivalence relation on sequences of matrices that yields a complete isomorphism invariant closely related to the classical ones referred to above . in the recent paper @xcite , busaniche , cabrer , and mundici
prove an isomorphism criterion for directed colimits of sequences of finitely presented lattice - ordered abelian groups with an order - unit , where all transition morphisms are assumed to be surjective .
their proofs make use of the non - trivial representation of these structures in terms of compact polyhedra and piecewise linear maps .
( for further information on the topic , the interested reader may consult ( * ? ? ? * ; * ? ? ? * ; * ? ? ? * and references therein ) . ) nonetheless , their isomorphism criterion ( * ? ? ?
* theorem 3.3 ) is an immediate consequence of corollary [ cor : var ] , given that by a well - known theorem of mundici ( * ? ? ?
* theorem 3.9 ) the category in question is equivalent to a variety of algebras , namely , chang s mv - algebras @xcite . in fact , our results show that the assumption in ( * ? ? ?
* theorem 3.3 ) that all transition morphisms be surjective may be dropped .
[ [ ss : bratteli ] ] the bratteli - elliott isomorphism criterion for dimension groups , and af @xmath1-algebras .
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ for background on dimension groups see @xcite .
a _ partially ordered abelian group _ is an abelian group @xmath200 ( which we always write additively ) equipped with a _ translation - invariant _ partial order : @xmath201 implies @xmath202 for each @xmath203 .
a morphism of partially ordered abelian groups is a group homomorphism @xmath204 that is order - preserving .
the _ positive cone _ of @xmath200 is @xmath205 .
one says that @xmath200 is _ directed _ if it is generated ( as a group ) by @xmath206 ; and that it is _ unperforated _ ( or has _ isolated order _ ) if @xmath207 for some @xmath208 and @xmath209 implies @xmath210 .
further , @xmath200 has ( _ riesz _ ) _ interpolation _ if whenever @xmath211 satisfy @xmath212 , there exists @xmath213 such that @xmath214 .
a _ dimension group _ is any partially ordered abelian group that is directed , unperforated , and has interpolation .
the _ category of dimension groups _ is the full subcategory of the category of partially ordered abelian groups whose objects are dimension groups .
simplicial group _ is ( any partially ordered abelian group isomorphic to ) the free abelian group @xmath215 of rank @xmath216 , equipped with the pointwise order inherited from @xmath199 , i.e. with the positive cone @xmath217 .
it is clear that each simplicial group is a dimension group .
the category of dimension groups has directed colimits , and they are computed as directed colimits of abelian groups , where the colimit group is equipped with the positive cone obtained as the set - theoretic colimit of the positive cones in the diagram ; see @xcite .
specifically , consider a sequence @xmath218 of simplicial groups and order - preserving homomorphisms . if @xmath219 is the colimit of ( [ eq : dimgrp ] ) in the category of abelian groups , and @xmath220 are the colimit maps , then setting @xmath221 yields a positive cone on @xmath219 that makes it into a partially ordered abelian group .
it is elementary to check that , with this order , @xmath219 is in fact a dimension group .
let @xmath31 be a dimension group that is the colimit object of a directed diagram @xmath123 of dimension groups and order - preserving group homomorphisms , and write @xmath130 for the colimit morphisms .
consider an order - preserving group homomorphism @xmath222 , for some integer @xmath209 .
let @xmath223 denote the standard @xmath199-module basis of @xmath224 .
by @xcite , for each @xmath225 there exists @xmath226 and @xmath227 such that @xmath228 . by the directedness of the index set ,
there exist @xmath229 and @xmath230 such that @xmath228 holds for each @xmath225 . since simplicial groups are free abelian groups , there exists a unique group homomorphism @xmath231 that satisfies @xmath232 , for each @xmath225 .
clearly @xmath233 .
note that @xmath121 preserves the positive cone of @xmath224 ( i.e. is _ positive _ ) , and therefore it is order - preserving .
this proves ( [ eq : factorization ] ) .
it is also clear that @xmath121 is essentially unique in the sense of ( [ eq : essential - unique ] ) , because any two group homomorphisms @xmath234 are uniquely determined by their action on @xmath223 .
[ rem : ind]the converse of lemma [ lem : fpsimplicial ] is also true , though not as easy to prove : it can be shown that it follows from the important effros - handelman - shen theorem ( * ? ? ?
* theorem 2.2 ) that each dimension group is the colimit object of a directed diagram of simplicial groups , the colimit being taken in the category of partially ordered abelian groups . thus , the characteristic approximation property of dimension groups is a notable instance of the much - studied abstract theory of accessing objects in a category via directed colimits of finitely presentable objects @xcite . in this connection , an obvious question for further research
is whether dimension groups are the ind - completion of the category of simplicial groups .
we plan to pursue this question elsewhere , and for now refer the interested reader to @xcite , where related issues are considered from the perspective of universal algebra .
we close this paper with some remarks on af @xmath1 algebras ; see e.g. @xcite for background . recall that a _ @xmath1-algebra _ is a ( not necessarily commutative ) algebra over @xmath235 , equipped with a norm @xmath236 and an involution @xmath237 , such that @xmath31 is complete with respect to @xmath236 , and satisfies @xmath238 and @xmath239 for all @xmath240 .
we always assume @xmath1-algebras to be _ unital _ ,
i.e. to come with a multiplicative identity
.- algebras here to simplify references to the literature . ]
morphisms of @xmath1-algebras are taken to be the _
@xmath237-homomorphisms _ ,
i.e. the ( unit - preserving ) algebra morphisms that commute with the involution .
it is proved in @xcite that arbitrary directed colimits exist in the category of @xmath1-algebras .
we write @xmath241 for the set of all @xmath242 matrices with complex entries .
then @xmath243 is a @xmath2-algebra with the involution and the norm obtained from identifying it with the bounded linear operators on the hilbert space @xmath244 . in particular ,
the involution is the conjugate transpose involution that takes a matrix @xmath245 to @xmath246 , where @xmath247 is the conjugate of @xmath248 .
the @xmath1-algebras @xmath249 , for integers @xmath250 , are defined on the product algebra using componentwise operations and the supremum norm . by a standard result ( @xcite , @xcite ) , up to an isomorphism the non - trivial @xmath1-algebras that are finite - dimensional ( as algebras ) are isomorphic to @xmath249 for some choice of the integers @xmath251
. in a seminal paper , bratteli @xcite defined a @xmath1-algebra @xmath31 to be _ approximately finite - dimensional _ ( or an _
af @xmath2-algebra _ , for short ) if it satisfies @xmath252 for some inclusion - increasing subsequence of finite - dimensional subalgebras @xmath47 , where @xmath253 denotes the norm - closure .
it is known @xcite that this is the same as defining af @xmath1-algebras as the directed colimits of sequences of finite - dimensional @xmath1-algebras ( with arbitrary transition @xmath237-homomorphisms ) .
the _ bratteli isomorphism criterion _
@xcite asserts that two af @xmath1-algebras @xmath252 and @xmath254 are isomorphic if , and only if , the two approximating sequences @xmath255 and @xmath256 of finite - dimensional @xmath1-algebras and inclusions are confluent in the sense of definition [ d : confluence ] .
it is important to emphasise that _
bratteli s criterion does not follow from our theorem [ t : confluence = iso ] .
_ this is because in the category of @xmath1-algebras ( or even in the full subcategory of af @xmath1-algebras ) a finite - dimensional algebra is in general neither finitely generated nor finitely presentable in the sense of gabriel and ulmer .
indeed , consider a @xmath237-homomorphism @xmath257 , with @xmath258 finite - dimensional , and pick a set of matrix units @xmath259 in @xmath258 , for integers @xmath260 and @xmath261 .
( see @xcite , @xcite , and @xcite for background on matrix units . )
writing @xmath262 for the colimit @xmath237-homomorphisms , there _ need not _ exist an integer @xmath153 such that the images @xmath263 of the chosen matrix units are contained in @xmath264 : by bratteli s intrinsic characterisation of af @xmath1-algebras @xcite , the most that can be said is that there exist an integer @xmath265 , elements @xmath266 in @xmath264 , and a real number @xmath267 , such that @xmath268 .
but by the universal property of matrix units @xcite , the required essentially unique factorisation of @xmath95 only exists if the stronger containment condition @xmath269 can be satisfied for some @xmath265 .
the problem here is that , unlike the situation for varieties of algebras , directed colimits in the category of @xmath1-algebras are not computed as the set - theoretic colimit with additional structure ; see @xcite .
even though dimension groups are not defined as a variety of algebras , their directed colimits are computed as in sets , as we recalled above .
this offers a genuine simplification with respect to af @xmath1-algebras , and explains why upon passing from af @xmath1-algebras to dimension groups by taking elliott s ordered @xmath0 ( see @xcite ) one can obtain the easy proof above of corollary [ cor : elliottbratteli ] .
* acknowledgments . * both authors acknowledge partial support from the italian national research project 2010fp79lr .
parts of this article were written while the authors were kindly hosted by the argentinian national research council ( conicet ) in buenos aires within the european fp7-irses project _ matomuvi _ ( pief - ga-2009 - 247584 ) .
the second - named author gratefully acknowledges partial support by a marie curie intra - european fellowship within the project _ adams _ ( pief - ga-2011 - 299071 ) .
the first - named author acknowledges partial support by the italian firb `` futuro in ricerca '' grant rbfr10dgua .
f. ulmer .
locally @xmath272-presentable and locally @xmath272-generated categories . in _ reports of the midwest category seminar ,
v ( zurich , 1970 ) _ , pages 230247 .
lecture notes in math .
springer , berlin , 1971 . | using the general notions of finitely presentable and finitely generated object introduced by gabriel and ulmer in 1971 , we prove that , in any ( locally small ) category , two sequences of finitely presentable objects and morphisms ( or two sequences of finitely generated objects and monomorphisms ) have isomorphic colimits ( = direct limits ) if , and only if , they are _
the latter means that the two given sequences can be connected by a back - and - forth chain of morphisms that is cofinal on each side , and commutes with the sequences at each finite stage . in several concrete situations ,
analogous isomorphism criteria are typically obtained by _
ad hoc _ arguments .
the abstract results given here can play the useful rle of discerning the general from the specific in situations of actual interest .
we illustrate by applying them to varieties of algebras , on the one hand , and to _ dimension groups_the ordered @xmath0 of approximately finite - dimensional @xmath1-algebras on the other .
the first application encompasses such classical examples as kurosh s isomorphism criterion for countable torsion - free abelian groups of finite rank .
the second application yields the bratteli - elliott isomorphism criterion for dimension groups .
finally , we discuss bratteli s original isomorphism criterion for approximately finite - dimensional @xmath1-algebras , and show that his result does not follow from ours .
filtered colimit , directed colimit , direct limit , finitely presentable object , finitely generated object , partially ordered abelian group , dimension group , af @xmath2-algebra , bratteli - elliott isomorphism criterion , variety of algebras . |
a theoretical framework for constructing model many - body systems that can exhibit phases with fractionalized quantum numbers has been proposed by senthil and fisher.@xcite here , we report results obtained from a numerical simulation of one such model and examine a recently proposed experimental test for detecting the fractionalized phase.@xcite the quantum many - body model that we will study consists of `` chargons '' coupled to a fluctuating @xmath0 gauge field in two spatial dimensions .
it was obtained by senthil and fisher by considering the special case of @xmath1-wave pairing with an even number of electrons per unit cell , and integrating out the spinon degrees of freedom .
this model has the advantage of being straightforward to simulate , while allowing a test of some of the underlying ideas associated with the fractionalized phases .
the hamiltonian for the 2d quantum lattice has the form @xmath2 - h \sum_{\langle ij\rangle } \sigma^x_{ij } \label{qham}\ ] ] where the chargon creation operator is @xmath3 and @xmath4 is conjugate to @xmath5 so that @xmath6=i\ , \delta_{ij}$ ] .
the gauge field operator on the link between nearest - neighbor sites @xmath7 and @xmath8 is @xmath9 , and @xmath10 is the product of gauge operators around a plaquette .
the sum @xmath11 is over nearest neighbor sites on the 2d spatial lattice .
a `` vison '' excitation consists of a plaquette for which @xmath10 is equal to @xmath12 .
visons are always joined in pairs by a string of plaquettes with two links flipped relative to the rest of the links in the vicinity .
a schematic example of this is shown in fig .
[ fig : visons](a ) . here and in the text , we have taken the flipped links to have @xmath13 for definiteness .
for the purpose of carrying out simulations , we work with the 3d ( two space and one euclidean time ) classical action associated with the hamiltonian of eq .
( [ qham ] ) , @xmath14 . \label{oneone}\ ] ] this action has an @xmath15 angular variable @xmath16 corresponding to the eigenvalue of the operator @xmath5 and a @xmath0 gauge field @xmath17 corresponding to the eigenvalue of the operator @xmath9 . here
, @xmath11 indicates all nearest neighbor sites on the 3d lattice .
this action has rotational symmetry as well as a local @xmath0-gauge symmetry in which @xmath16 at a site transforms to @xmath18 and all of the @xmath19 gauge fields linked to the @xmath20 site change sign . as discussed in ref .
, this model is expected to have the type of phase diagram illustrated in fig .
[ fig : phase ] .
both the gauge field @xmath19 and the @xmath15 field @xmath16 are disordered in region i when @xmath21 and @xmath22 are small . in the 2d quantum version ,
this corresponds to the fluctuating @xmath9 gauge field confining the chargons so that there are no free @xmath23 excitations , only @xmath24 excitations . in region
iii , the @xmath15 rotation symmetry is broken , as well as the @xmath0 gauge symmetry .
this is just the usual @xmath15 phase with a finite helicity modulus . here
, chargon pairs @xmath24 condense to form a superfluid .
region ii corresponds to a `` fractionalized '' ( unconfined ) phase .
we find that the @xmath25 field is disordered , as in region i , but the @xmath0 gauge field is ordered , or equivalently in the quantum version , the visons are gapped.@xcite this allows the chargon pairs to `` fractionate , '' and individual @xmath23 chargon excitations are present in the quantum version .
as one enters the superconducting phase , region iii , it is the @xmath23 field that condenses .
the fractionalized phase , region ii , is characterized by a `` topological '' order .
that is , on a manifold with a non - trivial topology , the ground state of the 2d quantum system has a degeneracy that depends upon the topology . with periodic boundary conditions the 2d quantum system has the topology of a torus .
a non - trivial topological excitation occurs if a string of plaquettes with two flipped links associated with a vison pair cuts through the torus , as illustrated schematically in fig .
[ fig : visons](b ) . in order to minimize the energy ,
the @xmath25 field has a discontinuity of @xmath26 ( mod @xmath27 ) across the flipped links denoted by the dashed lines of fig .
[ fig : visons](b ) . because the @xmath25 field does not have long - range order in region ii , this disturbance dies out within a correlation length , and its energy
does not grow with the lattice size .
when a vison loop threads the 3d torus used in our simulations , this topological configuration is trapped because the free energy barrier the system must go over to reach the no - vison state grows as the lattice size , @xmath28 , which goes to infinity in the bulk limit . as proposed by senthil and fisher , the existence of this topological order can be probed by driving this system into the superfluid state iii , where the @xmath25 field does have long - range order . in this case , when a trapped vison is present in region ii and the system is driven from region ii to region iii ( by , for example , increasing @xmath21 ) , the @xmath26 phase shift in @xmath25 associated with the vison will induce a circulating current of bosons.@xcite thus , by going into the superconducting phase iii one can look for trapped visons by measuring the boson current .
if they exist , the trapped visons tell us that we have come from a `` fractionalized '' phase .
abelian gauge theories have been studied extensively by high - energy physicists since the earliest days of lattice gauge theory.@xcite the work most closely related to our own is the study of the abelian higgs model with a @xmath0 gauge field coupled to an ising matter field,@xcite and a @xmath29 gauge field coupled to an @xmath15-matter field.@xcite here we study a @xmath0 gauge field coupled to an @xmath15 field .
more recently , an action similar to that of eq . [ oneone ] , but with an @xmath30 matter field , was used by lammert , rokhsar , and toner to study a classical model for nematics.@xcite the existence of three distinct phases in the model we study is consistent with conclusions drawn from the study of more general abelian higgs models , in which the higgs field is not in the fundamental representation of the gauge group.@xcite in particular , as pointed out by senthil and fisher@xcite , each of the phases discussed here has an analogue in the @xmath30 model of nematics.@xcite by contrast , the @xmath0 gauge theory coupled to an ising matter field has only a confined and a deconfined phase with the higgs and confined phases being analytically connected,@xcite as is expected in general for abelian higgs theories in which the matter field is in the fundamental representation of the gauge group.@xcite in section 2 , we will discuss monte carlo results for the polyakov loop ( the product of @xmath19 wrapped periodically around the lattice ) and the helicity modulus that give us numerical results for the phase diagram . then in section 3 , we will discuss visons and the senthil - fisher test for fractionalization .
section 4 contains our conclusions .
, dotted lines denote @xmath31 .
all visons on the lattice are labeled @xmath12 . ]
using the 3d euclidean action , we measure the polyakov loop and the helicity modulus to find the transitions in both the gauge field and the bosonic field . on our lattices
the polyakov loop provides a useful probe of the gauge field , as we will discuss .
the helicity modulus measures the stiffness of the bosonic field to rotating the spins . it can indicate the phase of the boson field , since in our @xmath15 spin formulation the superconducting state is characterized by a finite spin stiffness . it is well known that for @xmath32 the action of eq . [ oneone ] gives rise to a second - order phase transition on an infinite three - dimensional lattice.@xcite the confined ( strong coupling ) phase is characterized by area law behavior of the wilson loops , and the deconfined ( weak coupling ) phase by perimeter law behavior .
however , on a finite lattice there is a cross - over , rather than a _ bona fide _ phase transition . for the relatively small lattices on which we perform our simulations , measurement of the polyakov loop provides a convenient way to locate the cross - over .
a polyakov loop is the product of a line of @xmath19 wrapped periodically around the lattice , @xmath33 where all the links are pointing in the @xmath34 direction . for strong coupling
it vanishes order by order in perturbation theory , and in our simulations we find that it fluctuates about zero , as is seen in fig .
[ fig : polymc1 ] . at weak coupling an expansion about the state in which all the @xmath19 have the same sign gives , to leading order,@xcite @xmath35 where @xmath28 is the number of lattice points in the temporal direction
. a reference state in which the polyakov loop has the same magnitude , but opposite sign , can be obtained by the addition of a vison through the torus that flips all the links in the temporal direction between two adjacent time slices . by tracking @xmath36 in our simulations
, we observe the system tunneling between these two degenerate states , as is seen in fig .
[ fig : polymc2 ] . on our finite lattice @xmath36 averages to zero because of cancellations between states in which it takes on positive and negative values , but above the transition the tunneling occurs much less frequently than our period of observation so that we obtain a non - zero average .
it follows from eq .
[ expfalloff ] that for @xmath37 the jump in @xmath36 goes to zero .
however , in our simulations , @xmath38 , and the cross - over is in the neighborhood of @xmath39 .
for these values the magnitude of the polyakov loop at the crossover is approximately @xmath40 .
we then measured the polyakov loop along lines of constant @xmath21 and @xmath22 .
figure [ fig : poly ] shows the expectation value of polyakov loops along lines of constant @xmath22 .
these monte carlo measurements were taken using a local metropolis updating scheme on a lattice with @xmath41 sites .
we took 150 measurements for each point with each measurement separated by 35 monte carlo updates of the lattice .
after the measurements at a point are done , @xmath21 is increased and the system is allowed to equilibrate for 200 monte carlo updates . in this case , each run is started with the system completely ordered so that no vison loops get frozen into the system .
figure [ fig : polyj ] shows the expectation value of polyakov loops along lines of constant @xmath21 .
the measurements were taken in the same way with @xmath22 increased during the run instead of @xmath21 .
additionally , 500 monte carlo steps were used in between measurements to equilibrate . for this figure
each run is started from a completely disordered state .
when the system enters the deconfined phase ( ii ) from the confined phase ( i ) it has the opportunity to trap a vison . a trapped vison changes the expectation value of the polyakov loop from @xmath42 to @xmath12 and accounts for the run with @xmath43 .
the standard form of the helicity modulus for the @xmath15 model is not invariant under the @xmath0 gauge transformation .
it can be made gauge invariant by inserting factors of the gauge field .
this gives a helicity modulus of the form @xmath44 where @xmath34 points along the bonds in the @xmath45 , @xmath46 , or @xmath47 directions and @xmath48 is a unit vector pointing from the @xmath20 lattice site to the @xmath49 lattice site . since the helicity modulus measures the stiffness of the spins , it is @xmath50 where the bosons are disordered and finite where the bosons have long - range order.@xcite the helicity modulus measured along lines of constant @xmath22 is shown in fig .
[ fig : helmod ] .
these measurements were taken in the same way as the measurements of the polyakov loop : 150 measurements at each point separated by 35 monte carlo steps between measurements and 200 monte carlo measurements between points .
the transition is rounded due to finite size effects .
the results found by looking at the expectation value of the helicity modulus and the polyakov loop were used to construct the phase diagram shown in fig .
[ fig : phase ] .
the transitions were taken to be when the observable is statistically non - zero .
the nature of the transition between the deconfined phase and the confined phase and between the deconfined phase and the condensed phase are understood from studying the ising model and the xy model , and the boundary between the confining phase and the condensed phase has been studied by senthil and fisher.@xcite this is a transition at which both the gauge field and the boson field order .
this phase transition occurs down to @xmath51 where there is no plaquette term in the action , implying that it is the ordering of the bosons that forces the gauge field to order .
as discussed in the introduction , the fractionalized phase ii has a topological order caused by the presence of visons threading the torus .
these visons are trapped , as previously noted .
senthil and fisher recently proposed a method for detecting trapped visons by driving the system between the fractionalized phase ii and phase iii , in which the @xmath25 field is condensed.@xcite in our numerical experiments the presence of visons can be determined by observables such as the polyakov loop that measure the gauge field directly , while in real - world systems this is typically not possible .
it is therefore necessary to be able to determine the topological order without observing the gauge field directly .
however , as discussed by senthil and fisher , the boson field can be used to probe the topology on the lattice in the condensed phase , in which it has long - range order .
anti - periodic boundary conditions generated by a vison threading the torus cause the bosons to gradually twist by @xmath52 from one side of the vison string to the other .
this gives rise to a non - zero gauge invariant current , @xmath53 which can be measured .
here @xmath34 is a spatial unit vector perpendicular to the vison string . in this way ,
measuring the gauge - invariant current can determine the topological order of the lattice .
thus , when a single vison string threads the torus we expect a boson current to flow corresponding to the @xmath26 change in the phase produced by the vison .
in the ordered phase , the fluctuations in the boson phase lead to a renormalization of j , @xmath54 as the bare coupling @xmath21 increases and one goes deep into the boson condensed phase , the phase fluctuations decrease and @xmath55 goes to @xmath42 . in a real physical measurement , the quantity entering the circulation or flux is the renormalized coupling @xmath56 , so that a measurement of @xmath57 gives @xmath58 . for a finite @xmath59 lattice
we expect that when a vison is trapped and the system is switched into the condensed boson phase iii by increasing @xmath21 , one will find @xmath60 which goes to @xmath61 as @xmath28 goes to infinity .
the lattice is initially prepared at @xmath62 and @xmath63 in the condensed phase with a single vison string threading the torus and a gradual twist of the bosons by @xmath26 . in a real - world system , this would be achieved by threading an @xmath64 magnetic flux quantum through the sample.@xcite if the system is then moved to the fractionalized phase by decreasing @xmath21 to @xmath65 , the boson current disappears , but the vison remains trapped so that the boson current returns with the same magnitude if the system is moved back to the condensed phase .
we have done this numerically on an @xmath41 lattice using our euclidean action , with the results shown in fig .
[ fig : visa ] .
here we have plotted @xmath66 and we see that this ratio is non - zero in the condensed phase , signifying the presence of a vison that remains trapped in the fractionalized phase . for @xmath62 and @xmath63 on an @xmath41 , lattice @xmath55 is measured to be @xmath67 so that @xmath68 in agreement with fig . [
fig : visa ] . note that the boson current can be either positive or negative reflecting the direction of the gradual twist induced by the vison .
if the system with a single trapped vison is moved to a phase without fractionalization , the confined phase i , the vison can escape .
for example , at @xmath69 and @xmath70 we have prepared the system in the condensed phase with a single vison string threading the torus and a boson current , as shown over the first @xmath71 sweeps in fig .
[ fig : visb ] . for @xmath69 and @xmath70 on an @xmath41 lattice @xmath55 is measured to be @xmath72 , so that initially @xmath73 .
then the system is taken into the confined phase i by decreasing @xmath21 to @xmath65 at constant @xmath69 .
when , after another @xmath71 monte carlo steps , the system is taken back into the boson condensed phase iii by increasing @xmath21 to @xmath74 , the boson current is seen to vanish .
this means that the vison that was initially trapped in the condensed phase escaped when @xmath21 was decreased to @xmath65 with @xmath69 .
the boson current remains zero through further cyclings between the phases .
this measurement shows that @xmath75 corresponds to a point in the confined phase .
this type of measurement can also be used to determine the line separating the confined and fractionalized phases .
we can determine whether a given point in the phase diagram @xmath76 , is in the fractionalized or confined phase by the following set of measurements .
we start by equilibrating a trapped vison at the point @xmath77 , which we know to be in the fractionalized phase for sufficiently small @xmath78 .
we next decrease @xmath22 to @xmath79 , equilibrate the system , and then increase @xmath22 back to @xmath42 , where we again equilibrate the system .
finally we move to @xmath80 , a point at which we know the bosons are condensed .
if @xmath81 is in the confined phase , then the trapped vison will escape ; however , the system may trap visons when re - entering the fractionalized phase .
if an odd number of visons are trapped , there will be a non - zero bosonic current in the condensed phase , while if an even number of vison are trapped , no bosonic current will flow in the condensed phase .
thus we expect to find a current in @xmath82 of the runs . on the other hand ,
if @xmath83 is in the fractionalized phase , then the initial vison will remain trapped , and a non - zero current will be observed in the condensed phase .
figure [ fig : vescape ] shows the fraction of @xmath84 trials in which we observed a trapped vison for @xmath85 and at different values of @xmath79 .
the time to equilibrate the lattice in the condensed phase was highly variable because of false energy minima created by trapping several visons .
the system has probability @xmath86 of trapping an odd number of visons in moving from the confined phase to the fractionalized phase , but when the system is kept in the fractionalized phase the initial vison remains trapped .
the transition is rounded due to visons tunneling out of the finite torus close to the transition .
in this paper we have reported on an investigation of the xy model coupled to a @xmath0 gauge field using monte carlo techniques .
this model was proposed by senthil and fisher for the study of fractionalization . by observing the polyakov loop and the helicity modulus ,
we have determined the structure of the phase diagram of the theory .
in addition , we considered an experiment proposed by senthil and fisher to determine the existence of a fractionalized phase . using phase iii , in which the bosons are condensed , to measure the existence of a vison
, we see that the vison remains trapped in the fractionalized phase .
a vison that was trapped in the phase iii region is able to escape when the system is taken into the confined phase i. we would like to thank m.p.a .
fisher and t. senthil for many insightful discussions .
we would also like to thank r. moessner , s. l. sondhi and j. zannan for their helpful comments regarding the polyakov loop .
this work was supported by department of energy grant 85-er-45197 , national science foundation grant cda96 - 01954 and by silicon graphics inc .
t. senthil and matthew p.a .
fisher , cond - mat/9910224 t. senthil and matthew p.a .
fisher , cond - mat/0006481 t. senthil and matthew p.a .
fisher , cond - mat/0008082 j.b .
kogut , rev .
phys . * 51 * , 659 , ( 1979 ) , and references contained therein .
r. balian , j.m .
drouffe , and c. itzykson , phys .
d * 11 * , 2098 , ( 1975 ) .
e. fradkin and s.h .
shenker , phys .
d * 19 * , 3682 , ( 1979 ) .
m. creutz , phys .
d * 21 * , 1006 , ( 1980 ) .
m. einhorn and r. savit , phys .
d * 17 * , 2583 , ( 1978 ) , and phys .
d * 19 * , 1198 , ( 1979 ) .
jones , j. kogut , and d.k .
sinclair , phys .
d * 19 * , 1882 , ( 1979 ) .
lammert , d.s .
rokhsar , and j. toner , phys .
lett . * 70 * , 1650 ( 1993 ) .
l. mclerran and b. svetitsky , phys .
d * 24 * , 450 ( 1981 ) ying - hong li and s. teitel , phys .
b * 40 * , 9122 ( 1989 ) .
t. senthil and matthew p.a .
fisher ( private communication ) . | we study a model with fractional quantum numbers using monte carlo techniques .
the model is composed of bosons interacting though a @xmath0 gauge field .
we find that the system has three phases : a phase in which the bosons are confined , a fractionalized phase in which the bosons are deconfined , and a phase in which the bosons are condensed .
the deconfined phase has a `` topological '' order due to the degeneracy in the ground state of the gauge field .
we discuss an experimental test proposed by senthil and fisher that uses the topological order to determine the existence of a deconfined , fractionalized phase . |
star clusters are recognized as key probes for tracing the star - forming history ( sfh ) and chemical evolution of galaxy disks .
however , the substantial numbers of young field stars in the galaxy ( e.g. lada & lada 2003 ; bonatto & bica 2011a ) and in nearby galaxies ( e.g. silva - villa & larsen 2011 ) indicate that most clusters are short - lived , and this should be considered when interpreting cluster surveys . those clusters that survive their early evolution
likely do so because of a compbination of initial conditions and the timing of gas ejection ( e.g. smith et al . 2011 ) .
even massive clusters that survive their early evolution may not remain intact for more than a few disk crossing times .
for example , a 10@xmath6 m@xmath7 cluster which is substantially more massive than the majority of clusters in the solar neighborhood may survive for no more than 1.3 gyr in the relatively benign environment of the galactic disk ( e.g. lamers et al .
2005 ) .
once free of their natal clusters , stars can disperse over large distances on cosmologically short timescales .
bastian et al .
( 2011 ) conclude that the timescale for stars to diffuse over galaxy - wide scales in nearby dwarf irregular galaxies is @xmath8 myr .
davidge et al .
( 2011 ) investigate the clustering properties of stars in the outer disk of m33 , and conclude that cluster stars in this environment may disperse over kpc scales on timescales of tens of myr .
the study of star clusters with ages of a few tens of myr will provide clues into the dispersal of stars from their birth environments .
in addition , as demonstrated by bonatto & bica ( 2011b ) , studies of young star clusters that are physically close together , and so may have shared a common birth place , will provide constraints on the sfhs and lifetimes of giant molecular clouds ( gmcs ) . in the present study , moderately deep @xmath9 and @xmath1 images obtained with megacam on the canada - france - hawaii telescope ( cfht ) are used to examine the stellar contents of the young clusters king 12 , ngc 7788 , and ngc 7790 .
these clusters are separated on the sky by only a few tens of arc minutes , and so can be observed with a single megacam pointing . that these clusters are close together on the sky ( projected separation @xmath10 parsecs ) and have ages that agree to within a few tens of myr , opens the possibility that they may have originated in the same star - forming complex , and possibly even the same gmc .
various properties of the clusters , taken from the wedba database ( mermilliod 1995 ) , are summarized in table 1 .
king 12 and ngc 7788 do not have published color - magnitude diagrams ( cmds ) that are based on data obtained with modern detectors .
the sole photometric study of king 12 is that of mohan & pandey ( 1984 ) , who present photoelectric photometry of 30 stars with @xmath11 , and rank king 12 between ngc 2362 and ngc 884 in terms of age .
the only published cmd of ngc 7788 was produced by becker ( 1965 ) from photographic observations .
he measures @xmath12 for this cluster , which is markedly lower than the reddenings of king 12 and ngc 7790 .
ngc 7790 has been the subject of various studies , motivated in large part by the presence of 3 cepheids , making this cluster an important rung in the cosmic distance ladder .
deep imaging of ngc 7790 suggests that @xmath13 , with a distance modulus of @xmath14 and an age in the range 50 100 myr ( romeo et al . 1989 ) .
matthews et al .
( 1995 ) investigate the photometric and radial velocity variations of the cluster cepheid cf cas , and compute a distance of @xmath15 pc ( distance modulus 12.5 ) .
matthews et al .
( 1995 ) also calculate a mass of 5 m@xmath7 for cf cas . adopting this as the main sequence turn - off ( msto )
mass then suggests an age of 60 - 70 myr , although this is an upper limit given that the cepheid progenitor may have experienced mass loss .
the paper is structured as follows .
the observations and the procedures used to produce the final images are described in 2 , while the photometric measurements are discussed in 3 .
the cmds and lfs of the clusters are presented in
4 and 5 , while the spatial distributions of stars in the clusters are investigated in 6 .
a summary and discussion of the results follows in 7 .
the data were recorded with the cfht megacam ( boulade et al .
2003 ) during the night of october 11 , 2010 .
the megacam detector is a mosaic of thirty six @xmath16 e2v ccds that are arranged in a @xmath17 format .
a @xmath18 degree@xmath19 field is imaged exposure@xmath20 with 0.185 arcsec pixel@xmath20 sampling .
there are two 80 arcsec gaps between detector banks , and 13 arcsec gaps between individual ccds .
images were recorded through @xmath21 and @xmath1 filters .
a single 60 sec exposure was recorded in @xmath22 , while single 30 second exposures were recorded in @xmath23 and @xmath1 .
the angular resolution of the final images is 0.8 arcsec fwhm .
the initial processing , which included bias subtraction and flat - fielding , was done with the cfht elixir pipeline .
each elixer - processed megacam image was parsed into six @xmath24 ccd sub - mosaics for subsequent analysis .
experience has shown that such sub - mosaics are a manageable size for subsequent photometric analysis .
chromatic abberations in the megacam optics introduce filter - to - filter positional offsets that increase in size as one moves away from the center of the science field .
differential offsets on the order of an arcsec ( i.e. a few pixels ) were found between @xmath22 and @xmath23 images near the edge of the megacam field , while much smaller offsets were found between the edges of the @xmath1 and @xmath23 images .
these distortions were removed by mapping the @xmath22 and @xmath1 images into the @xmath23 image reference frame using the geomap and geotrans routines in iraf .
photometric measurements were made with the point spread function ( psf)fitting routine allstar ( stetson & harris 1988 ) .
the input source catalogues , preliminary magnitudes , and psfs were obtained by running tasks in the daophot ( stetson 1987 ) package .
the psfs were constructed from isolated bright stars , and faint companions were removed from these stars in an iterative manner using progressively improved versions of the psf .
allstar rejects sources that are not well - matched by the psf and/or that hinder convergence .
still , sources that have poor photometric measurements and/or that are non - stellar may be present in the allstar output . as in previous studies ( e.g. davidge 2010 ) , such objects were identified and removed using the magnitude error computed by daophot , @xmath25 , as a guide .
stars that are neither saturated nor in crowded environments define a monotonic relation on the @xmath26 magnitude diagram ( e.g. figure 2 of davidge 2010 ) , while resolved galaxies , unresolved blends , saturated stars , and cosmetic defects depart from this relation following davidge ( 2010 ) , objects that did not fall on the unblended stellar sequence on the @xmath26 magnitude diagram were deleted from the photometric catalogue .
objects with @xmath27 magnitudes , most of which plumb the faint limits of the photometry , were also rejected .
standard stars are observed during each megacam observing block .
the zeropoints extracted from the standard star observations are then placed in image headers as part of elixer processing .
this standard star information was used to transform the instrumental magnitudes into the sdss @xmath28 system .
the effective wavelengths of the @xmath22 and @xmath29 filters differ by @xmath30 , in the sense that @xmath22 is at a longer wavelength , and this is a source of concern in the @xmath31 transformation .
previous experience with @xmath22 observations of the brightest main sequence stars in nearby galaxies found that a linear @xmath31 transformation is adequate for these objects ( e.g. davidge & puzia 2011 ; davidge et al . 2012 ) .
the veracity of the transformation process was checked in those studies by making comparisons with published photometry .
good agreement was also found between the predicted and observed locus of the upper main sequence on the @xmath32 cmds .
the successful application of a linear @xmath31 transformation in these studies is due to the modest size of the balmer discontinuity in the spectral - energy distributions ( seds ) of massive upper main sequence stars . for the present study ,
a linear @xmath31 transformation was adopted .
intrinsic colors were computed by assuming a single reddening for the entire field .
star - to - star reddening variations are to be expected because ( 1 ) the reddening material is not uniformly distributed across the field , and ( 2 ) the stars in the field are at a range of distances , and so are attenuated by different amounts of dust .
however , experiments indicated that perturbing the adopted @xmath33 by @xmath34 magnitudes did not affect greatly the transformed magnitudes .
clem et al . ( 2008 ) find that a linear @xmath31 transformation may result in large uncertainties at intermediate spectral types , where the balmer discontinuity is most pronounced , and so it is important to check the @xmath29 magnitudes computed here .
to do this , published @xmath35 photometry for king 12 and ngc 7790 from mohan & pandey ( 1984 ) and alcala & arellano ferro ( 1988 ) were transformed into @xmath28 using the relations from smith et al .
( 2002 ) , and the results were compared with the megacam photometry . while ccd measurements are available for ngc 7790 ( e.g. romeo et al .
1989 ) , the photoelectric photometry of alcala & arellano ferro ( 1985 ) are prefered because they include the @xmath36 measurements that are essential for checking the transformed @xmath29 magnitudes .
in 4 it is shown that the megacam and transformed photoelectric measurements mesh together nicely on the @xmath32 and @xmath37 cmds .
model isochrones also provide a reasonable match to the cluster sequences on the @xmath32 cmds over much of the main sequence ( e.g. figure 6 ) .
still , recognizing the potential shortcomings in the @xmath31 transformation , the @xmath29 measurements play a subordinate role to the @xmath23 and @xmath1 data when computing cluster distance moduli and reddenings in 4.2 .
data completeness and the uncertainties in the photometry were assessed in two ways .
first , artificial star experiments were run .
the photometric properties and statistics of the recovered artificial stars can be used to measure the uncertainies in the photometry and calculate sample completeness in a direct way .
second , fields with densities that are comparable to those in the clusters were simulated by stacking sub - sections of the megacam images .
a comparison of the photometric measurements made from the unstacked and stacked images can then be used to assess the impact of crowding in a differential manner .
artificial stars were assigned colors and brightnesses that fall along the field star main sequence .
as with the actual observations , an artificial star was considered to be detected only if it was recovered in at least two filters ( either @xmath38 and @xmath39 ) .
the results from allstar were filtered with the @xmath25-based criteria that were applied to the science frames .
the artificial star experiments indicate that the data are complete when @xmath40 and @xmath41 , while 50% completeness occurs near @xmath42 and @xmath43 .
the random and systematic uncertainties in the photometry climb rapidly towards fainter magnitudes when sample completeness drops below 50% , and so the point at which 50% completeness occurs is a pragmatic estimate of the faint limit of data .
this is demonstrated in the next section where it is shown that the 50% completeness curve falls just above the lower envelope of data on the cmds .
the clusters have stellar densities that are a few times higher than in the field , and so the faint limit of the cluster photometry differs from that of the surrounding field . to assess the size of this difference
, fields with stellar densities that are three times higher than is typical throughout most of the megacam image were simulated by stacking three @xmath44 pixel sub - sections of the megacam images .
the image sub - sections selected for this experiment do not include the clusters .
source brightnesses were then measured in the stacked images using the procedures described in 3.1 .
the composite @xmath37 cmd and @xmath1 lf of objects in the three sub - fields are compared in figure 1 with the cmd and lf obtained from the stacked images .
significant differences between the two cmds and lfs become evident between @xmath2 and 21 .
it is evident that tripling the stellar density affects the photometry when @xmath45 , and so @xmath2 is adopted as the faint limit for the cluster observations .
the @xmath32 and @xmath37 cmds of king 12 , ngc 7788 , ngc 7790 are shown in the top panels of figures 2 and 3 .
the king 12 and ngc 7788 cmds include sources that are within 90 arcsec of the cluster centers , which is comparable to the cluster radii measured in 6 .
the ngc 7790 cmd includes sources within 180 arcsec of the cluster center .
while ngc 7790 extends out to larger radii ( 6 ) , it was decided to limit coverage to this region to reduce contamination from non - cluster members .
the @xmath46 uncertainties in @xmath47 and @xmath48 computed from the artificial star experiments are shown in the upper right hand panel of each figure .
a line that corresponds to 50% completeness is also shown in the ngc 7790 panels , and this falls just above the faint envelope of the ngc 7790 cmds .
the faint limit of these data is set by the @xmath29 measurements in figure 2 , and the @xmath23 measurements in figure 3 . while not shown in figures 2 and 3 ,
the artificial star experiments indicate that the data are 100% complete roughly 1 magnitude above the 50% completeness relation .
stars in the megacam images saturate near @xmath49 , and this produces a sharp cut - off at the bright end of the observations that prevents the msto from being sampled .
therefore , the cluster sequences in figures 2 and 3 were extended to brighter magnitudes by adding photoelectric observations from mohan & pandey ( 1984 ) for king 12 and alcala & arellano ferro ( 1988 ) for ngc 7790 , which were transformed into the sdss system as discussed in 3.2 . only stars with @xmath50 from these studies
are shown , as there is a marked increase in scatter at fainter magnitudes among the photoelectric measurements from both studies .
corresponding observations of ngc 7788 made with linear detectors are not available in the literature .
a prominent blue main sequence with @xmath51 and @xmath52 dominates the upper half of each cluster cmd .
the vertical arm of the main sequence in the upper portions of the cmds have similar @xmath47 and @xmath48 colors , which in turn suggests that the cluster reddenings are similar .
this foreshadows one result of this paper that the reddening of ngc 7788 has heretofore been underestimated .
the clusters are at very low galactic latitudes and with the exception of the brightest cluster members contamination from stars along the line of sight complicates efforts to identify cluster sequences on the cmds .
the extent of field star contamination is examined in the bottom rows of figures 2 and 3 , where the cmds of sources in annuli surrounding the clusters are shown .
the control fields have an inner radius that is @xmath53 that of the radius used to construct the cmds in the top rows of figures 2 and 3 , and an outer radius that allows the same total area on the sky as in the cluster to be sampled ; thus , star counts in the cmds in the lower panels of figures 2 and 3 can be used to estimate directly field star contamination in the cluster cmds . given the large angular size of ngc 7790 ( 6 ) ,
some stars that belong to ngc 7790 are undoubtedly present in the control field .
the relative numbers of blue main sequence stars in the cmds of the various cluster and control fields suggest that the clusters have a stellar density that is @xmath54 that in the surrounding field .
it is difficult to track the cluster sequences in the lower half of each cmd .
this is because the cluster and field star sequences in the cmds overlap at intermediate magnitudes .
in addition , while the number of field main sequence stars grows towards fainter magnitudes as progressively larger volumes of the disk are sampled , the cluster lfs are such that the number of cluster stars per magnitude interval does not change greatly with @xmath1 ( 5 ) . finally , age plays a role in explaining the dearth of faint cluster members in the king 12 cmd , as the lower part of the main sequence is not yet populated in this young cluster ( see below ) .
the cluster cmds are compared with isochrones from girardi et al .
( 2004 ) in figure 4 .
the distances and reddenings listed in table 1 have been adopted for this initial comparison .
the relations provided in table 6 of schlegel et al .
( 1998 ) , which draw on the extinction laws of cardelli et al .
( 1989 ) and odonnell ( 1994 ) , were used to redden the models .
isochrones with z = 0.019 and ages of 10 , 20 , 40 , and 80 myr are shown . a 80 myr sequence with z = 0.030 is also plotted to allow metallicity effects to be assessed .
a clear result from figure 4 is that the reddening for ngc 7788 in the wedba database is too small , as the models fail to match the vertical plume of main sequence stars with @xmath55 .
it is also apparent that the ages of these clusters are difficult to establish given the modest number of intrinsically bright stars .
this holds even after including the photoelectric observations of bright stars in king 12 and ngc 7790 .
the reddenings and distance moduli used in figure 4 can be tuned to obtain better agreement between the models and observations , and the results of such an exercise are shown in figure 5 .
an initial reddening for each cluster was selected so that the isochrones match the blue envelope of the vertical part of the main sequence with @xmath55 on the @xmath37 cmd .
the distance modulus was then adjusted with the reddening fixed and the best match with the lower envelope of the cluster sequence near the elbow in the cluster cmd near @xmath56 was found . a revised reddening estimate
was then made , and the procedure was iterated until convergence .
the revised distance modulus and color excess for each cluster are listed in figure 5 above the cluster cmds .
the estimated uncertainties in the distance moduli are @xmath57 , while the uncertainties in the reddenings are @xmath34 magnitude .
the distance modulus found for ngc 7790 agrees with other recent estimates , which are summarized in table 2 .
in contrast , the distance moduli of king 12 and ngc 7788 differ significantly from the values in the wedba database .
while the @xmath32 cmds provide the tightest constraints on cluster ages ( see below ) , preliminary ages can be obtained from figure 5 for king 12 and ngc 7790 . based largely on the three stars with @xmath58 from mohan & pandey ( 1984 ) , the comparisons in figure 5 suggest that king 12 has an age @xmath59 myr . as for ngc 7790 ,
the brightest star in the alcala & arrelano ferro ( 1988 ) observations is consistent with an age near 30 myr , although the majority of fainter data points are consistent with an older age .
an age can not be estimated for ngc 7788 based solely on the megacam @xmath23 and @xmath1 data , although the situation is different when the @xmath29 measurements are considered ( see below ) . with the distance moduli and reddenings set from the @xmath37 cmds , additional insights into the cluster ages
can be obtained from the @xmath32 cmds , and isochrones are compared with the @xmath32 cmd in figure 6 .
the @xmath29 photometry provides an independent check of the reddening , while the @xmath47 color is more sensitive to age and metallicity than @xmath48 .
this being said , the @xmath47 color can be difficult to reproduce in model isochrones , with uncertainties of up to a tenth of a magnitude ( e.g. an et al . 2009 ) .
the bluest stars in the @xmath32 cmd of king 12 are consistent with an age @xmath0 myr . for ngc 7790 ,
the photoelectric measurements with @xmath50 suggest an age between 60 and 80 myr . finally , given the increased age sensitivity of @xmath47 among intermediate mass stars it is possible to estimate an age for ngc 7788 , and the megacam data are consistent with an age of 20 40 myr .
this is the least reliable age estimate in this study , and photometry of brighter stars in this cluster are needed to confirm the age of ngc 7788 .
star clusters as young as king 12 and ngc 7788 should harbor a population of pre - main sequence ( pms ) objects , and these will fall redward of the main sequence on the cmds . in fact , some parts of the king 12 and ngc 7788 @xmath37 cmds , indicated in figure 3 , contain an excess number of objects when compared with the control fields .
the area of the king 12 and ngc 7788 cmds indicated in figure 3 will be refered to as the ` pms box ' in subsequent discussion .
the pms box in king 12 contains 40 objects , while there are 32 objects in ngc 7788 . for comparison ,
each of the control fields contain 18 objects in the pms box .
the excellent agreement between the number density of these objects in the two control fields indicates that outside of the clusters the objects in the pms box are uniformly distributed , as expected if they are non - cluster field objects . after subtracting the number of objects in the control fields , the content of the pms box differs from zero at the @xmath60 level in king 12 , and the @xmath61 level in ngc 7788 .
the lower portions of the cluster cmds are compared with models of pms evolution from tognelli et al .
( 2011 ) in figure 7 .
there are considerable uncertainties in models of pms evolution , and these have been reviewed by siess ( 2001 ) .
the largest single source of uncertainty in the input physics is the recipe used to account for convection , although the equation of state , boundary conditions , rotation , and mass accretion will also affect the models .
the translation of models onto the observational plane are also prone to uncertainties in the bolometric corrections , the t@xmath62 scale , abundances , and circumstellar extinction , although the latter likely affects the observational properties of pms objects only for systems with ages @xmath63 myr ( baume et al .
2003 ) . the luminosity @xmath64 effective temperature ( t@xmath62 ) pairs tabulated by tognelli et al .
( 2011 ) were transformed onto the observational plane using relations between t@xmath62 , broad - band colors , and bolometric corrections derived from main sequence stars .
the use of relations that are based on main sequence stars is not ideal , as the seds of pms stars differ from those of main sequence stars .
in addition , no allowance was made for circumstellar extinction , although the good agreement between age estimates based on pms stars and more evolved main sequence stars in young cluster such as ngc 2362 ( e.g. moitinho et al .
2001 ) and ngc 3293 ( baume et al . 2003 ) suggests that natal dust envelopes dissipate on timescales that are shorter than the ages of king 12 and ngc 7788 .
indeed , observations and dynamical arguments both suggest that circumstellar disks likely dissipate over timescales of @xmath65 myr ( mayne et al .
2007 ) .
the properties of pms models depend on factors including metallicity , mixing length ( @xmath66 ) , helium abundance , and the initial deuterium abundance ( @xmath67 ) .
the models in figure 7 as well as those used later in this paper assume z = 0.015 , y=0.278 , @xmath68 , and @xmath69 . despite the inherent uncertainties in the transformation procedure coupled with the sensitivity to input physical parameters
the pms models pass through the portions of the king 12 and ngc 7788 cmds that contain an excess population of red objects , suggesting that pms objects have been detected in these clusters .
the models also predict that pms stars in ngc 7790 with an age @xmath70 myr may also lurk in the megacam images .
such objects would be fainter ( and hence more difficult to detect ) than their closer and younger brethren in king 12 and ngc 7788 .
in fact , the ngc 7790 lf appears to be fully populated to @xmath71 ( 5 ) , suggesting that pms stars in that cluster are fainter than @xmath2 .
the @xmath1 lfs of objects in king 12 , ngc 7788 , and ngc 7790 are shown in figure 8 .
the lfs are restricted to @xmath72 , which is the magnitude range that the stacking experiments discussed in 3.3 indicate is not affected by crowding .
given this cut - off then main sequence stars in these clusters with masses @xmath73 are sampled .
the lfs have been corrected for field star contamination by subtracting star counts over the entire megacam field , but excluding the areas occupied by the clusters .
drawing on number counts over such a large area reduces statistical errors , so that the random uncertainties in the final cluster lfs mainly reflect statistical flucuations in the cluster stellar contents .
a potential shortcoming of estimating contamination from source counts that cover a large area near the disk plane is that reddening variations may affect the results .
still , the overall trends of the background - subtracted lfs in figure 8 do not change greatly if number counts taken from the control fields shown in figure 3 , which are less prone to large - scale reddening variations given their modest size and physical proximity to the clusters , are used instead of those based on the entire megacam field .
the lfs of ngc 7788 and ngc 7790 in figure 8 have broadly similar appearances , with a more - or - less flat section when @xmath74 , with the break near @xmath56 due to an inflexion point in the mass - magnitude relation among main sequence stars . in contrast
, there is an apparent deficiency of stars with @xmath1 between 16.5 and 19 in king 12 when compared with other clusters .
the ratio of sources at the faint and bright end of the king 12 lf is also larger than in the other clusters .
the dotted line in figure 8 is a model lf constructed from the z = 0.019 marigo et al .
( 2008 ) sequences , which was generated using the interface at the _ padova database of evolutionary tracks and isochrones _ web site .
the model assumes an age of 30 myr .
while the clusters span a range of ages , it should be recalled that the megacam data sample only stars that are below the msto , and so age effects near the bright end of the lfs in figure 8 should be negligible
. a kroupa ( 1998 ) imf was adopted , and the model in each panel was scaled to match the cluster lf between @xmath75 and 16 .
the model lf does not include binaries .
simulations indicate that the feature used below to estimate cluster ages the lower main sequence cut - off is not sensitive to binarity ( cignoni et al . 2010 ) .
the model lf does not match the king 12 observations very well , with number counts at some intermediate magnitudes differing from the model at more than the @xmath76 level . such a disgreement is perhaps not unexpected , as the main sequence will not be fully populated at intermediate masses in young clusters like king 12 , and the magnitude at which such departures occur provides information about cluster age .
mayne et al . (
2007 ) compare main sequence and pms sequences in a number of open clusters , and identify a gap that is associated with the transition from a fully convective pms object to one with a radiative core .
the rate at which objects traverse this gap depends on the mass of the protostar , in the sense that more massive stars traverse the gap faster .
the gap will appear as a discontinuity in the cluster lf .
the model provides a better match to the lfs of ngc 7788 and ngc 7790 , and the overall agreement between the model and the cluster lfs near the bright end provides a ( loose ) consistency check on their distance moduli .
still , the ngc 7788 lf departs from the model lf at @xmath77 , or m@xmath78 .
the simulations discussed in 3.3 suggest that this is not due to incompleteness .
the integrated lfs of the clusters will provide additional insights into any differences between the observed and model brightness distributions in king 12 and ngc 7788 .
therefore , the differential lfs shown in figure 8 were integrated , and the differences between the observed and model lfs of these clusters
@xmath79 are shown in figure 9 . a deficiency in the observed number of stars with respect to the model produces a positive value of @xmath79 .
in the case of king 12 , @xmath79 differs from zero with a significance in excess of the @xmath76 level at @xmath80 , and at fainter @xmath81 is persistently positive .
this result , coupled with the comparison between the diffential lfs of king 12 and the model in figure 8 , suggests that the onset of the pms occurs at @xmath82 in king 12 .
this corresponds to m@xmath83 , and the mass - magnitude relation for main sequence stars shown in table viii of habets & heintze ( 1981 ) indicates that main sequence stars with this m@xmath84 have a mass @xmath85 m@xmath7 .
this lower mass limit for main sequence stars can be used to estimate the age of king 12 . adopting the same parameters as those used to generate the pms isochrones shown in figure 7 , the tognelli et al .
( 2011 ) models predict that a star with a mass @xmath85 m@xmath7 will take @xmath86 myr to relax onto the main sequence , and this is adopted as the pms - based age for king 12 .
the age calculated from the main sequence cut - off thus falls within the range that was obtained from the cmds in 4.2 .
with such an age , the red objects at the faint end of the king 12 lf in figure 8 are likely to be pms stars , and in 6 it is demonstrated that the spatial distribution of these objects is consistent with cluster membership .
there is a substantial increase in @xmath79 near @xmath87 in the upper panel of figure 9 could this be the onset of the pms in king 12 rather than the more modest feature discussed above ? in order for this to be the case
then the non - zero @xmath79 at @xmath88 would have to be a statistical fluke .
if the onset of the pms occurs at @xmath87 then stars as massive as 1.2 m@xmath7 are on the king 12 main sequence .
the age of king 12 based on the main sequence cut - off is then @xmath89 myr , which still agrees with the upper end of the range predicted by bright main sequence stars .
the evidence for the onset of a pms sequence in the lf of ngc 7788 is not as robust as in king 12 .
it is evident from the bottom panel of figure 9 that the the observed and model integrated lfs of this cluster may start to diverge near @xmath90 , although @xmath79 only differs from zero at the @xmath91 level at this @xmath1 .
still , it is encouraging that @xmath92 when @xmath93 , and so @xmath94 is adopted as the tentative onset of the pms in ngc 7788 .
this corresponds to m@xmath95 , and the habets & heintze ( 1981 ) calibration indicates that a main sequence star with this m@xmath84 has a mass @xmath96 m@xmath7 . the tognelli et al .
( 2011 ) models predict that such an object will take @xmath97 myr to relax onto the main sequence .
this age estimate favors the old end of the age range for ngc 7788 measured in 4.2 .
there is not a statistically significant dfference between the observed and model lfs of ngc 7790 when @xmath98 , which corresponds to m@xmath99 .
this suggests that the main sequence of this cluster is populated by stars with masses @xmath100 m@xmath7 .
the tognelli et al .
( 2011 ) models then predict that ngc 7790 should be older than @xmath101 myr , and this is consistent with the age of this cluster estimated in 4.2 .
the large - scale clustering properties of objects in king 12 , ngc 7788 , and ngc 7790 are investigated using the two - point correlation function ( tpcf ) .
the tpcf is based on a simple observable - the separation between two points on the sky and so the interpretation of the tpcf is not complex .
most previous applications of the tpcf to star clusters examined spatial scales that sample the transition between multiple star systems ( binaries , triples , etc ) and the larger - scale clustering of such systems ( e.g. gomez et al . 1993 ; larson 1995 , simon 1997 ) .
larson ( 1995 ) finds that the clustering properties of stars in the taurus and ophiucus star - forming regions changes at separations @xmath102 parsecs , and notes that this spatial scale is comparable to the jeans length in molecular clouds .
the separation at which the correlation properties change can then be interpreted as the maximum spatial extent of individual stellar systems that form within a single proto - stellar envelope .
power in the correlation spectrum at larger spatial scales is due to clustering between individual star systems .
three samples of objects are defined for this investigation .
the first consists of bright main sequence stars ( bmss ) , which have @xmath1 between 14 and 16 , and @xmath48 between 0 and 0.7 .
the second consists of red plume sources ( rpss ) , which have @xmath103 and @xmath48 between 1.2 and 2.5 .
the extraction region for the rps sample was selected to avoid possible pms stars in king 12 and ngc 7788 ( 5 ) .
the third sample consists of candidate pms objects .
this sample is restricted to king 12 and ngc 7788 , and contains objects in the pms region of the @xmath37 cmd indicated in figure 3 .
these samples probe very different populations .
the intermediate mass main sequence stars that make up the bms sample are present in all three clusters ( e.g. figure 3 ) , and so provide a homogeneous basis for comparing cluster properties .
the objects in the pms sample are a mix of cluster and non - cluster objects , and this complicates efforts to compare their spatial distribution with other objects ( see below ) . as for the rpss
, these isotropically distributed objects are not cluster members , and so are used to normalize the tpcfs .
to the extent that the sources in the bms , pms , and rps samples occupy comparable areas of the science detector , then this normalization corrects for issues arising from detector geometry , such as gaps between detectors and finite area coverage .
the tpcfs obtained by ratioing the separation functions ( sfs ) of bms and rps sources within a 9 arcmin radius of the three clusters are shown in figure 10 .
the curves in figure 10 have been normalized based on the number of objects in each sample .
the edge of the megacam science field is 6 arcmin from the center of ngc 7790 , and so there is incomplete spatial coverage of this cluster for separations @xmath104 arcmin .
however , as both the bms and rps samples are vignetted , then the impact of incomplete spatial coverage on the shape of the tpcf is removed .
while there is not a discontinuity near 12 arcmin separations in the ngc 7790 tpcf , the shape of the tpcf of ngc 7790 at separations in excess of 12 arcmin will be affected by incomplete spatial coverage if the cluster is not azimuthally symmetric on the sky - that is , any sub - structures in the southern part of the cluster will be missed .
following simon ( 1997 ) , jackknife re - sampling was used to assess uncertainties in the correlation functions .
this approach automatically tracks the influence of stars over the full range of separations .
results of re - sampling experiments for ngc 7788 are shown in the lower panel of figure 10 .
each realization shows the tpcf that is obtained after randomly extracting 25% of the sources in the bms and rps samples .
a number of broad conclusions can be reached from the re - sampling experiments .
first , the dispersion amongst the tpcfs increases towards smaller separations . this is to be expected given that the number of possible object - object pairings drops as one moves to smaller separations , leading to greater sensitivity to stochastic flucuations .
a similar affect occurs at very large separations .
second , the dispersion in the re - sampled ngc 7788 tpcfs is comparable to the bin - to - bin chatter in the full ngc 7788 sample .
the local bin - to - bin chatter thus provides an estimate of the random uncertainty in the tpcf at a given separation .
finally , the overall shapes of the tpcfs in the lower panel of figure 10 are similar .
this indicates that the structural features identified in the tpcf of ngc 7788 ( see below ) are robust detections .
if the separation distributions of two uniformly distributed populations are ratioed then the result will be a tpcf that is flat , with bin - to - bin noise introduced by stochastic variations in the spatial distributions of the stars .
the tpcfs of the clusters in figure 10 are not flat , indicating that the bms objects are not uniformly distributed .
rather , the tpcfs of all three clusters show a general trend of increasing coherence towards smaller separations , which is a consequence of the central concentration of stars in the clusters .
however , this broad similarity aside , the tpcfs of king 12 and ngc 7788 differ from those of ngc 7790 , and this is likely due to structural differences
. the tpcf of an isolated cluster with a well - defined boundary that is surrounded by a uniformly distributed field population will at separations that exceed the cluster diameter
be flat out to the largest separations probed by the sample .
this is clearly not the case for king 12 and ngc 7788 , as the tpcfs of these clusters outside of their central peaks have distinct plateaus . the first plateau , labelled ` halo ' in the ngc 7788 panel , likely results from a population of bms objects that are not as tightly clustered as objects in the main body of the star cluster , but still have a source density that exceeds that at larger separations .
the halo component extends out to 10 arcmin separations ( 5 arcmin radius ) around both clusters .
the tpcf then drops to a second plateau at larger separations , labelled ` field ' in the ngc 7788 panel , where the bms and rps objects are again uniformly distributed , but with a lower density than in the halo .
the inflexion points at small separations , where the peak in the cluster tpcf blends into the surrounding halo , indicate that the main body of king 12 has a radius of @xmath18 arcmin , while the main body of ngc 7788 has a radius of 1.5 arcmin .
the tpcf of ngc 7790 does not have the prominent peak at small separations that is the signature of a central density concentration ; rather , the tpcf of ngc 7790 is flat at separations @xmath105 arcmin . given that the separation statistics are dominated by cluster objects , then this suggests that bms objects are more - or - less uniformly distributed within 2.5 arcmin of the center of ngc 7790 .
the tpcf of ngc 7790 drops steadily for separations between 5 and 15 arcmin , and a uniform distribution of bms stars is encountered at separations in excess of 15 arcmin . for comparison , gupta et al .
( 2000 ) measure a core radius of 3.7 arcmin for ngc 7790 , and find that cluster stars largely disappear at @xmath53 this value .
there is no evidence of a halo around ngc 7790 .
if ngc 7790 has a uniform - density halo then it must be located at much larger distances from the cluster center than is the case for the other two clusters .
in addition , the ngc 7790 tpcf at large radii has a level of coherence that is comparable to the tpcfs of the field regions around the two other clusters .
if there is a halo surrounding ngc 7790 then it must have a much lower density relative to the main body of the cluster than in either king 12 or ngc 7788 .
the clustering properties of objects in the king 12 and ngc 7788 pms samples are investigated in figure 11 .
the top row of this figure shows the tpcfs obtained by ratioing the sfs of the pms and rps samples , while the lower row shows the ratio of the sfs of pms objects and bms stars in each cluster . with the possible exception of separations @xmath106 arcmin ,
pms objects in king 12 have a roughly uniform distribution for separations @xmath107 arcmin . at larger separations
the degree of clustering drops .
this is similar to the clustering properties of bms stars in king 12 , providing additional evidence that the red sources detected in king 12 are pms sources .
pms stars should have a lower mass than objects that are already on the main sequence . that pms stars are clustered out to 10 arcmin separations in king 12
indicates that the halo around this cluster ( 1 ) is populated by stars having a range of masses , and ( 2 ) has a radial extent that is similar for objects of all masses .
still , the ratio of the pms and bms sfs in the lower left hand panel of figure 11 indicates that the objects in these samples have different clustering properties .
this may be due in part to different levels of contamination from non - cluster sources in each sample .
in contrast to king 12 , the tpcf of pms objects in ngc 7788 is more - or - less flat over all separations , indicating that the candidate pms objects are uniformly distributed , with no evidence of clustering .
the ratio of pms and bms sfs in the vicinity of ngc 7788 , shown in the lower right hand panel of figure 11 , indicates that the objects in these samples have very different clustering properties .
the size of the pms sample in ngc 7788 has a lower statistical significance than in king 12 ( @xmath76 versus @xmath108
4.3 ) , and there is a higher fractional contribution from non - cluster objects .
contamination from non - cluster objects , which are uniformly distributed , will cause the tpcf to become flatter .
a deep emission line survey of ngc 7788 will help assess the spatial distribution of any pms stars in this cluster .
images obtained with the cfht megacam have been used to investigate the young star clusters king 12 , ngc 7788 , and ngc 7790 .
these clusters are at distances of 2 3 kpc and have angular sizes of at least a few arcmin .
they are also close together on the sky , and all three can be observed with a single megacam pointing . neither king 12 nor ngc 7788 have been extensively studied in the past .
the properties of main sequence stars and supected proto - stars have been explored in these clusters . in particular , the megacam data have been used ( 1 ) to estimates ages based on stars that are on the upper and lower parts of the main sequence , ( 2 ) to estimate distances , and ( 3 ) to explore the spatial distribution of stars in these clusters .
the properties of faint main sequence and pre - main sequence stars , the implications of the cluster distances and ages on a possible common origin , and the structural properties of the clusters are discussed in turn in the following sections .
ages have been estimated for all three clusters using the photometric properties of stars near the msto .
comparisons with isochrones suggest that ngc 7790 has an age 60 80 myr , and this is consistent with the age infered from the mass of cf cas ( matthews et al .
1995 ) , with the assumption that the cepheid has experienced no significant mass loss during its evolution ( 1 ) .
the properties of stars near the mstos of king 12 and ngc 7788 suggest an age @xmath0 myr for the former and 20 40 myr for the latter .
the msto is not an ironclad chronometer for low mass clusters , as the morphology of cluster sequences near the msto on the cmds of such systems are susceptible to stochastic effects .
unresolved binaries will also complicate efforts to identify the msto .
stochastics effects are much less of an issue for lower mass cluster members , as these objects are more numerous than the most massive cluster members .
for this reason , the main sequence cut - off is a potentially powerful alternate means of measuring ages in young , spatially diffuse , low mass clusters .
the use of the main sequence cut - off to estimate cluster ages is not new , and belikov et al .
( 1997 ) , mayne et al .
( 2007 ) , and cignoni et al .
( 2010 ) have used discontinuities in the magnitude distribution of cluster stars to estimate ages .
the amplitude of the discontinuities in lfs depends on cluster age ( e.g. mayne et al .
the feature found by belikov et al .
( 1997 ) in the lf of the pleiades , which has an age @xmath8 myr , occurs near m@xmath109 and has a modest size , whereas the models examined by cignoni et al .
( 2010 ) predict features of a much more substantial size for systems with ages @xmath110 myr .
cignoni et al .
( 2010 ) suggest that main sequence cut - off / pms onset signatures in lfs are most useful as age estimators in systems that are younger than @xmath111 myr .
the main sequence cut - off is not without its difficulties as an age indicator .
there are uncertainties in the models of pms evolution ( e.g. siess 2001 ) , and the identification of the lower end of the main sequence may be complicated by contamination by non - cluster members .
another potential complication is that the stellar content of a cluster may not be coeval , and age dispersions of a few myr may be present ( e.g. da rio et al . 2010 ) . the pms stars that are only now relaxing onto the main sequence may then be a few myr younger than the more massive stars near the msto .
this being said , good agreement between upper and lower main sequence age estimates have been found in studies of ngc 2362 ( moitinho et al .
2001 ; dahm 2005 ; delgado et al .
2006 ) and ngc 3293 ( baume et al . 2003 ) .
ages have been estimated for king 12 and ngc 7788 using the properties of the faintest main sequence stars .
given the various uncertainties in models of pms evolution ( 4.3 ) , it is encouraging that the ages estimated for these cluster from features at intermediate and faint magnitudes in the @xmath1 lf do not differ greatly from those obtained from stars near the msto .
for king 12 , the main sequence cut - off near @xmath88 yields an age @xmath86 myr , which is consistent with the msto age estimate of @xmath0 myr .
even if the main sequence cut - off in king 12 is 1 magnitude fainter ( 5 ) , then the age estimated for king 12 increases by only 7 myr . as for ngc 7788 , the lower end of the main sequence yields an age @xmath97 myr , whereas upper main sequence stars give an age in the range 20 40 myr .
we close this section by noting that candidate pms stars have been detected in both king 12 and ngc 7788 .
the detection of pms stars is most secure in king 12 , as these objects have large - scale distributions that are similar to those of bright main sequence stars ( 6 ) .
if these objects are pms stars then they should be sources of h@xmath66 emission , and this can be checked directly by obtaining moderately deep narrow - band images of each cluster .
a population of pms stars should also be present in ngc 7790 , although these objects likely have @xmath45 , and so are below the faint limit of the megacam data .
prior to the current study , some properties of king 12 , ngc 7788 , and ngc 7790 hinted at a possible common origin , either in the same gmc or star - forming complex .
in particular , the projected separations of these clusters is only @xmath101 parsecs , and previous studies yielded ages that agree to within a few tens of myr . neither the projected distribution of the clusters on the sky nor the large age range rule out an origin in the same gmc .
random motions in molecular clouds are a few km sec@xmath20 ( solomon et al . 1987 ) , and so clusters that formed in the same gmc might disperse over @xmath8 parsecs after 30 myr .
in addition , a typical gmc may survive for @xmath112 myr ( roughly three free - fall times murray 2011 ) , suggesting that gmcs have the potential to form clusters over timescales of tens of myr .
the possibility that these clusters may have originated in the same gmc and the prospect of probing the sfh of that gmc was the prime motivation for obtaining the megacam images used in this study .
however , the distances measured in the current work rule out these clusters forming in the same gmc .
king 12 and ngc 7788 are separated by @xmath113 parsecs along the line of sight whereas ngc 7790 is @xmath114 kpc more distant than king 12 . still ,
while an origin in the same gmc is ruled out , these clusters may have formed in the same star - forming complex .
indeed , large star - forming complexes may extend over many hundred of parsecs ( e.g. review by efremov 1995 ) , and can have sfhs that span many tens of myr ( e.g. garcia - benito et al .
2011 ) .
while the tpcf has not traditionally been used to probe the _ large - scale _ structure of star clusters ( but see sanchez & alfaro 2009 ) , it is used here to investigate the distribution of stars in all three clusters out to separations of 20 arcmin .
the tpcf multiplexes information from all objects in a sample , thereby making efficient use of the available information .
the tpcf is also based on a simple observable the separation between two points on the sky providing an intuitive basis for interpreting the results . the star - star separation function ( s3f ) is a related statistic , which davidge et al . (
2011 ) used to investigate the diffusion timescales of stars from star - forming complexes in the outer disk of m33 .
the tpcfs of king 12 and ngc 7788 indicate that bms stars in both clusters have compact cores that are confined to the central @xmath115 parsecs of each cluster .
the main bodies of these clusters are in turn surrounded by bms stars that have a uniform distribution on the sky .
these halos extend out to @xmath116 arcmin from the cluster centers , which is a few times the radius of the main body of the clusters .
as the oldest cluster , ngc 7790 is likely to be more dynamically evolved than king 12 and ngc 7788 , which in turn may affect the core regions .
bastian et al .
( 2008 ) investigate the structural properties of young star clusters , and find that the structure of the core may change with time , with clusters having ages @xmath0 myr showing the greatest range of structural diversity when compared with younger clusters .
indeed , sanchez & alfaro ( 2009 ) find that clusters that are as old as 100 myr may still contain sub - structures that have not yet been obliterated by dynamical processes , although other clusters with this age show no evidence for such sub - structures .
the tpcf of ngc 7790 clearly differs from those of king 12 and ngc 7788 , in that ngc 7790 has a much larger angular size and there is no obvious central core .
if ngc 7790 once had a core then it has since dissipated . where did the bms stars in the halos that surround king 12 and ngc 7788 originate
? the extended halos are probably not the product of dynamical relaxation .
signatures of mass segregation might be expected to develop after a few cluster crossing times , and prisinzano et al .
( 2003 ) find evidence for mass segregation in the @xmath8 myr cluster ngc 2422 , in the form of a spatially diffuse population of low mass main sequence stars .
the lfs of main sequence stars in ngc 7788 and ngc 7790 are similar in shape over most of the magnitude range probed here , and are well matched by models that are based on the kroupa ( 1998 ) solar neighborhood mass function . that the lfs of ngc 7788 and ngc 7790 , which have different ages and physical sizes , are consistent with the solar neighborhood mass function suggests that dynamical evolution has not yet affected the spatial distribution of objects with masses @xmath117 m@xmath7 in these clusters .
this indicates that the bms stars that are used to construct the tpcf , which are among the most massive stars in these clusters , have not been displaced to large radii by mass segregation .
the possibility that the halo stars may have formed _ in situ _ in the peripheral regions of the density enhancement that evolved into the main bodies of the clusters seems unlikely . with their uniform density distribution ,
the halos are structurally distinct from the central cluster .
star formation has a fractal distribution over spatial scales in excess of 10 parsecs ( e.g. gomez et al .
1993 ; references in elmegreen & elmegreen 2001 ) , and there is no evidence for the formation of distinct , large - scale halo components in star - forming environments .
we suggest that the halo stars aroung king 12 and ngc 7788 originated at smaller radii in the clusters and migrated outwards , possibly in response to the rapid expulsion of large quantities of star - forming material earlier in the evolution of the cluster . while once thought to be catastrophic for cluster survival , simulations suggest that the loss of substantial amounts of gas may not lead to the immediate disruption of a young cluster .
rather , stars that disperse from a cluster due to such an event may linger for @xmath118 myr in a low surface brightness region surrounding the cluster remnant ( e.g. bastian & goodwin 2006 ) . with an age in excess of 30 myr
then if ngc 7790 once had a halo of this nature then it may have by now dissipated into the surrounding field ( bastian & goodwin 2006 ) .
the large - scale spatial distributions of objects that have very different masses may provide clues into the dynamical state of the cluster at the time of gas loss .
objects in the king 12 pms and bms samples have similar clustering properties , with a uniform distribution of objects in both samples for separations between 5 and 10 arcmin ( figure 11 ) , and a drop in coherence for separations @xmath119 arcmin . despite having very different masses ,
the similarity in clustering properties at separations associated with the cluster halo suggests that the objects in the pms and bms samples had similar kinematic properties at the time of gas ejection .
this is consistent with king 12 still experiencing dynamical evolution at the time of gas loss , in agreement with simulations of early cluster evolution discussed by delgado et al .
( 2011 ) .
ccc e(@xmath120 ) & @xmath121 & reference + @xmath122 & @xmath123 & gupta et al .
( 2000 ) + @xmath124 & @xmath125 & lee & lee ( 1999 ) + & @xmath126 & mathews et al .
( 1995 ) + @xmath127 & @xmath14 & romeo et al .
( 1989 ) + & & + @xmath128 & @xmath129 & current study + | the stellar contents of the open clusters king 12 , ngc 7788 , and ngc 7790 are investigated using megacam images .
comparisons with isochrones yield an age @xmath0 myr for king 12 , 2040 myr for ngc 7788 , and 60 80 myr for ngc 7790 based on the properties of stars near the main sequence turn - off ( msto ) in each cluster .
the reddening of ngc 7788 is much larger than previously estimated .
the luminosity functions ( lfs ) of king 12 and ngc 7788 show breaks that are attributed to the onset of pre - main sequence ( pms ) objects , and comparisons with models of pms evolution yield ages that are consistent with those measured from stars near the msto .
in contrast , the @xmath1 lf of main sequence stars in ngc 7790 is matched to @xmath2 by a model that is based on the solar neighborhood mass function .
the structural properties of all three clusters are investigated by examining the two - point angular correlation function of blue main sequence stars .
king 12 and ngc 7788 are each surrounded by a stellar halo that extends out to 5 arcmin ( @xmath3 parsecs ) radius .
it is suggested that these halos form in response to large - scale mass ejection early in the evolution of the clusters , as predicted by models . in contrast , blue main sequence stars in ngc 7790 are traced out to a radius of @xmath4 arcmin ( @xmath5 parsecs ) , with no evidence of a halo .
it is suggested that all three clusters may have originated in the same star - forming complex , but not in the same giant molecular cloud .
= 1.0 cm = 0.8 cm |
in the last few years it has become apparent that quantum information may lead to further insight into other areas of physics such as condensed matter and statistical mechanics @xcite .
the attention of the quantum information community to study in condensed matter has stimulated an exciting cross fertilization between the two areas .
it has been found that entanglement plays a crucial role in the low - temperature physics of many of these systems , particularly in their ground state@xcite .
quantum phase transition ( qpt ) happens at zero temperature and shown non analyticity in the physical properties of the ground state by the change of a parameter @xmath2 of the hamiltonian @xmath3 .
this change is driven only by quantum fluctuations@xcite .
since qpt occurs at @xmath4 , the emerging correlations have a purely quantum origin .
therefore , it is reasonable to conjecture that entanglement is a crucial ingredient for the occurrence of the qpt @xcite .
@xcite have shown that a discontinuity in a bipartite entanglement measure ( concurrence@xcite and negativity @xcite ) is a necessary and sufficient indicator of a first - order quantum phase transition , negativity being characterized by a discontinuity in the first derivative of the ground state energy .
they have also shown that a discontinuity or a divergence in the first derivative of the same measure ( assuming it is continuous ) is a necessary and sufficient indicator of a second - order qpt , that is characterized by a discontinuity or a divergence of the second derivative of the ground state energy .
dzyaloshinsky has shwon@xcite that , in crystal with no inversion center , the usual isotropic exchange @xmath5 is not the only magnetic interaction and antisymmetric exchange @xmath6 is allowed .
later , moriya has shown@xcitethat inclusion of spin orbit coupling on magnetic ions in 1st and 2nd order leads to antisymmetric and anisotropic exchange respectively .
this interaction is , however , rather difficult to handle analytically , but it is one of the agents responsible for magnetic frustration . since this interaction may induce spiral spin arrangements in the ground state@xcite , it is closely involved with ferroelectricity in multiferroic spin chains@xcite . besides , the dm interaction plays an important role in explaining the electron spin resonance experiments in some one - dimensional antiferromagnets@xcite .
moreover , the dm interaction modifies the dynamic properties@xcite and quantum entanglement@xcite of spin chains@xcite . in the present paper , we are interested to study the one - dimensional spin-1/2 ising model with added dm interaction from quantum entanglement point of view using variational matrix product state and numerical exact diagonalization methods .
the hamiltonian is given by @xmath7 where @xmath8 is spin-@xmath9 operator on the @xmath10-th site , and @xmath11 @xmath12 denotes antiferromagnetic ( ferromagnetic ) coupling constant . in very recent
works@xcite respectively studied the ground state phase diagram of the ferromagnetic and antiferromagnetic ising chain in the presence of the uniform dm interaction .
it is found that the ground state phase diagram of both systems consists of spiral - ferromagnetic and spiral - antiferromagnetic phases respectively and a commensurate - incommensurate ( c - ic ) quantum phase transition occurs at @xmath13 .
however at the critical value @xmath14 , a metamagnetic phase transition occurs into the chiral gapless phase in the ground state phase diagram of the ferromagnetic chain .
this paper is structured as follows . in section
ii , we will discuss about bipartite and multipartite entanglement measures as qpt indicators of our model and we will present our numerical study . in section iii , the variational matrix product states will be outlined and bipartite entanglement will be obtained . in section iv , we willstudy entanglement of three qubits in symmetric and antisymmetric way . finally we conclude and summarize our results in section v.
the occurrence of collective behavior in many - body quantum systems is associated with classical and quantum correlation .
the quantum correlation , which known as entanglement , can not be measured in terms of classical physics and represents the impossibility of giving a local description of many - body quantum state .
the issue of finding entanglement measures has recently attracted an increasing interest@xcite .
concurrence and tangle are the most widely used measures in qpt related entanglement studies .
both of these measures are for bipartite states and because of monogamous nature of entanglement they are expected to decrease at the quantum critical point , if entanglement is shared by the whole system .
therefore , in order to manifest the presence of ( qpt ) in the model described by eq.([ising ] ) , we focus on the entanglement of formation @xcite in the quantum spin system and make use of the _ one - tangle _ and of the concurrence .
the one - tangle @xcite quantifies the zero temperature entanglement of a single spin with the rest of the system and defines as @xmath15 where @xmath16 is the one - site reduced density matrix , @xmath17 , and @xmath18 . in terms of the spin expectation values @xmath19 , one has : @xmath20 on other hand , the concurrence@xcite quantifies instead the pairwise entanglement between two spins and defines as @xmath21 where @xmath22 and @xmath23 is the correlation function between spins on sites @xmath24 and @xmath25 and @xmath26 . the notation @xmath27 represents the ground state expectation value . , ( a ) ferromagnetic and ( b ) antiferromagnetic cases for different lengths n=12 , 16 , 20 , 24.,title="fig:",width=377,height=302 ] + , ( a ) ferromagnetic and ( b ) antiferromagnetic cases for different lengths n=12 , 16 , 20 , 24.,title="fig:",width=377,height=302 ] + one - tangle and concurrence are related by coffman - kundu - wootters ( ckw ) conjecture @xcite , which had been proved by osborne and verstraete @xcite , stating that @xmath28 which expresses the crucial fact that pairwise entanglement does not exhaust the global entanglement of the system , as entanglement can also be stored in 3-spin correlations , 4-spin correlations , and so on .
authors in ref.(@xcite ) , have proposed that , due to ckw conjecture , the minimum of the entanglement ratio @xmath1 , as a novel estimator of qpt , fully based on entanglement quantifiers . by doing an experiment , one can find a clear picture of the entanglement phenomenon in the ground state magnetic phases of the model . since a real experiment
can not be done at zero temperature , the best way is doing a virtual numerical experiment . a very famous and accurate method in the field of the numerical experiments
is known as the lanczos method .
however , the strong role of a numerical experiment to examine quantum phase transitions is not negligible . to explore the nature of the entanglement in different magnetic phases
, we used lanczos method to diagonalize numerically chains with length up to @xmath29 and coupling constant @xmath30 .
the ground state eigenvector , @xmath31 , was obtained for chains with periodic boundary conditions .
the numerical lanczos results on the concurrence for the ising chain with dm interaction , are shown in fig .
[ c - f ] . as is clearly seen from fig .
[ c - f](a ) , in the case of ferromagnetic chain and for @xmath32 , the concurrence is equal to zero which shows that the ground state of the system is in the fully non - entangled polarized ferromagnetic phase . at the critical value @xmath33 , the concurrence jumps to a non - zero value which confirms the metamagnetic phase transition . by more increasing the dm vector , @xmath34 ,
the ground state is in the chiral phase and nearest neighbors are entangled . on the other hand , in the case of the antiferromagnetic ising model , as can be clearly seen from fig .
[ c - f](b ) , in the absence of the dm interaction , the ground state is non - entangled which is related to the saturated n@xmath35el phase .
as soon as the dm vector applies , nearest neighbors will be entangled and concurrence between them increases from zero .
thus in the case of antiferromagnetic ising chains , the dm interaction induces the quantum correlations of the two spins and nearest neighbor spins will be entangled as soon as the dm interaction applies . in contrast , in the case of the ferromagnetic ising chains the dm interaction only induces the quantum correlations after the critical value @xmath14 and nearest neighbor spins will not be entangled up to the critical value @xmath14 .
an additional insight into the nature of different phases can be obtained by studying the entanglement ratio .
therefore , we have calculated the entanglement ratio by using lanczos method in both ferromagnetic and antiferromagnetic cases .
we have plotted our numerical results in fig .
[ c1-f ] . for ferromagnetic case , as it can be seen from fig .
[ c1-f](a ) , the entanglement ratio remains zero up to the critical dm interaction @xmath14 which is expected from the saturated ferromagnetic phase .
as soon as the dm interaction increases from the critical @xmath14 , the entanglement ratio starts to increase from zero . in the inset of fig .
[ c1-f](a ) , the first derivative entanglement ratio is plotted . as it is seen in the ferromagnetic phase , @xmath36 ,
the derivative is equal to zero and an abrupt change took place exactly at @xmath37 which is an indication of the quantum phase transition . in the antiferromagnetic case ,
[ c1-f](b ) , as soon as the dm interaction applies the entanglement ratio creates and decreases by increasing the dm vector up to @xmath38 which a change took place and in the @xmath39 region the ratio becames monotonous . in the inset of fig .
[ c1-f](b ) , the first derivative entanglement ratio is plotted .
it is clear that in the @xmath39 region derivative of ratio is equal to zero and an abrupt change took place exactly at @xmath40 which is an indication of the quantum phase transition .
is plotted as a function of dm vector @xmath41 , ( a ) ferromagnetic and ( b ) antiferromagnetic cases for different lengths n=20 , 24 .
inset : the first derivative of entanglement ratio . , title="fig:",width=377,height=302 ]
+ is plotted as a function of dm vector @xmath41 , ( a ) ferromagnetic and ( b ) antiferromagnetic cases for different lengths n=20 , 24 .
inset : the first derivative of entanglement ratio . ,
title="fig:",width=377,height=302 ] + because of many different kinds of entanglement , quantifying of multipartite entanglement states ( mes ) is more difficult .
global - entanglement ( @xmath42 ) measure defined by meyer and wallach @xcite which can measure the total nonlocal information per particle in a general multipartite system@xcite @xmath43.\ ] ] @xmath42 is the average of tangles per particles ( @xmath44 ) , without giving detailed knowledge of tangle distribution among the individual particles .
therefore , @xmath42 is an average quantity and can not distinguish between entangled states which have equal @xmath45 yet different distributions of tangles , like @xmath46 ( greenberger - horne - zeilinger ) and @xmath47 state .
@xmath42 has ability to discriminate between @xmath48 from @xmath49 states because of their different values of tangle
. de oliveira et al .
@xcite also introduced a slight extension of global entanglement as generalized global entanglement ( gge ) which , in contrast to global entanglement , the proposed gge measure can distinguish three paradigmatic entangled @xmath46 ( greenberger - horne - zeilinger ) , @xmath47 and w states .
@xmath50\nonumber\\ & = & \frac{4}{3}\left[1-\frac{1}{4}\sum_{\alpha,\beta=0}^{3}\big<\sigma_{j}^{\alpha}\sigma_{j+1}^{\beta}\big>^{2}\right ] . \label{gge}\end{aligned}\ ] ] as such the generalized measure can detect a genuine multipartite entanglement and is maximal at the critical point@xcite . here
, we have calculated multipartite entanglement eq.([gge ] ) by using lanczos numerical method with periodic boundary conditions in antiferromagnetic case .
we have plotted our numerical results in fig.[multipartite ] for antiferromagnetic case . as it can be seen from fig.[multipartite](a ) the multipartite entanglement starts to increase by increasing dm up to the critical dm interaction @xmath14 which a change took place exactly at @xmath37 and then after that the multipartite entanglement reaches the saturation value .
our calculation shows that multipartite entanglement is maximal around the critical point @xmath51 . in order to get better insight into the ability of multipartite entanglement as a quantum phase transition toolkit
we have plotted the first derivative entanglement multipartite entanglement in the inset of fig .
[ multipartite](a ) .
it shows divergent behavior at the critical point @xmath51 .
we have also plotted @xmath52 in fig .
[ multipartite](b ) .
our calculation shows @xmath53 increases as @xmath54 at the critical point . here
, we have finally calculated global entanglement @xmath42 of model eq.([ising ] ) and compar it with multipartite entanglement .
it can be seen from fig .
[ multipartite3 ] that both @xmath53 and @xmath42 are maximal at the critical point @xmath14 and their behavior is qualitatively the same .
is plotted as a function of dm vector @xmath41 for antiferromagnetic ising chain with length n=20 , 24 .
inset : the first derivative of multipartite entanglement@xmath55 .
( b ) @xmath53 for the chain size n=24 , is plotted in the limite of @xmath54.,title="fig:",width=377,height=302 ] + is plotted as a function of dm vector @xmath41 for antiferromagnetic ising chain with length n=20 , 24 .
inset : the first derivative of multipartite entanglement@xmath55 .
( b ) @xmath53 for the chain size n=24 , is plotted in the limite of @xmath54.,title="fig:",width=377,height=302 ] + and global entanglement @xmath42 are plotted as a function of dm vector @xmath41 , for antiferromagnetic ising chain with length n=24.,title="fig:",width=377,height=302 ] +
the matrix product state is defined as@xcite @xmath56 where @xmath57 , and @xmath58 and @xmath59 are probability amplitudes for two spin configurations at site @xmath10 . in
what follow , we intend to determine the ground state energy of ferromagnetic ising spin system with dm interaction . in this respect , by using the above formalism , the variational energy is obtained by @xmath60 } , \label{var}\end{aligned}\ ] ] where @xmath61 and @xmath62 here @xmath63 and @xmath64 .
the minimum of the variational energy function corresponds to the ground state energy of the system . using the normalization condition
@xmath65 , we can mapped variational parameter to @xmath66 , @xmath67 .
therefor one can obtain @xmath68 , @xmath69 where @xmath70 and by choosing @xmath71 , it is found that : @xmath72 by minimizing above equation , the ground state energy @xmath73 in the ferromagnetic case @xmath74 shall be obtained .
it was shown that the ground state energy has the constant value , @xmath75 for @xmath76 and decreasing linearly with dm interaction for @xmath77 as @xmath78 .
now , buy using the minimized variational parameters of @xmath79 we are able to calculate tangle eq.([tangle2 ] ) and concurrence eq.([concurrence ] ) . for @xmath80
one can obtain @xmath81 and @xmath82 so we have @xmath83 and @xmath84 .
for @xmath85 again using the above conditions , one can obtain @xmath86 and @xmath87 which give @xmath88 and @xmath89 .
in this section we focus on the entanglement of formation of three - qubit in two inequivalent , symmetric and non - symmetric pairwise entanglement ways . by labeling the 3-qubits as @xmath90 sequentially .
the symmetric reduced density matrix @xmath91 is defined as @xmath92 , where @xmath93 is the density matrix of 3-qubits .
the non - symmetric reduced matrix is @xmath94 . before present our results
, we briefly review the definition of concurrence@xcite .
let @xmath95 be density matrix of a pair of qubit @xmath96 and @xmath10 .
the concurrence corresponding to density matrix is defined as @xmath97 where the quantities @xmath98 are square roots of the eigenvalue of the operator @xmath99 the concurrence @xmath100 corresponds to an non - entangled state and @xmath101 corresponds to a maximally entanglement sate .
a straightforward calculation gives the eigenstates and the eigenvalues of 3-qubit of eq.([ising ] ) .
the square roots of the operator @xmath102 for symmetric and nonsymmetric cases are presented in sequence . in the symmetric case
@xmath103 @xmath104 and for the non - symmetric case @xmath105 @xmath106 where @xmath107 . in order to determine the existence of entanglement
, we have considered ferromagnetic and antiferromagnetic cases .
fig.[c1 - 3t0 ] and fig.[c1 - 2t0 ] show symmetric , @xmath108 , and non - symmetric , @xmath109 , concurrences . in the symmetric way ,
[ c1 - 3t0 ] , it is clear that in the absence of dm interaction the antiferromagnetic case is entanglement , in contrast to ferromagnetic case which is fully unentangled . in antiferromagnetic case ,
the @xmath108 is zero up to @xmath110 and starts to increase by increasing dm and reaches a saturation value .
our calculation shows a competition between ising exchange @xmath111 and the dm interaction which entanglement starts to decreasing by increasing ising exchange . in the symmetric way , ferromagnetic ising chain with added dm interaction
does not show any entanglement and by increasing dm interaction nothing will not happen .
, for antiferromagnetic case as function of dm and j exchange.,title="fig:",width=377,height=302 ] + , for both ferromagnetic an antiferromagnetic cases as function of dm and j exchange . , title="fig:",width=377,height=302 ] + in the non - symmetric way , fig .
[ c1 - 2t0 ] , ferromagnetic ising chain with added dm interaction does not show any entanglement up to @xmath110 .
but for @xmath112 entanglement starts to increase by increasing dm interaction and shows a competitive behavior between dm and j exchange which entanglement shows decreasing by increasing @xmath113 . in antiferromagnetic chain , through non - symmetric case , opposite to ferromagnetic case , is fully entangled .
entanglement starts to increasing from zero as soon as turn on dm and reaches its saturation value around @xmath110 .
it should be mentioned that in antiferromagnetic case , increasing @xmath111 exchange can enhance the amount of entanglement .
the concept of thermal entanglement was introduced and studied within one - dimensional isotropic heisenberg model@xcite .
here we study this kind of entanglement within three - qubit ising chain with added dm interaction through symmetric and non - symmetric way .
the state of the system at thermal equilibrium is @xmath114 , where @xmath115 $ ] is the partition function and @xmath116 is the boltezmann s constant . as @xmath117 represents thermal state , the entanglement in the state is called thermal entanglement@xcite .
the square roots of the operator @xmath118 for symmetric and nonsymmetric cases are presented in sequence . for the non - symmetric case @xmath119 @xmath120 where @xmath121,\nonumber\\ y_{12}&=&\frac{1}{\omega}\left[\frac{1}{2}+\frac{(a^{2}+1)e^{-\beta \varepsilon_{3}}}{a^{2}+2}+\frac{(b^{2}+1)e^{-\beta \varepsilon_{7}}}{b^{2}+2 } \right],\nonumber\\ z_{12}&=&\frac{1}{\omega}\left[\frac{2ae^{-\beta \varepsilon_{3}}}{a^{2}+2}+\frac{2be^{-\beta \varepsilon_{7}}}{b^{2}+2 } \right],\end{aligned}\ ] ] and for the non - symmetric case @xmath122 @xmath123 where @xmath124,\nonumber\\ y_{13}&=&\frac{1}{\omega}\left[1+\frac{(2e^{-\beta \varepsilon_{3}}}{a^{2}+2}+\frac{2e^{-\beta \varepsilon_{7}}}{b^{2}+2 } \right],\nonumber\\ z_{13}&=&\frac{1}{\omega}\left[1-\frac{(2e^{-\beta \varepsilon_{3}}}{a^{2}+2}-\frac{2e^{-\beta \varepsilon_{7}}}{b^{2}+2 } \right],\end{aligned}\ ] ] and @xmath125 $ ] , @xmath126 , @xmath127 . in fig.[thermal - c - af ] and fig.[thermal - c-2-af ] we give two plots of the thermal concurrence of antiferromagnetic and ferromagnetic ising chains as functions of temperature and dm interaction respectively .
in antiferromagnetic case , fig.[thermal - c - af ] , model dose not shows any entanglement through symmetric way and model lives in fully unentangled phase .
in contrast to symmetric way , non - symmetric way shows entanglement . as it can be seen from fig.[thermal - c - af ] , there is a region which surprisingly temperature can enhance entanglement and this region will become wide spread for higher dm . for higher temperature dm can not overcome
temperature and entanglement gradually goes to zero by increasing temperature . , for antiferromagnetic case as function of dm and temperature.,title="fig:",width=415,height=359 ] + , and symmetric way(bottom ) , @xmath108,for ferromagnetic case as function of dm and temperature .
, title="fig:",width=415,height=340 ] + , and symmetric way(bottom ) , @xmath108,for ferromagnetic case as function of dm and temperature .
, title="fig:",width=415,height=340 ] + in ferromagnetic case , fig.[thermal - c-2-af ] , both symmetric and non - symmetric way have qualitatively the same behavior . here , same as antiferromagnetic case
, temperature can enhance entanglement .
thermal and quantum fluctuations have a tight competition to drive system in favorable regime , but in low temperature , both temperature and dm interaction has effective influence on the degree of entanglement of the system and both of them could be used to increase the entanglement of the spin system .
+ either fig.[c1 - 3t0 ] and fig.[c1 - 2t0 ] are plotted for @xmath128 , because we could not find any prominent behavior for @xmath129 .
actually both symmetric and non - symmetric entanglement are zero in the @xmath129 region .
to summarize , we have investigated the effect of a dzyaloshinskii - moriya ( dm ) interaction on the ground state phase diagram of the one - dimensional ( 1d ) ising spin-1/2 model using the variational matrix product state and numerical lanczos methods from entanglement point of view .
our results show that there is a critical point in either ferromagnetic and antiferromagnetic cases . in the ferromagnetic ising chain
this critical point was predicted by variational matrix approach exactly at @xmath110 , which by using numerical study we have confirmed it .
we have also used the minimum of the entanglement ratio @xmath1,to check the presence of this quantum critical point . for both ferromagnetic and antiferromagnetic cases
our numerical study gave the minimum of @xmath130 at @xmath13 , .
we also used generalized multipartite entanglement tools to check how entanglement is share in our model and check their ability to detect critical point . for antiferromagnetic case
we have calculated global @xmath42 and generalized multipartite entanglement @xmath53 and either of them show that a quantum phase transition took place at @xmath110 . we have calculated ground state entanglement of symmetric , @xmath108 , and non - symmetric , @xmath109 , concurrences for ferromagnetic and antiferromagnetic cases . in the symmetric way , in the absence of dm interaction
both antiferromagnetic and ferromagnetic cases are fully unentangled . by increasing dm , in antiferromagnetic case ,
the @xmath108 is zero up to critical points @xmath110 .
after the critical point entanglement starts to increase until its saturation .
for ferromagnetic case , in symmetric way @xmath108 is always zero and does not show any entanglement . in the non - symmetric way , in the absence of dm interaction ,
@xmath109 is equal zero for either antiferromagnetic and ferromagnetic cases . by turning dm nothing , in the ferromagnetic case
, will not happen up to critical point @xmath51 and then after that @xmath109 starts to increase by increasing dm interaction .
in contrast to ferromagnetic case , in the antiferromagnetic case @xmath109 starts to increasing immediately after turning dm and reaches its saturation value around @xmath110 .
we have also studied thermal entanglement of symmetric , @xmath131 , and non - symmetric , @xmath132 for both ferromagnetic and antiferromagnetic cases .
our calculations show that for either symmetric and non - symmetric cases thermal(unentanglement favorable ) and quantum(entanglement favorable ) fluctuations have competition to drive system to their favorable regime and surprisingly thermal decoherence can enhance entanglement in some part of low temperature region .
10 de oliveira t r , rigolin g , de oliveira m c 2006 phys .
a * 73 * , 010305(r ) ; rigolin g , de oliveira t r , de oliveira m c 2006 phys .
a * 74 * , 022314 ; de oliveira
t r , rigolin g , de oliveira m c and miranda e 2006 phys .
lett . * 97 * , 170401 klumper a , schadschneider a and zittartz 1991 j. phys . a 24 , l955 ; klumper a , schadschneider a and zittartz 1992 z. phys .
b * 87 * , 281 ; klumper a , schadschneider a and zittartz 1993 europhys . lett . * 24 * , 293 | we have studied occurrence of quantum phase transition in the one - dimensional spin-1/2 ising model with added dzyaloshinsky - moriya ( dm ) interaction from bi - partite and multi - partite entanglement point of view . using exact numerical solutions , we are able to study such systems up to @xmath0 qubits .
the minimum of the entanglement ratio @xmath1 , as a novel estimator of qpt , has been used to detect qpt and our calculations have shown that its minimum took place at the critical point .
we have also shown both the global - entanglement ( ge ) and multipartite entanglement ( me ) are maximal at the critical point for the ising chain with added dm interaction . using matrix product state approach
, we have calculated the tangle and concurrence of the model and it is able to capture and confirm our numerical experiment result .
lack of inversion symmetry in the presence of dm interaction stimulated us to study entanglement of three qubits in symmetric and antisymmetric way which brings some surprising results . |
quantum information and condensed matter physics are currently two areas of very active research where several links between the two branches have been discovered in recent years . in particular
, the relation of entanglement properties to critical points of quantum phase transitions has been studied in some detail . in analogy to classical phase transitions , the latter are typically characterized in terms of the scaling behavior of equilibrium properties , such as the correlation length .
analogous scaling phenomena have now been found for the entanglement properties of a spin chain in the vicinity of a critical point . @xcite . in quantum communication ,
condensed matter systems have recently received some attention as interesting candidates for quantum channels @xcite .
this approach may be of advantage in situations where photonic quantum communication is not possible for example because the two units are only separated by a few optical wavelengths or are inside a material that does not permit for the well - controlled propagation of light .
the quantum mechanical properties of correlated many body system are to some extent determined by the energetic gap between their ground states and the lowest excited states .
in particular the system is short range correlated if it features a finite gap , while some quantities can become long range correlated when the gap closes @xcite .
the latter happens at the critical points of quantum phase transitions @xcite .
motivated by these findings the dynamical entanglement properties of quantum many body systems undergoing a quantum phase transition are receiving increasing attention @xcite . to pursue the analysis of quantum many body systems as possible quantum channels a step further , we study here the connection between the spectral gap and the systems ability to transfer excitations and entanglement .
however , our aim is twofold : besides being interested in the systems transfer capacity , we also want to explore , to what degree the transfer properties can be employed to detect and characterize critical points of quantum phase transitions . in particular , we consider spin chains and harmonic chains , where in both cases , two ancillas couple weakly to the chain at distant sites .
we study the transfer of quantum information and excitations from one ancilla to the other numerically , by employing newly developed matrix product state techniques @xcite , and analytically with a master equation approach .
the numerics is applied to the spin chains while the master equation is used to describe the case of an harmonic chain , where its validity is confirmed numerically for gaussian initial states @xcite .
both systems show a very similar behavior : the transfer properties sensitively depend on the energy gap between the ground state and the lowest excited states , but do not significantly dependent on the detailed structure of the hamiltonian .
the quality and speed of state transfer through such systems may thus be used to detect critical points experimentally .
this paper is organized as follows : section ii begins with a discussion of spin chains , mainly employing numerical methods .
in section iii a simplified picture unraveling the origin of the observed behavior is presented .
this picture is then further corroborated in section iv where we study harmonic chains both numerically and analytically with a master equation approach .
section v discusses connection to quantum channel capacities and section vi summarizes our results .
a variety of quantum spin chains exhibit quantum phase transitions .
therefore we begin with the study of a linear chain of spins with nearest neighbor interactions and open boundary conditions described by the hamiltonian @xmath0 here @xmath1 is the number of spins , @xmath2 a magnetic field and @xmath3 , @xmath4 and @xmath5 the interaction between neighboring spins .
two ancilla systems ( named @xmath6 for `` sender '' and @xmath7 for `` receiver '' ) couple to the chain at spins @xmath8 and @xmath9 , which are near the center of the chain in order to avoid boundary effects .
the total hamiltonian of chain and ancillas then reads @xmath10 where @xmath11 is the zeeman splitting of the ancillas , which might differ from @xmath2 , and @xmath12 the coupling of the ancillas to the chain , which is taken to be weak , i.e. @xmath13 . as an initial state of the system we consider @xmath14 i.e. the sender is spin up and the receiver is spin down , while the chain is in its ground state , @xmath15 .
this will allow us to explore the propagation of this excitation through the chain .
the dynamics is simulated with a matrix product state technique using matrices of dimension @xmath16 @xcite .
the accuracy of the simulations was confirmed by varying the matrix dimension and the size of the time steps .
furthermore , we confirmed that the dynamics conserves the total energy of the system .
figure [ simulnonres ] shows the probability @xmath17 that `` sender '' @xmath6 is in its excited state @xmath18 and the `` receiver '' @xmath7 in its ground state @xmath19 , together with @xmath20 and @xmath21 for a model with @xmath22 , @xmath23 , @xmath24 , @xmath25 , @xmath26 , @xmath27 , @xmath28 and @xmath29 .
@xmath30 is always less than @xmath31 .
@xmath32 ( light gray ) , @xmath33 ( gray ) and @xmath34 ( black ) for @xmath35 , @xmath26 , @xmath27 , @xmath28 and @xmath29 as given by the simulation for the open boundary model with @xmath22 spins .
@xmath6 couples to spin @xmath36 and @xmath7 to spin @xmath37 .
figure taken from @xcite.,width=264 ] the plots show that the excitation that was initially located in @xmath6 oscillates back and forth between @xmath6 and @xmath7 .
figure [ simulres ] shows @xmath38 , @xmath20 and @xmath21 for a model with @xmath39 , @xmath40 , @xmath41 , @xmath35 , @xmath26 , @xmath27 , @xmath42 and @xmath29 .
again , @xmath30 is always less than @xmath31 .
@xmath32 ( dotted line ) , @xmath33 ( dashed line ) and @xmath34 ( solid line ) for @xmath35 , @xmath26 , @xmath27 , @xmath42 and @xmath29 as given by the simulation for the open boundary model with @xmath39 spins . @xmath6 couples to spin @xmath43 and @xmath7 to spin @xmath44 . figure taken from @xcite.,width=264 ] for these parameters , the excitation is not fully transferred to @xmath7 , contrary to figure [ simulnonres ] . both , @xmath6 and @xmath7 relax to their ground states with the excitation only being partially and temporarily transferred to @xmath7 , even for close - lying spins .
note that the parameters chosen in figures [ simulnonres ] and [ simulres ] are the same except for @xmath45 which in figure [ simulres ] is significantly larger than in figure [ simulnonres ] .
the two observed scenarios are rather generic .
to demonstrate this , we have done the same simulations for a xxz - model .
the results , shown in figures [ simulnonres1 ] and [ simulres1 ] , clearly agree with our findings for the previous coupling parameters . again
@xmath45 in figure [ simulres1 ] is significantly larger than in figure [ simulnonres1 ] , while all other parameters are equal .
@xmath32 ( light gray ) , @xmath33 ( gray ) and @xmath34 ( black ) for @xmath35 , @xmath46 , @xmath47 , @xmath48 , @xmath49 and @xmath29 as given by the simulation for the open boundary model with @xmath22 spins .
@xmath6 couples to spin @xmath36 and @xmath7 to spin @xmath37 .
figure taken from @xcite.,width=264 ] @xmath32 ( dotted line ) , @xmath33 ( dashed line ) and @xmath34 ( solid line ) for @xmath35 , @xmath26 , @xmath47 , @xmath48 , @xmath50 and @xmath29 as given by the simulation for the open boundary model with @xmath39 spins .
@xmath6 couples to spin @xmath43 and @xmath7 to spin @xmath44 .
figure taken from @xcite.,width=264 ]
the exact dynamics of a quantum spin chain is extraordinarily complex .
nevertheless , underlying the dramatic difference between the almost perfect transfer scenarios in figures [ simulnonres ] and [ simulnonres1 ] and the damped scenario in figures [ simulres ] and [ simulres1 ] is a simple physical mechanism .
the dynamics we have simulated is given by the schrdinger equation containing the hamiltonian ( [ hamtot ] ) . as a consequence ,
all moments of the hamiltonian are conserved .
the initial state @xmath51 , where one of the ancillas is excited , is not an eigenstate of @xmath52 as given by ( [ hamtot ] ) .
the mean value is @xmath53 and the variance of the energy is given by @xmath54 given that the evolution is assumed to be hamiltonian , both are in fact time invariant .
thus in the entire dynamics only those states with an energy expectation value @xmath55 in the range @xmath56 play a significant role as indicated by figure [ energycon ] .
@xmath6 and @xmath7 are depicted as two level systems , while for the chain there is a unique ground state and a quasi continuous band of excited states sketched as the gray area .
the dots indicate the initial occupations .
the relevant energy range given by the variance lies between the two horizontal dashed lines .
if the spectral gap is larger than the zeeman splitting of the ancillas ( left plot ) it is not possible to create a real excitation of the chain , only processes involving virtual excitations of the chain are then permitted and the excitation will be almost completely transferred from @xmath6 to @xmath7 .
if however the spectral gap is smaller than the zeeman splitting of the ancillas , real excitations may be created in the chain .
these excitations dephase rapidly due to the large number of accessible energy levels which prevents their reabsorption the chain acts as a bath and information is lost .
sketch of the energy levels of the system .
the dots indicate the occupations of the initial state . for this initial state
only the energy levels between the two horizontal dashed lines are accessible , resulting in almost perfect transfer for the left scenario and damping for the right one.,title="fig:",width=113 ] sketch of the energy levels of the system .
the dots indicate the occupations of the initial state . for this initial state
only the energy levels between the two horizontal dashed lines are accessible , resulting in almost perfect transfer for the left scenario and damping for the right one.,title="fig:",width=113 ] we believe that this heuristic picture captures the essential physics . to further corroborate our simple physical model and to underline the generality of our findings
, we now turn to a different model for the chain which also features an adjustable energy gap above its unique ground state .
harmonic chains whose time evolution is governed by a hamiltonian that is quadratic in position and momentum can be solved exactly in a compact form and therefore lend themselves to analytical investigations .
furthermore , gaussian initial states , ie states whose wigner function is a gaussian , preserve their gaussian character under the time evolution . as a gaussian
is fully described by its first and second moments this allows for a very compact , polynomial in the number of oscillators , description of the state of the systems and its evolution in time ( see e.g. @xcite for details ) , which permits the numerical investigation of very large systems .
this is the basic motivation for the study of harmonic systems in gaussian states in this section .
we consider a harmonic chain with periodic boundary conditions described by @xmath57 with the @xmath58 being the momenta and the @xmath59 the positions ( @xmath60 ) . in this case , the two ancillas are harmonic oscillators that couple to oscillators @xmath8 and @xmath9 of the chain .
the complete hamiltonian reads @xmath61 we are interested in the time evolution of the ancillas only .
thus , we will now derive an approximate master equation for the reduced density matrix @xmath62 of the two ancilla systems .
for weak coupling @xmath63 we find @xmath64 { \right } ] { \right}\ } \
, , \ ] ] where @xmath62 is the reduced density matrix of the ancillas , @xmath65 is the trace over the degrees of freedom of the chain and @xmath66 denotes the ground state of the chain ( [ harmchain ] ) .
@xmath67 is the interaction between ancillas and chain in the interaction picture . here
@xmath68 with @xmath69 and @xmath70 and @xmath71 eq . ( [ nakajima ] ) is an expansion in the coupling strength @xmath72 up to second order , which is a good approximation if the integral approaches a constant value for @xmath73 , where @xmath74 . since @xmath75 only changes significantly on time scales @xmath76 , the approximation @xmath77 can be used .
performing the trace on the rhs of ( [ nakajima ] ) yields @xmath78 { \right}. { \nonumber}\\ & & \enspace + { \left}. { \left}(x_1 + ( x_0 - x_1)\delta_{jl}{\right } ) { \left } ( \ { a_j a_l^{\dagger } , \sigma \ } - 2 ( a_j \sigma a_l^{\dagger } ) { \right } ) { \right } ) \ , , \end{aligned}\ ] ] where @xmath79 $ ] and @xmath80 denote commutators and anti - commutators and @xmath81 and @xmath82 are the annihilation operators of @xmath6 and @xmath7 , respectively @xmath83 for @xmath84 . in deriving ( [ masterform ] ) , we applied a rotating wave approximation .
the coefficients read @xmath85 with @xmath86 given by @xmath87 where @xmath88 . due to the symmetries of the model , we have @xmath89 and @xmath90 .
( [ masterform ] ) is a good approximation whenever @xmath91 .
since the @xmath92 do not depend on @xmath72 themselves , there is always a sufficiently small @xmath72 such that this holds , provided @xmath93 exists .
the validity of the master equation can be confirmed by a numerical simulation for a chain with 1400 oscillators .
we found good agreement between our analytical and numerical solutions , with the relative errors being less than @xmath94 .
the harmonic chain can be diagonalized via a fourier transform @xcite . in the limit of an infinitely long chain , @xmath95 ,
the correlation functions read @xmath96 where @xmath97 with @xmath98 . and all @xmath93 exist except for the case where @xmath99 . as in master equations for system bath models ,
we now insert the asymptotic expressions , @xmath100 , into eq .
( [ masterform ] ) .
this replacement assumes that all internal dynamics of the chain happens on much shorter time scales than the dynamics caused by the interaction of the ancillas with the chain .
furthermore , it does not treat the initial evolution for short times with full accuracy since @xmath101 .
the master equation thus obtained is therefore valid in a regime where the couplings @xmath72 are weak enough such that the time it takes for an excitation to travel from @xmath6 to @xmath7 is completely determined by @xmath72 , i.e. by the time it takes to be transferred into and from the chain . from eq .
( [ masterform ] ) we find the following solution for the expectation values of the occupation numbers of @xmath6 and @xmath7 , @xmath102 and @xmath103 : @xmath104 here , @xmath105 ,
@xmath106 and @xmath107 .
note that @xmath108 and @xmath109 . inserting the correlation functions into eq .
( [ mastercoeff ] ) one sees that @xmath110 and @xmath111 are only non - zero if @xmath112 for at least one mode @xmath113 , that is if our initial state is in resonance with states where both ancillas are in their ground states and the chain is in one of its lowest - lying excited states ( c.f .
figure [ energycon ] ) .
the dispersion relation shows that this only happens for @xmath114 . as in figures
[ simulnonres ] and [ simulres ] or [ simulnonres1 ] and [ simulres1 ] , we thus observe two different scenarios : if @xmath115 , and therefore @xmath116 , the excitation that is initially in @xmath6 oscillates back and forth between @xmath6 and @xmath7 at a frequency @xmath117 , i.e. @xmath118 note in particular that the excitation is entirely transferred to @xmath7 at times @xmath119 .
figure [ offresplot ] shows the frequencies @xmath117 of the excitation s oscillations between @xmath6 and @xmath7 for cases where @xmath115 , for @xmath120 , @xmath29 and @xmath121 as a function of @xmath122 .
as @xmath123 decreases , the transfer becomes faster and the oscillation frequency increases .
frequencies @xmath117 of the excitation s oscillation between @xmath6 and @xmath7 for @xmath120 , @xmath29 and @xmath124 .
the transfer speed increases as @xmath125 .
figure taken from @xcite.,width=264 ] let us examine the dependence of the oscillation frequency @xmath126 on the distance @xmath127 . with the definitions of @xmath128 and @xmath129 and eqs .
( [ mastercoeff ] ) , ( [ correlfun ] ) and @xmath130 where @xmath131 denotes the principal value , we find @xmath132 using the substitution @xmath133 , this integral can be converted to an integral over the unit circle , which in turn can be evaluated via the residue theorem @xcite .
the result is @xmath134 where @xmath135 and @xmath136 .
the frequency @xmath117 and hence the transfer speed decrease exponentially with increasing @xmath127 .
if , on the other hand , @xmath114 , the chain acts similarly to a bath . here , @xmath137 , @xmath138 , and both ancillas
relax into their ground state transferring their energy into the chain . during this process , however , a fraction of the energy initially located in @xmath6 appears momentarily in @xmath7 before it is finally damped into the chain .
the maximal excitation of the receiver throughout the entire evolution depends on the distance @xmath127 . for a given initial energy in @xmath6 , only a narrow range of the excitation spectrum
is relevant for the dynamics .
hence all relevant modes of the chain have very similar wavelength .
if the distance @xmath127 is a multiple of half that wavelength , a wave that has high amplitude next to @xmath6 will have high amplitude next to @xmath7 , too .
in that case , the maximal excitation probability in @xmath7 will be relatively large , although less than @xmath139 .
if on the other hand , a wave that has high amplitude next to @xmath6 has zero amplitude next to @xmath7 , the transfer vanishes completely .
figure [ mastersolplot2 ] shows the solution ( [ mastersol ] ) for a harmonic chain and ancillas with @xmath120 , @xmath124 , @xmath29 , @xmath140 and @xmath141 .
the solution ( [ mastersol ] ) for a harmonic chain and ancillas with @xmath120 , @xmath124 , @xmath29 , @xmath140 and @xmath141 . figure taken from @xcite.,width=264 ] one might try to derive the same type of master equation for the spin chain ( [ hamchain ] ) . in the next section , we explain why such an approach is problematic when attempted with standard techniques .
the standard procedure to diagonalize say the transverse ising chain is a jordan wigner transformation followed by fourier and bogoliubov transformations @xcite .
this procedure however only diagonalizes the subspace of odd or even number of excitations at a time but not both . since the operator @xmath142 maps between the odd to the even number subspaces for any @xmath143 , an exact expression for the time dependent correlation functions @xmath144 , which would be required for deriving a master equation , can not be obtained via the above procedure @xcite .
if one uses only say the even excitation number subspace as an approximation , the errors for the obtained eigenvalues scale as @xmath145 , where @xmath1 is the length of the chain .
it is therefore often argued that only using one subspace is a good approximation for long chains . in our problem
, this reasoning can not be applied : an approximation of the required correlation functions , @xmath144 , by only considering one subspace leads to unphysical results for the master equation .
an example is given in figure [ nonphys ] .
we have therefore avoided a treatment of the spin chains with master equations .
@xmath33 ( dashed line ) and @xmath34 ( solid line ) for @xmath35 , @xmath26 , @xmath146 , @xmath42 and @xmath29 as given by the solution of a master equation for the transverse ising spin chain . the solution is not physical since probabilities are not less or equal to 1.,width=264 ]
quantum spin chains may be used as quantum channels . here
we have studied the dependence of the transfer quality and speed on the size of the spectral gap between the ground and the lowest excited state of the considered system .
two different scenarios are observed : provided the spectral gap of the chain is larger than the amount of energy initially located in the sender , close to perfect transfer happens on rather long time scales .
the transfer becomes even more perfect but slower if the coupling strength between ancillas and chain or the energy initially in the sender is lowered further .
if the energy in the sender is initially larger than the spectral gap , the behavior changes significantly . here ,
the excitations get lost into the chain implying a bad transfer quality .
this second scenario takes place on much shorter time scales than the first one .
these results may be used for two main applications : for quantum information purposes , they show , that by using a system with a large enough spectral gap one can build good quantum channels .
these channels are furthermore considerably robust against decoherence because the chain remains to a high precision in its ground state which is not significantly affected by an environment at zero temperature . in condensed matter physics , on the other hand
, the transfer speed and quality can be used to infer the size of the spectral gap and hence the location of quantum critical points .
this method is admittedly inferior to neutron scattering in magnetic composites , but it may by advantageous in optical lattices where neutron scattering is not that successfull .
the authors would like to thank sougato bose and daniel burgarth for discussions at early stages of this project as well as comments by andrew fisher .
this work is part of the qip - irc supported by epsrc ( gr / s82176/0 ) , the integrated project qubit applications ( qap ) supported by the ist directorate as contract number 015848 and was supported also by the alexander von humboldt foundation , hewlett - packard and the royal society .
bose s. phys .
lett . * 91 * , 207901 ( 2003 ) .
+ burgarth d and bose s. new j. phys . * 7 * , 135 2005 . plenio m b , hartley j , and eisert j. new j. phys . * 6 * , 36 ( 2004 ) .
+ eisert j , plenio m b , bose s , and hartley j. phys .
lett . * 93 * , 190402 ( 2004 ) . + plenio m b and semio f. new j. phys .
* 7 * , 73 ( 2005 ) .
+ perales a and plenio m b j. opt .
b 7 , 601 ( 2005 ) hastings m b phys .
lett . * 93 * , 126402 ( 2004 ) .
+ hastings m b phys .
b * 69 * , 104431 ( 2004 )
. + hastings m b phys .
b * 73 * , 085115 ( 2006 )
. + hastings m b phys .
lett . * 93 * , 140402 ( 2004 ) .
cramer m , eisert j , plenio m b and dreissig j. phys .
a * 73 * , 012309 ( 2006 ) .
+ cramer m and eisert j. new j. phys .
8 , 71 ( 2006 ) .
+ wolf m m , giedke g and cirac j i. phys . rev .
lett . * 96 * , 080502 ( 2006 ) .
sachdev s _ quantum phase transitions_. cambridge univ . press , cambridge , ( 1999 ) .
osterloh a , amico l , falci g and fazio r. nature * 416 * , 608 ( 2002 ) . verstraete f , martn - delgado m a and cirac j i. phys .
* 92 * , 087201 ( 2004 ) .
+ pachos j k and plenio m b. phys . rev .
* 93 * , 056402 ( 2004 ) .
+ kay a , lee d k k , pachos j k , plenio m b , reuter m e and rico e. optics and spectroscopy * 99 * , 339 - 356 ( 2005 ) .
larsson d and johannesson h. phys .
lett . * 95 * , 196406 ( 2005 ) .
amico l , osterloh a , plastina f , fazio r and massimo palma g phys . rev .
a * 69 * , 022304 ( 2004 ) .
+ dr w , hartmann l , hein m , lewenstein m and briegel h - j phys .
lett . * 94 * , 097203 ( 2005 ) .
+ sen de a , sen u , ahufinger v , briegel h - j , sanpera a and lewenstein m quant - ph/0507172 ( 2005 ) .
sen a , sen u and lewenstein m. phys .
a * 72 * , 052319 ( 2005 ) .
yi x x , cui h t and wang l c. quant - ph/0511026 ( 2005 ) . zurek w h , dorner u and zoller p. phys .
lett . * 95 * , 105701 ( 2005 ) .
+ polkovnikov a. phys .
b * 72 * , 161201(r ) ( 2005 )
. + dziarmaga j. phys .
95 * , 245701 ( 2005 ) hartmann m j , reuter m e , and plenio m b. new j. phys . *
8 * , 94 ( 2006 ) .
stlund s , rommer s phys .
75 * , 3537 - 3540 ( 1995 ) .
+ vidal g. phys .
lett . * 93 * , 040502 ( 2004 ) .
+ daley a j , kollath c , schollwck u and vidal g. j. stat .
exp . , p04005 ( 2004 ) .
we like to thank andrew fisher for bringing this point to our attention .
plenio m b , eisert j , dreissig j , and cramer m. phys .
* 94 * , 060503 ( 2005 ) .
lieb e , schultz t and mattis d. ann .
* 16 * , 407 ( 1961 ) .
mccoy b m. phys .
rev . a * 4 * , 2331 ( 1971 ) .
eisert j. and plenio m.b .
* 1 * , 479 ( 2003 ) | recently , there has been growing interest in employing condensed matter systems such as quantum spin or harmonic chains as quantum channels for short distance communication .
many properties of such chains are determined by the spectral gap between their ground and excited states . in particular
this gap vanishes at critical points of quantum phase transitions . in this article
we study the relation between the transfer speed and quality of such a system and the size of its spectral gap . we find that the transfer is almost perfect but slow for large spectral gaps and fast but rather inefficient for small gaps . |
this year marks the 500th anniversary of the copperplate engraving _
melencolia i _ by the great artist - mathematician albrecht d " urer ( 14711528 ) .
a good way for math lovers to celebrate this anniversary is to play a time - honored party game " that has attracted art historians and scientists for many years : guessing the nature and meaning of the composition s enigmatic stone polyhedron , illustrated in figure [ setup ] .
published attempts at playing the game date back at least a century @xcite .
in fact , it s a game that may go on indefinitely , because ( 1 ) we have no writings by d " urer definitively saying what the polyhedron is , ( 2 ) our ability to measure and analyze the polyhedron is subject to at least some small errors , and ( 3 ) the image itself has anomalous features , some of which we discuss in the appendix .
nevertheless the polyhedron , sometimes known as d " urer s solid " , seems to be accurately drawn for the most part , and authors such as macgillavry @xcite have been able to get good fits of their models by allowing for a certain amount of human error in both drawing and measuring . in this article
we discuss a possible model of the solid and compare it to the results of macgillavry and others .
we will try to make the case that one of the most convenient and effective ways to score the game is to use the cross ratio .
we show how the cross ratio works as a projectively invariant shape parameter " of the polyhedron , and how it can be used in analyzing various theories of this figure .
a commonly accepted rule of the game is the assumption that the polyhedron is formed by starting with a _ trigonal trapezohedron _ or _
trigonal deltohedron_a three - dimensional figure whose faces are six congruent rhombi ( a cube is such a figure ) .
then , with the longest diagonal held vertically , congruent tetrahedrons are cropped off the top and bottom by horizontal planes , leaving congruent equilateral triangles as the top and bottom faces .
the other six faces the truncated rhombi are thus congruent pentagons .
the shape of the solid is completely determined by the shape of one of the pentagonal faces , hence many authors describe their model of the solid by describing the shape of one of these pentagons .
in addition to rules , the game should have some method for scoring various attempts at it , at least in an informal way .
first of all , it makes no sense to play the game if we assume that d " urer was such a poor draftsman that we can simply ignore the engraving and conjecture any shape we want .
we therefore require that a proposed model of the solid should have a shape that is close to that derived by perspective analysis of the engraving .
in addition to respecting d " urer as an artist , we must acknowledge his reputation as a mathematician . in the words of the respected mathematician morris kline , of all the renaissance artists ,
the best mathematician was the german albrecht d " urer " @xcite . among other things , d
" urer owned a copy of euclid and wrote well - respected treatises on proportion , including
significantly for us the golden ratio .
thus we assume that d " urer incorporated some kind of interesting mathematical relationships into the design of the solid , rather than just sketching until he found something that looked pleasing to him .
a proposed model of the solid should include the description of such relationships .
under the assumptions of the game , the shape of the solid is completely determined by two numbers , which we refer to as shape parameters " . as we mentioned earlier , the shape of the solid is determined by the shape of one of the six congruent pentagonal faces .
the shape of a pentagonal face ( see figure [ generic - face ] ) can be specified by the acute angle @xmath0 of the rhombus and another parameter @xmath1 , which determines the level at which the rhombus is truncated .
one obvious candidate for @xmath1 would be the ratio of distances @xmath2 in figure [ generic - face ] ; that is , the fraction of @xmath3 left after truncation . however , this parameter has one disadvantage ; it can not always be measured directly in a perspective drawing of the face , because perspective often distorts the ratios of lengths . and
@xmath1 determine the shape of the truncated rhombic face and hence the shape of the solid . ] fortunately , given four collinear points @xmath4 , @xmath5 , @xmath6 , @xmath7 as in figure [ generic - face ] , there is a quantity whose value is not changed by perspective , namely the _ cross ratio _ , among other things . ]
@xmath8 defined by @xmath9 for a truncated rhombic face like the one in figure [ generic - face ] , there is a convenient relationship between the cross ratio and the ratio @xmath2 . letting @xmath10 , we have @xmath11 solving ( [ lambda ] ) for @xmath2 gives @xmath12 from ( [ lambda ] ) and ( [ bc / ac ] ) we see that the truncation of the rhombus determines the value of the cross ratio @xmath1 , and the value of @xmath1 determines the truncation of the rhombus .
since @xmath1 is projectively invariant that is , unchanged by the distortion of perspective we choose it as the second shape parameter .
perhaps the most thorough and frequently cited perspective analysis of the solid is the one done by macgillavry @xcite .
macgillavry estimated the shape parameters of the pentagonal faces by computing them several different ways , applying the rules of perspective to different parts of the engraving and averaging the results .
macgillavry estimated @xmath13 and @xmath14 .
figure [ m - face](a ) shows such a pentagon in solid gray , with macgillavry s minimum value of @xmath15 . from this value of @xmath2 and equation ( [ lambda ] )
we see that macgillavry s work implies @xmath16 our own estimate of @xmath1 is close to that of macgillavry . in figure [ setup ]
we have extended sides of the three visible pentagonal faces to locate the truncated vertices @xmath17 and @xmath18 of the rhombi . for @xmath19 , the lines @xmath20 are the centerlines of the rhombi , the points @xmath21 are the centers , and the points @xmath22 are the truncation points of the centerlines .
if our assumptions about the polyhedron are correct , and if the drawing and our measurements were perfect , then the cross ratios @xmath23 would be exactly the same for @xmath19 . in an imperfect world , we of course expect some variation
. one of the authors computed these cross ratios by importing a digital image of _ melencolia i _ into the program geogebra , and found @xmath24 to allow for imperfections as macgillavry did , we average these three values to get @xmath25 . the reader may have noticed that these estimates of @xmath26 and @xmath27 are suspiciously close to a famous number : the golden ratio @xmath28 , where @xmath29 following this clue , our model of the solid is based on the golden ratio , with parameters we denote by @xmath30 and @xmath31 .
figure [ m - face](b ) shows our proposed model of a face in black outline .
the rhombus is inscribed in a pair of golden rectangles , each with height @xmath32 and width @xmath28 .
the truncation line is easily found by drawing a @xmath33 line from the center of the rhombus until it meets an edge of the rhombus as shown .
it is straightforward to show that @xmath34 , resulting in a cross ratio , via ( [ lambda ] ) , of @xmath35 version of macgillavry s pentagonal face in solid gray . in ( b ) the golden " pentagon in black outline , with the rhombus inscribed in a pair of golden rectangles . in ( c ) macgillavry s pentagon in solid gray , and the golden pentagon in dashed black outline .
when drawn at this size , they are practically identical . ] thus the cross ratio is the golden ratio .
the acute angle @xmath30 is given by @xmath36 the values of @xmath31 and @xmath30 are very close to macgillavry s values of @xmath37 and @xmath15 .
in fact , as shown in figure [ m - face](c ) , the two pentagons are nearly indistinguishable ( macgillavry s is solid gray and ours is the dashed black outline ) . for the above reasons we ( somewhat playfully ) refer to the pentagon in ( b ) as a _
golden pentagon_. to analyze how close the two pentagons really are , observe that the half - width of the pentagon in figure [ m - face](a ) , divided by the half - width of the pentagon in ( b ) is @xmath38 .
thus if the half - widths of the pentagons are about an inch ( they re actually a little less ) , then the difference in the half - widths is less than a thousandth of an inch .
the ratios of the heights of each pentagon above the horizontal centerline is @xmath39 .
thus if these heights are about a half inch ( again they re a little less ) , then the difference between them is about three thousandths of an inch .
we conclude that from an intuitive , visual standpoint , macgillavry s minimum - angle solution of @xmath37 and @xmath15 is essentially the same as ours , minus the golden " formulation of the shape parameters : @xmath40 and @xmath41 .
over the years many theories have been advanced as to the shape of d " urer s solid , or equivalently , its pentagonal faces . with so many possibilities , any particular one of them ( including this one ) probably stands only a small chance of being right .
the advantage of this abundance of theories is that it can be as much fun testing old theories as it is cooking up new ones .
following our methods of the previous section , we show how the cross ratio can be applied to this variation of the game .
figure [ 4-grid ] depicts a model conjectured by lynch @xcite , who guessed that vertices of the solid project orthogonally onto the lattice points of a @xmath42 grid , thus linking it with a @xmath42 magic square that appears in _
melencolia i_. however , if this were true , we would have ( assuming for simplicity that the squares are 1 unit on a side ) , @xmath43 which differs significantly ( about @xmath44 ) from our measured value of @xmath25 and macgillavry s estimate of @xmath45 .
we conclude that lynch s idea is not in good agreement with d " urer s perspective rendering of the solid .
figure [ 72 ] depicts a model by schreiber having an acute rhombus angle of @xmath46 and a truncation chosen so that the face is inscribed in a circle . but
a straightforward computation shows that this requires a ratio of @xmath47 .
when substituted into ( [ lambda ] ) , this gives a cross ratio value of @xmath48 , which differs from macgillavry s estimate and ours by more than @xmath49 .
this can be checked approximately by using a ruler and protractor to draw the @xmath46 rhombus , and a circle template to choose a circle that passes through its left , right , and bottom vertices .
the circle then determines the truncation . in either case
the cross ratio of the resulting pentagon contradicts the evidence of the engraving itself .
in addition , the @xmath46 angle is @xmath50 less than macgillavry s minimum value of @xmath51 . ,
circle - inscribed face conjectured by schreiber @xcite . ]
figure [ sketch ] shows a drawing from a d " urer sketchbook found by weitzel @xcite .
weitzel conjectured that this drawing is a preliminary sketch of a face of the solid in _
melencolia i _ , and estimated the rhombus angle to be @xmath52 , apparently based on the angle at the upper left in figure [ sketch](b ) .
however , we noticed that the drawing is not perfectly symmetrical , as the other angle is about @xmath53 .
the average of these values is close to macgillavry s lower bound of @xmath51 .
a feature of the sketch that is _ not _ consistent with macgillavry s results is the truncation ratio of approximately @xmath54 indicated in figure [ sketch](b ) , apparently intended to allow for a circumscribed circle .
this is significantly different from macgillavry s value of @xmath55 for the faces of d " urer s solid . alternatively , letting @xmath56 in equation ( [ lambda ] ) yields a cross ratio @xmath1 for weitzel s model of @xmath57 which is again significantly different from our measured value of @xmath25 and macgillavry s estimate of @xmath45 .
this suggests that d " urer may have used an outer rhombus much like that in figure [ sketch ] for the face of his solid , but truncated it differently for some reason . assuming that the sketch really is for _ melencolia i _
, our suggestion for the different truncation would be that the resulting face has a cross ratio equal to the golden ratio .
under the usual assumptions , the shape of d " urer s solid can be specified by a pair @xmath58 of shape parameters , as illustrated in figure [ generic - face ] .
of the two , the cross ratio @xmath1 is the easiest to investigate empirically , because it is projectively invariant .
the two parameters , which are independent of one another , are equally important in determining the solid , but in the literature the angle @xmath0 seems to have attracted more attention .
for example , ishizu @xcite provides a table of values of @xmath0 proposed by various authors going back to 1900 , but no corresponding table of values of @xmath1 , or any other parameter that would determine the truncation of the rhombus .
we urge future players of the game to give equal consideration to both parameters , and we suggest the projectively invariant cross ratio as the parameter to determine the truncation .
interestingly , ishizu s table shows that in the literature , the proposed values of @xmath0 cluster around either @xmath59 ( largely because it is consistent with the perspective of the engraving ) or @xmath46 because of its connection with the golden ratio a connection that might have attracted d " urer , namely @xmath60 but if that is a nice connection with the golden ratio one that would appeal to d " urer the mathematician then surely the double connection @xmath61 which we propose , @xmath62 must have some appeal as well , especially since it is compatible with the perspective of d " urer s rendering of the solid . fortunately for the anniversary celebration , neither this model nor any other will end the game of guessing the intended shape of the solid .
any proposed values of @xmath0 and @xmath28 can agree only approximately with the engraving and , as we show in the appendix , the engraving itself has some anomalous features that make measurements of it even more uncertain .
there s plenty of room for more ideas and more players .
we invite the reader to play the game and invent a new and better theory ! 12 a. bogomolny , golden ratio in geometry , from interactive mathematics miscellany and puzzles , http://www.cut-the-knot.org/do_you_know/goldenratio.shtml
. accessed 27 march 2014 .
k. enomoto , d " urer no tamentai ni tsukarete , _ mizue _ * 891 * ( 1979 ) , 8895
. h. ishizu , another solution to the polyhedron in d " urer s melencolia : a visual demonstration of the delian problem , _ aesthetics _ * 13 * ( 2009 ) , 179194 . m. kline , _ mathematical thought from ancient to modern times , vol . 1 _ , oxford university press , new york , 1972 .
t. lynch , the geometric body in d " urer s engraving melencolia i , _ journal of the warburg and courtauld institutes _ * 45 * ( 1982 ) , 226232 .
c. h. macgillavry , the polyhedron in a. d " urer s melencolia i : an over 450 years old puzzle solved ? , _ nederl .
b _ * 84 * ( 1981 ) , 287294 .
s. r " osch , die bedeutung des polyeders in d " urers kupferstich ` melancholia i ' ( 1514 ) , _ fortschritte der mineralogie _ * 48 * ( 1970 ) , 8385 .
p. schreiber , a new hypothesis on d " urer s enigmatic polyhedron in his copper engraving melencolia i " , _ historia mathematica _ * 26 * ( 1999 ) , 369377 .
h. weitzel , a further hypothesis on the polyhedron of a. d " urer s engraving ` melencolia i ' , _ historia mathematica _ * 31 * ( 2004 ) , 1114 .
d " urer s rendering of the solid has certain anomalous features , a couple of which are depicted in figure [ anomaly ] .
the solid line segments @xmath63 , @xmath64 , and @xmath65 are translations of one another and therefore parallel .
observe that segments @xmath63 and @xmath64 very nearly coincide with edges of the solid , while @xmath65 noticeably diverges from the nearest edge .
so let us turn our attention away from the lines in the image to the lines in the ( imagined ) actual solid object , sitting in space .
if the solid is a trigonal trapezohedron as is usually assumed , the three edges that we are considering are parallel in space to one another , hence their images must be concurrent at a vanishing point .
but the images nearest @xmath63 and @xmath64 are essentially parallel , hence their vanishing point is at infinity ( a so - called _ ideal point _ ) . on the other hand ,
the image of the edge nearest @xmath65 is clearly not parallel to the images of the other two edges , so the images of the three edge lines are not concurrent at any point , ideal or ordinary .
this feature is either an inconsistency in the drawing , or else d " urer s conception of the solid contradicts the usual assumptions . to illustrate another anomaly in figure [ anomaly ] ,
we have also drawn the dashed line segments @xmath66 and @xmath67 parallel to one another .
segment @xmath66 coincides with the image of a face diagonal , while @xmath67 is close to but deviates slightly from the image of a bottom edge . under the usual assumptions about the solid , the diagonal and the edge are parallel in space ,
hence their images in the engraving should converge to a vanishing point on the horizon line , above and to the left of the picture in figure [ anomaly ] .
but notice that as the image of the bottom edge goes to the left , it dips below line @xmath67 , hence it actually diverges from line @xmath66 ( or equivalently , it converges in the wrong direction ) .
the bottom edge should in fact rise above line @xmath67 ( or the diagonal should dip below line @xmath66 ) as it goes to the left .
the above errors in the drawing , if they are errors , are not drastic .
indeed , most authors seem not to have noticed them . however , they highlight the importance of computing the parameters of the truncated rhomboid in multiple ways , using different parts of the image , perhaps even avoiding certain parts if they seem to be too inaccurately drawn . | on the 500th anniversary of albrecht d " urer s copperplate engraving _
melencolia i _ , we invite readers to join in a time - honored party game " that has attracted art historians and scientists for many years : guessing the nature and meaning of the composition s enigmatic stone polyhedron . our main purpose is to demonstrate the usefulness of the cross ratio in the analysis of works in perspective .
we show how the cross ratio works as a projectively invariant shape parameter " of the polyhedron , and how it can be used in analyzing various theories of this figure .
d " urer ; _ melencolia i _ ; engraving ; solid ; polyhedron ; rhombohedron ; perspective ; cross ratio 51n15 ; 51n05 |
the solar neutrino problem is one of the central issues in the neutrino physics @xcite .
nowadays , the neutrino oscillation is assumed to be the solution of the problem and the focus of the solar neutrino physics is to determine oscillation parameters : the mass difference among @xmath8 , @xmath9 and @xmath10 , and their mixing angles @xcite .
the astrophysical factor @xmath11 , defined by @xmath12 $ ] with @xmath13 the cross section of the @xmath14-capture reaction @xmath3be(@xmath15)@xmath0b and @xmath16 the sommerfeld parameter , plays an essential role in the investigation of neutrino oscillation , since the prediction value for the flux of the @xmath0b neutrino , which is intensively being detected on the earth , is proportional to @xmath1 .
the required accuracy from astrophysics is about 5% in errors . because of difficulties of direct measurements for the @xmath14-capture reaction at very low energies , alternative indirect measurements were proposed : @xmath14-transfer reactions and @xmath0b coulomb dissociation are typical examples of them . in the former the asymptotic normalization coefficient ( anc ) method
@xcite is used , carefully evaluating its validity , while in the latter the virtual photon theory ( vpt ) is adopted to extract @xmath1 ; the use of vpt requires the condition that the @xmath0b is dissociated through its pure e1 transition , the validity of which is not yet clarified quantitatively . in the present paper
we propose systematic analysis of @xmath0b coulomb dissociation by means of the anc method , instead of vpt .
an important advantage of the analysis is that one can evaluate the error of @xmath1 coming from the use of the anc method ; the fluctuation of @xmath1 , by changing the @xmath0b single - particle wave functions , can be interpreted as the error of the anc analysis @xcite . for the calculation of @xmath0b dissociation cross sections , we use the method of continuum - discretized coupled - channels ( cdcc ) @xcite , which was proposed and developed by kyushu group .
cdcc is one of the most accurate methods being applicable to breakup processes of weakly - bound stable and unstable nuclei . as a subject of the present analysis ,
four experiments of @xmath0b coulomb dissociation done at riken @xcite , gsi @xcite , msu @xcite and notre dame @xcite are available . among them
we here take up the notre dame experiment at 25.8 mev and extract @xmath1 by the cdcc + anc analysis , quantitatively evaluating the validity of the use of the anc method .
it was shown in ref .
@xcite that cdcc can successfully be applied to the msu data at 44 mev / nucleon .
however , the cdcc calculation requires extremely large modelspace ; typically the number of partial waves is 15,000 .
thus , preceding systematic cdcc + anc analysis of the experimental data at intermediate energies , we propose hybrid ( hy ) coupled - channels ( cc ) calculation by means of the standard cdcc and the eikonal - cdcc method ( e - cdcc ) , which allows one to make efficient and accurate analysis .
e - cdcc describes the center - of - mass ( c.m . )
motion between the projectile and the target nucleus by a straight - line , which is only the essential difference from cdcc . as a consequence ,
the resultant e - cdcc equations have a first - order differential form with no huge angular momenta , hence , one can easily and safely solve them .
because of the simple straight - line approximation , results of e - cdcc may deviate from those by cdcc .
one can avoid this problem , however , by constructing hy scattering amplitude from results of both cdcc and e - cdcc .
this can be done rather straightforwardly , since the resultant scattering amplitude by e - cdcc has a very similar form to the quantum - mechanical one , which is one of the most important features of e - cdcc . in the latter part of the present paper
we show how to perform the hy calculation and apply it to @xmath2ni(@xmath0b,@xmath3be@xmath4)@xmath2ni at 240 mev . in sec .
[ anc+cdcc ] we describe the cdcc + anc analysis for @xmath2ni(@xmath0b,@xmath3be@xmath4)@xmath2ni at 25.8 mev : the anc method and cdcc are quickly reviewed in subsections [ anc+cdcca ] and [ anc+cdccb ] , respectively , and numerical results and the extracted @xmath1 are shown in subsection [ anc+cdccc ] . in sec .
[ hybrid ] the hy calculation for coulomb dissociation , with the formalism of e - cdcc , is described ( subsection [ hybrida ] ) and its validity is numerically tested for @xmath2ni(@xmath0b,@xmath3be@xmath4)@xmath2ni at 240 mev ( subsection [ hybridb ] ) .
finally , summary and conclusions are given in sec .
[ summary ] .
in this section we propose cdcc + anc analysis for @xmath0b coulomb dissociation to extract @xmath1 .
first , in subsection [ anc+cdcca ] , we give a quick review of the anc method and discuss advantages of applying it to @xmath0b coulomb dissociation .
second , calculation of @xmath0b breakup cross section by means of cdcc is briefly described in subsection [ anc+cdccb ] .
finally , we show in subsection [ anc+cdccc ] numerical results for @xmath2ni(@xmath0b,@xmath3be@xmath4)@xmath2ni at 25.8 mev ; the extracted value of @xmath1 , with its uncertainties , is given .
the anc method is a powerful tool to extract @xmath1 indirectly .
the essence of the anc method is that the cross section of the @xmath3be(@xmath15)@xmath0b at stellar energies can be determined accurately if the tail of the @xmath0b wave function , described by the whittaker function times the anc , is well determined .
the anc can be obtained from alternative reactions where peripheral properties hold well , i.e. , only the tail of the @xmath0b wave function has a contribution to observables .
so far the anc method has been successfully applied to @xmath14-transfer reaction such as @xmath17be(@xmath3be,@xmath0b)@xmath18be @xcite , @xmath19n(@xmath3be,@xmath0b)@xmath20c @xcite , and @xmath21be(@xmath22)@xmath23b @xcite .
also trache _ et al . _
@xcite showed the applicability of the anc method to one - nucleon breakup reactions ; @xmath1 was extracted from systematic analysis of total breakup cross sections of @xmath0b @xmath24 @xmath3be + @xmath14 on several targets at intermediate energies . in the present paper
we apply the anc method to @xmath0b coulomb dissociation , where @xmath1 has been extracted by using vpt based on the principle of detailed balance . in order to use vpt ,
the previous analyses neglected effects of nuclear interaction on the @xmath0b dissociation , which is not yet well justified .
additionally , roles of the e2 component , interference with the dominant e1 part in particular , need more detailed investigation , although recently some attempts to eliminate the e2 contribution from measured spectra have been made . on the contrary ,
the anc analysis proposed here is free from these problems .
we here stress that as an important advantage of the present analysis , one can evaluate quantitatively the error of @xmath1 by the fluctuation of the anc with different @xmath0b single - particle potentials . comparing with ref .
@xcite , in the present anc analysis angular distribution and parallel - momentum distribution of the @xmath3be fragment , instead of the total breakup cross sections , are investigated , which is expected to give more accurate value of @xmath1 .
moreover , our purpose is to make systematic analysis of @xmath0b dissociation at not only intermediate energies but also quite low energies .
thus , the breakup process should be described by a sophisticated reaction theory , beyond the extended glauber model used in ref .
@xcite . for that purpose
, we use cdcc , which is one of the most accurate methods to be applicable to @xmath0b dissociation .
generally cdcc describes the projectile ( c ) + target ( a ) system by a three - body model as shown in fig . 1 ; in the present case c is @xmath0b and 1 and 2 denote @xmath3be and @xmath14 , respectively .
the three - body wave function @xmath25 , corresponding to the total angular momentum @xmath26 and its projection @xmath27 , is given in terms of the internal wave functions @xmath28 of c : _ jm = _
l y_jm^_0 l _ 0(r ) + _ l y_jm^l _ 0^_(k , r ) dk ; [ expand1 ] _
jm^l_jm , where @xmath29 is the total spin of c and @xmath30 is the orbital angular momentum for the relative motion of c and a ; the subscript 0 represents the initial state .
we neglect all intrinsic spins of the constituents and as a consequence c has only one bound state in the present case .
the first and second terms in the r.h.s . of eq .
( [ expand1 ] ) correspond to the bound and scattering states of c , respectively . in the latter
the relative momentum @xmath31 between c and a is related to the internal one @xmath32 of c through the total - energy conservation . in cdcc
the summation over @xmath29 and integration over @xmath32 are truncated at certain values @xmath33 and @xmath34 , respectively . for the latter ,
furthermore , we divide the @xmath32 continuum into @xmath35 _ bin_-states , each of which is expressed by a discrete state @xmath36 with @xmath37 denote a certain region of @xmath32 , i.e. , @xmath38 .
after truncation and discretization , @xmath25 is approximately expressed by @xmath39 with finite number of channels : _
jm^cdcc = _
l y_jm^_0 l _ 0(r ) + _
= 0^_max _ i=1^n _ l y_jm^l _ i
( r ) [ expand2 ] with @xmath40 .
the @xmath41 and @xmath42 are the discretized @xmath31 and @xmath43 , respectively , corresponding to the @xmath37th bin state @xmath36 . inserting @xmath44 into a three - body schrdinger equation ,
one obtains the following ( cc ) equations : _ ( _ i , r ) = _
^ v_^(r ) _ ^ ( _ i^,r ) [ eqs : cdcc ] for all @xmath45 including the initial state , where @xmath46 is the reduced mass of the c + a system and @xmath47 is the form factor defined by @xmath48 with @xmath49 the sum of the interactions between a and individual constituents of c. the cdcc equations ( [ eqs : cdcc ] ) are solved with the asymptotic boundary condition : @xmath50 where @xmath51 and @xmath52 are incoming and outgoing coulomb wave functions .
thus one obtains the @xmath53-matrix elements @xmath54 , from which any observables , in principle , can be calculated ; we followed ref .
@xcite to calculate the distribution of @xmath3be fragment from @xmath0b .
cdcc treats breakup chanels of a projectile explicitly , including all higher - order terms of both coulomb and nuclear coupling - potentials , which gives very accurate description of dissociation processes in a framework of three - body reaction dynamics .
detailed formalism and theoretical foundation of cdcc can be found in refs .
@xcite .
we here take up the @xmath0b dissociation by @xmath2ni at 25.8 mev ( 3.2 mev / nucleon ) measured at notre dame @xcite , for which vpt was found to fail to reproduce the data @xcite . the extended glauber model , used in ref .
@xcite , is also expected not to work well because of the low incident energy .
thus , the notre dame data is a good subject of our cdcc + anc analysis .
parameters of the modelspace taken in the cdcc calculation are as follows .
the number of bin - states of @xmath0b is 32 for s - state and 16 for p- , d- and f - states .
we neglected the intrinsic spins of @xmath14 , @xmath3be and @xmath2ni as mentioned in the previous subsection .
the maximum excitation energy of @xmath0b is 10 mev , @xmath55 ( @xmath56 ) is 100 fm ( 500 fm ) and @xmath57 is 1000 . for nuclear interactions of @xmath14-@xmath2ni and @xmath3be-@xmath2ni we used the parameter sets of becchetti and greenlees @xcite and moroz _
et al_. @xcite , respectively .
be fragment in the laboratory frame .
the solid and dashed lines represent the results of cdcc calculation with the parameter set of kim and esbensen - bertsch ( eb ) , respectively , for @xmath0b single particle potential .
results in the left panel correspond to @xmath58 and those with appropriate values of @xmath59 , i.e. , 0.96 for kim and 1.20 for eb , are shown in the right panel . the experimental data are taken from ref .
width=529 ] in fig .
2 we show the results of the angular distribution of @xmath3be fragment , integrated over scattering angles of @xmath14 and excitation energies of the @xmath3be + @xmath14 system . in the left panel
the results with the @xmath0b wave functions by kim _
et al . _
@xcite ( solid line ) and esbensen and bertsch @xcite ( dashed line ) , with the spectroscopic factor @xmath59 equal to unity , are shown .
after @xmath60 fitting , one obtains the results in the right panel ; one sees that both calculations very well reproduce the experimental data .
the resultant @xmath59 is 0.96 and 1.20 with the @xmath0b wave functions by kim and esbensen - bertsch , respectively , showing quite strong dependence on @xmath0b models .
in contrast to that , the anc @xmath61 calculated by @xmath62 with @xmath63 the single - particle anc , is found to be almost independent of the choice of @xmath0b wave functions , i.e. , @xmath64 ( @xmath65 ) .
thus , one can conclude that the anc method works in the present case within 1% of error .
following ref .
@xcite we obtained the following result : @xmath66 where the uncertainties from the choice of the modelspace of cdcc calculation ( 1.5% ) and the systematic error of the experimental data ( 10% ) are also included .
although the quite large experimental error prevents one from determining @xmath1 with the required accuracy ( 5% ) , the cdcc + anc method turned out to be a powerful technique to determine @xmath1 with small theoretical uncertainties .
more careful analysis in terms of the charge distribution of @xmath3be , nuclear optical potentials and roles of the intrinsic spins of the constituents , are being made and more reliable @xmath1 will be reported in a forthcoming paper .
in sec . [ anc+cdcc ] we showed that the cdcc + anc analysis for the @xmath0b dissociation at 25.8 mev gives @xmath1 with good accuracy , being free from rather ambiguous assumptions made in the previous analyses using vpt . in ref .
@xcite it was shown that the anc method works well for one - nucleon breakup reactions at intermediate energies .
also cdcc turned out to almost perfectly reproduce the parallel - momentum distribution of @xmath3be fragment from @xmath67pb(@xmath0b,@xmath3be@xmath4)@xmath67pb at 44 mev / nucleon @xcite .
thus , it is expected that accurate determination of @xmath1 can be done by the cdcc + anc analysis of @xmath0b coulomb dissociation measured at riken @xcite , gsi @xcite and msu @xcite . from a practical point of view
, however , cdcc calculation including long - ranged coulomb coupling - potentials requires extremely large modelspace rather difficult to handle ; typically the number of partial waves is 15,000 for the msu data @xcite .
although interpolation technique for angular momentum reduces the number of cc equations to be solved in terms of @xmath26 , those with huge angular momenta are rather unstable and careful treatment is necessary . in this sense , it seems almost impossible to apply cdcc to the gsi data at 250 mev / nucleon , where @xmath57 is expected to exceed 100,000 . on the contrary ,
semi - classical approaches , expected to work quite well at intermediate energies , are free from any problems concerned with huge angular momenta .
the accuracy of such semi - classical calculations , however , is difficult to be evaluated quantitatively , although one may naively estimate the error is only less than about 10% or so .
it should be noted that our goal is to determine @xmath1 with more than 95% accuracy , which requires definite estimation of the error of the calculation . in this section
we propose hy calculation of @xmath0b dissociation at intermediate energies , constructing hy scattering amplitude ( @xmath68 matrix ) from partial amplitudes with quantum - mechanical ( qm ) and eikonal ( ek ) cc calculations ; for the latter we use a new version of cdcc , that is , the eikonal - cdcc method ( e - cdcc ) .
the formalism of e - cdcc and calculation of the hy amplitude are described in subsection [ hybrida ] and the validity of the hybrid calculation is tested for @xmath2ni(@xmath0b,@xmath3be@xmath4)@xmath2ni at 240 mev in subsection [ hybridb ] .
we start with the expansion of the total wave function @xmath69 : ( * r*,*r * ) = _ i m _ i , m(*r * ) e^-i(m - m_0)_r _ i m(r,_r ) , [ psinew2 ] where @xmath70 is the projection of @xmath29 on the @xmath71-axis taken to be parallel to the incident beam ; @xmath72 is the discretized internal - wave - function of c , calculated just in the same way as in the standard cdcc .
the symbol @xmath73 used in subsection [ anc+cdccb ] , which denotes a discretized quantity , is omitted here for simplicity .
we make the following ek approximation : _
c(r,_r ) _ c(b , z ) e^i*k*_c(b ) , [ eikonal ] where @xmath74 denotes channels \{@xmath37 , @xmath29 , @xmath70 } together and the wave number @xmath75 is defined by k_c^2(b)= e-_i,- [ capk ] with @xmath63 the impact parameter ; the direction of @xmath76 is assumed to be parallel to the @xmath71-axis . inserting eqs . ( [ psinew2 ] ) and ( [ eikonal ] ) into a three - body schrdinger equation and neglecting the second order derivative of @xmath77 , one can obtain the following e - cdcc equations :
k_c^(b ) _ c^(b)(z ) = _ c f^(b)_cc(z ) _
c^(b)(z ) e^i(k_c^(b)-k_c^(b ) ) z [ cceq4 ] for all @xmath74 including @xmath78 , with @xmath79 $ ] .
we put @xmath63 in a superscript since it is not a dynamical variable but an input parameter .
equations ( [ cceq4 ] ) are solved with the boundary condition @xmath80 .
since the e - cdcc equations are first - order differential ones and contain no coefficients with huge angular momenta , they can easily and safely be solved . using the solutions of eq .
( [ cceq4 ] ) , the scattering amplitude with e - cdcc is given by f_c0^e = - _
c f^(b)_cc(z ) e^-i(m - m_0)_r e^i ( * k_c^(b)*-*k_c^(b ) * ) _ c^(b)(z ) d*r*. [ f ] making use of the following forward - scattering approximation : ( * k_c^(b)*-*k_c^(b ) * ) -k_c^(b)_f b+(k_c^(b)-k_c^(b))z , [ c ] one obtains f_i m , i_0 _ 0 m_0^e = k_i m^(b ) e^-i(m - m_0)_r e^-ik_i m^(b ) _ f b ( s_i m , i_0 _ 0 m_0^(b ) - _ i i_0__0_m m_0 ) bdbd_r , + [ f4 ] where the ek @xmath53-matrix elements are defined by @xmath81 .
we then _ discretize _ @xmath82 : f_i m , i_0 _ 0 m_0^e & = & _ l k_i m^(b_l^mid ) + & & ( s_i m , i_0 _ 0 m_0^(b_l^mid ) - _ i i_0__0_m m_0 ) _ b_l^min^b_l^max bdb , [ f6 ] where @xmath83 , @xmath84 and @xmath85 are defined through @xmath86 , @xmath87 and @xmath88 , respectively . in deriving eq .
( [ f6 ] ) we neglected the @xmath63-dependence of @xmath89 , @xmath90 $ ] and @xmath91 within a small size of @xmath63 corresponding to each @xmath30 .
after manipulation one can obtain f_i m , i_0 _ 0 m_0^e _
l i^(m - m_0 ) y_l m - m_0( ) ( s_i m , i_0 _ 0 m_0^(b_l^mid ) - _ i i_0__0_m m_0 ) , + [ f7 ] which has a similar form to that of the standard cdcc : f^q_i m , i_0 _ 0 m_0 & = & _ l _ j=|l-|^l- _ l_0=|j-_0|^j-_0 ( l_0 0 _ 0 m_0 | j m_0 ) ( l m_0-m m | j m_0 ) + & & ( s_i l , i_0 l_0 _ 0^j-_i i_0 _ l l_0__0 ) ( -)^m - m_0y_l m - m_0( ) .
[ fq ] the construction of the hy scattering amplitude @xmath92 can be done by : f_i m , i_0 _ 0 m_0^h _
ll_c f_l^q + _
l > l_c f_l^e , [ fh ] where @xmath93 ( @xmath94 ) is the @xmath30-component of @xmath95 ( @xmath82 ) and @xmath96 represents the connecting point between the qm and ek calculations , which is chosen so that @xmath94 coincides with @xmath93 for @xmath97 .
one sees that eq .
( [ fh ] ) includes all qm effects necessary through @xmath93 , and also interference between the lower and higher @xmath30-regions .
it should be noted that derivation of eq .
( [ f7 ] ) , which leads one to eq .
( [ fh ] ) rather straightforwardly , is one of the most important features of e - cdcc .
actually , the present ek calculation is very simple , i.e. , e - cdcc equations ( [ cceq4 ] ) contain no correction terms to the straight - line approximation .
however , this is not a defect but a merit of e - cdcc , since such a simplest calculation is enough to describe scattering processes , if combined with the result of the qm calculation taking an appropriate @xmath96 . in the above formulation we neglected the coulomb distortion . in order to include it , we use @xmath98 instead of eq .
( [ eikonal ] ) @xcite , where @xmath99 is the coulomb wave function .
the formulation of @xmath82 can then be done just in the same way as above . in order to see the validity of the hy calculation with cdcc and e - cdcc
, we analyze @xmath2ni(@xmath0b,@xmath3be@xmath4)@xmath2ni at 240 mev .
the number of bin - states of @xmath0b is 16 , 8 and 8 for s- , p- and d - states , respectively , and @xmath100 is 4000 . as for the @xmath0b wave function , the parameter set by kim _ et al . _
@xcite was adopted . for nuclear interaction between @xmath3be and @xmath2ni
we used the global potential for @xmath3li scattering by cook _
et al_. @xcite .
other parameters are taken just the same as in subsection [ anc+cdccc ] .
ni(@xmath0b,@xmath3be@xmath4)@xmath2ni at 240 mev .
the solid and dashed lines show , respectively , the results of the qm and ek calculations .
the dotted ( dash - dotted ) line represents the hy result with @xmath101 ( 400 ) .
, width=529 ] in the left and right panels in fig .
3 we show the elastic cross section ( rutherford ratio ) and the total breakup one , respectively , as a function of scattering angle in the center - of mass ( c.m . ) frame .
the solid , dashed and dotted lines represent the results of the qm , ek and hy calculations , where @xmath96 is taken to be 4000 , 0 and 200 , respectively . in the right panel the result of the hy calculation with @xmath102 is also shown by the dash - dotted line .
the agreement between the qm and hy calculations with an appropriate value of @xmath96 for the latter , namely , 200 ( 400 ) for elastic ( breakup ) cross section , is excellent ; the error is only less than 1% .
one also sees that difference between the ek and qm results is appreciable .
since the present ek calculation is quite simple , as mentioned in the previous subsection , this does not directly show the fail of ek approximation .
however , it seems quite difficult for ek calculation to obtain perfect agreement with the result of the qm one . on the contrary ,
the hy calculation turned out to be applicable to analyses of @xmath0b dissociation to extract @xmath1 , where very high accuracy is required .
( solid line ) and 400 ( dashed line ) .
b ) the s - state breakup cross sections by the qm calculation with @xmath103 ( dashed line ) and the ek calculation with @xmath104 ( dotted line ) .
the dash - dotted line is the incoherent sum of the dashed and dotted lines and the solid line is the hy result with @xmath102 , namely , the coherent sum of the two .
, width=529 ] we show in the left panel of fig .
4 the s - state breakup cross sections by the qm calculation ; the solid and dashed lines correspond to the calculation with @xmath105 and 400 , respectively .
one sees big difference between the two , which shows that the partial scattering amplitudes for larger @xmath30 indeed has an essential contribution to the breakup cross section . in the right panel
we show the s - state breakup cross sections by the qm calculation with @xmath103 ( dashed line ) and the ek calculation with @xmath104 ( dotted lines ) .
the dash - dotted line is the incoherent sum of the two , which deviates from the hy result shown by the solid line .
thus , one sees that the essence of the present hy calculation is the construction of the hy scattering amplitude not the hy cross section .
we show in fig .
5 the p - state breakup cross sections by the qm calculation with ( dashed line ) and without ( solid line ) adiabatic ( ad ) approximation .
one sees that the ad approximation increases the breakup cross section about 10% at forward angles .
the oscillation shown by the dashed line seems to indicate the ad calculation is not valid in the present case , probably for higher partial waves .
we found just the same features in the ek calculation .
( solid line ) and @xmath94 ( dashed line ) for the \{@xmath29 , @xmath70 } @xmath106 \{@xmath107 , @xmath108 } @xmath106 \{1 , 0 } component .
, width=245 ] ( solid line ) and @xmath94 ( dashed line ) for the \{@xmath29 , @xmath70 } @xmath106 \{@xmath107 , @xmath108 } @xmath106 \{1 , 0 } component . ,
width=264 ] comparison between @xmath93 ( solid line ) and @xmath94 ( dashed line ) in eq .
( [ fh ] ) , corresponding to the \{@xmath29 , @xmath70 } @xmath106 \{@xmath107 , @xmath108 } @xmath106 \{1 , 0 } component , is made in fig .
one sees that the difference between @xmath93 and @xmath94 is appreciable for smaller @xmath30 , around 100 in particular , and as the larger @xmath30 becomes , the better agreement is obtained . for @xmath109 ,
the difference is not visible , which is consistent with the result shown in the right panel of fig .
thus , the hy calculation of @xmath0b coulomb dissociation turned out to allow one to make efficient and accurate analysis .
the method is expected to be applicable to the experimental data measured at not only riken and msu ( at several tens of mev / nucleon ) but also gsi ( at 250 mev / nucleon ) , i.e. , systematic analysis of @xmath0b coulomb dissociation for wide energy regions can be done .
in the present paper we propose systematic analysis of @xmath0b coulomb dissociation with the asymptotic normalization coefficient ( anc ) method . an important advantage of the use of the anc method is that one can extract the astrophysical factor @xmath1 evaluating its uncertainties quantitatively , in contrast to the previous analyses with the virtual photon theory ( vpt ) . in order to make accurate analysis of the measured spectra in dissociation experiments , we use the method of continuum - discretized coupled - channels ( cdcc ) , which was developed by kyushu group .
the cdcc + anc analysis was found to work very well for @xmath2ni(@xmath0b,@xmath3be@xmath4)@xmath2ni at 25.8 mev measured at notre dame , and we obtained @xmath110 ( evb ) , which is consistent with both the latest recommended value @xmath111 evb @xcite and recent results of direct measurements @xcite .
the cdcc + anc analysis is expected to work well also at intermediate energies . from a practical point of view
, however , cdcc calculation for @xmath0b coulomb dissociation at several tens of mev / nucleon requires extremely large modelspace , typically about 15,000 partial waves are needed . in order to make efficient analysis at intermediate energies , we introduce a new version of cdcc , that is , the eikonal - cdcc method ( e - cdcc ) .
e - cdcc describes the center - of - mass ( c.m . )
motion between the projectile and the target nucleus by a straight - line and treats the excitation of the projectile explicitly , by constructing discretized - continuum - states same as in cdcc .
the resultant e - cdcc equations are easily and safely be solved , since they have a first - order differential form and no huge angular momenta , in contrast to the cdcc equations .
inclusion of the coulomb distortion can be done with the use of coulomb wave functions instead of plane waves in eikonal ( ek ) approximation .
one of the most important features of e - cdcc is that the resultant scattering amplitude @xmath82 has a very similar form to the quantum - mechanical ( qm ) one , i.e. , @xmath82 is expressed by the sum of partial amplitudes @xmath94 , using relation between the angular momentum @xmath30 and the impact parameter @xmath63 .
this allows one to construct the hybrid ( hy ) amplitude @xmath92 rather straightforwardly ; @xmath92 is given by the sum of the partial amplitudes calculated by cdcc for smaller @xmath30 , @xmath112 , and those by e - cdcc for larger @xmath30 , @xmath113 , where @xmath96 is the connecting value .
the hy calculation make the cdcc + anc analysis much simple , retaining its accuracy ; all qm effects necessary can be included through @xmath93 .
the validity of the hy calculation is tested for @xmath2ni(@xmath0b,@xmath3be@xmath4)@xmath2ni at 240 mev .
the hy calculation turned out to perfectly reproduce the elastic and total breakup cross sections obtained by the qm one , namely , the error is only less than 1% ; the appropriate value of @xmath96 is found to be 400 ( 200 ) for the total - breakup ( elastic ) cross section , which is much smaller than the required maximum value of @xmath30 , i.e. , @xmath105 .
calculation with a hy cross section , not hy amplitude , was found to fail to reproduce the corresponding qm result , which shows the importance of the interference between the lower ( qm ) and higher ( ek ) regions of @xmath30 . in conclusion ,
systematic analysis of @xmath0b coulomb dissociation with the anc method and the hy coupled - channels ( cc ) calculation is expected to accurately determine @xmath1 , with reliable evaluation of its uncertainties .
an extracted @xmath1 from the riken , msu and gsi data , combined with that from the notre dame experiment , will be reported in near future .
the authors wish to thank m. kawai , t. motobayashi and t. kajino for fruitful discussions and encouragement .
we are indebted to the aid of jaeri and rcnp , osaka university for computation .
this work has been supported in part by the grants - in - aid for scientific research of the ministry of education , science , sports , and culture of japan ( grant nos .
14540271 and 12047233 ) . | systematic analysis of @xmath0b coulomb dissociation with the asymptotic normalization coefficient ( anc ) method is proposed to determine the astrophysical factor @xmath1 accurately .
an important advantage of the analysis is that uncertainties of the extracted @xmath1 coming from the use of the anc method can quantitatively be evaluated , in contrast to previous analyses using the virtual photon theory ( vpt )
. calculation of measured spectra in dissociation experiments is done by means of the method of continuum - discretized coupled - channels ( cdcc ) . from the analysis of @xmath2ni(@xmath0b,@xmath3be@xmath4)@xmath2ni at 25.8 mev , @xmath5 ( evb )
is obtained ; the anc method turned out to work in this case within 1% of error . preceding systematic analysis of experimental data at intermediate energies , we propose hybrid ( hy ) coupled - channels ( cc ) calculation of @xmath0b coulomb dissociation , which makes numerical calculation much simple , retaining its accuracy .
the validity of the hy calculation is tested for @xmath2ni(@xmath0b,@xmath3be@xmath4)@xmath2ni at 240 mev .
the anc method combined with the hy cc calculation is shown to be a powerful technique to obtain a reliable @xmath1 .
# 1#2 # 1#2 # 1 # 1 # 1 # 1 kazuyuki ogata , m. yahiro@xmath6 , y. iseri@xmath7 , t. matsumoto , n. yamashita and + m. kamimura _ department of physics , kyushu university _
@xmath6department of physics and earth sciences , university of the ryukyus @xmath7department of physics , chiba - keizai college |
lhcb is a dedicated experiment on b - quark physics , currently under construction at the large hadron collider ( lhc ) .
it will profit from the 500 @xmath5 b - hadron production cross section in the 14 tev proton - proton collisions at lhc ( i.e. about 1% of the total visible cross section ) to make precise measurements of cp - violation and rare decays of the b - mesons @xcite . by over - constraining the cabibbo - kobayashi - maskawa matrix elements
, lhcb will hopefully be able to observe subtle inconsistencies with the standard model , therefore providing indications of new physics .
a sketch of the lhcb detector is shown in fig .
[ detector ] .
profiting from the large @xmath6 cross section , lhcb will collect large samples of @xmath0 and @xmath1 decays . combining the cp - violating observables of the decays @xmath2 and @xmath3
it is possible , by means of a method proposed by r. fleischer @xcite , to extract the @xmath4 angle of the unitarity triangle . in section 2 the selection strategy , implemented to identify
the @xmath2 and @xmath3 decays within the current lhcb monte carlo simulation framework , is described and the estimated annual event yields and background levels are quoted . by using the results shown in section 2 , the sensitivities achievable on the cp - violating observables can be determined .
this is described in section 3 .
to evaluate the performance of lhcb in the reconstruction of @xmath2 and @xmath3 decays , a full geant monte carlo simulation has been performed , followed by realistic pattern recognition algorithms for the track reconstruction and particle identification .
the monte carlo event samples used for this analysis are shown in tab .
[ mc samples ] . .[mc
samples]monte carlo event samples used to study the selection of @xmath2 and @xmath3 decays .
the other channels are studied as potential sources of background . [ cols="^,^",options="header " , ] the previous results for the @xmath3 channel are obtained for a value of the mixing parameter @xmath7 . due to the finite proper time resolution of the spectrometer ( about @xmath8 ) , higher values of @xmath9 , i.e. faster @xmath10 oscillations , lead to larger errors for the measurements of @xmath11 and @xmath12 .
thus , a study of the dependence of the resolution of @xmath11 and @xmath12 on @xmath9 , up to @xmath13 , has been performed .
the results of this study for @xmath11 are shown in fig .
[ xsstudy ] ( the dependence for @xmath12 is identical and is omitted here ) . as it can be seen the error on the asymmetries increases by about a factor 1.6 at @xmath13 with respect to @xmath7 .
the results presented in this paper show the potential of lhcb in collecting large samples of @xmath14 and @xmath3 decays .
they are obtained by using a full and realistic monte carlo simulation of the detector response .
the number of @xmath14 offline - selected decays are 27000/year , while for @xmath3 it is 35000/year , with background - to - signal ratios respectively of about 0.8 and 0.55 .
the sensitivity on the cp - violating observables achievable by lhcb for these channels has been studied in detail , yielding very encouraging results .
the estimated precision on @xmath11 and @xmath12 after one year of data taking , assuming @xmath9 to be equal to 20 , is 0.054 for the @xmath14 channel and 0.043 for the @xmath3 channel . by means of these observables it is possible to extract the @xmath4 angle of the unitarity triangle .
in fact , @xmath11 and @xmath12 , related to the @xmath2 decay , can be expressed in the framework of the standard model as : @xmath15 @xmath16 where @xmath17 is the @xmath18 mixing phase , while @xmath19 and @xmath20 parametrize - sloppily speaking - the penguin over tree amplitude ratio of the decay transition @xcite .
analogously , @xmath11 and @xmath12 , related to the @xmath3 decay , can be written as @xmath21 @xmath22 where @xmath23 is the @xmath10 mixing phase , @xmath24 , and the @xmath25 and @xmath26 parameters are analogous to @xmath19 and @xmath20 for the @xmath3 transition . in the limit of exact u - spin symmetry of the strong interactions
the relations @xmath27 and @xmath28 hold , and the measurements of the four asymmetry coefficients allow to determine @xmath29 and @xmath4 simultaneously , provided that @xmath30 is determined elsewhere ( e.g. through the @xmath31 transition ) or considered negligibly small , as it is expected in the standard model .
moreover @xmath29 will be accurately known at the time of lhcb , thus allowing a more precise determination of @xmath4 .
the resolution on @xmath4 achievable at lhcb by using this method , in case of negligible u - spin breaking effects , is of the order of few degrees .
the author is grateful to t. nakada and o. schneider for their encouragement and suggestions .
special thanks to u. marconi for his help and practical support . | lhcb will collect large samples of @xmath0 and @xmath1 decays . combining the cp - violating observables of the decays
@xmath2 and @xmath3 it is possible to extract the @xmath4 angle of the unitarity triangle .
the selection of these decays within the current lhcb simulation framework is outlined and the expected annual event yields and background - to - signal ratios are quoted . then , the results of a study on the sensitivity that lhcb can achieve for the corresponding cp - violating observables are presented . |
a wealth of activity in the field of graphene , following the seminal work of novosolev and geim @xcite , has paved way for silicene @xcite , a silicon analogue of graphene .
this new entrant might have a potential to begin another fertile direction in condensed matter science and technology . replicating a rich graphene physics , and seeing silicene as a ` carbon copy ' of graphene
has been a part of recent efforts .
stable silicene layer has been created on metallic substrates , ag , zrb@xmath3 and ir .
interesting arpes and stm results have started coming out @xcite .
as for germanene , fully hydrogenated stable germanane ( geh , analogue of ch , graphane ) layer with a wide band gap has been exfoliated @xcite . finding a suitable low energy model for strongly interacting quantum matter continues to be a challenge . in these efforts , a careful look at the very experimental results that we wish to understand can guide us to correct modeling .
experiments in turn get guidance from theoretical developments
. the synergy continues .
this is true from standard model building in elementary particle physics to standard model building for cuprate .
silicene is no exception .
currently prevalent view is that neutral silicene is a dirac metal , a semi - metal qualitatively similar to graphene @xcite .
purpose of the present paper is to offer a different view point that neutral silicene is a mott insulator .
if proved correct , our provocative proposal will make silicene different from semi metallic graphene in a fundamental fashion and open new avenues for physics and new technology , arising from strong electron correlations .
it is well appreciated now , thanks to the path breaking discovery of high tc superconducting cuprates by bednorz and muller @xcite and subsequent resonating valence bond theory by anderson and collaborators @xcite , that mott insulators and strong electron correlations are seats of a variety of rich physics and phenomena .
in addition to superconductivity it includes , quantum spin liquids , emergent fermions with fermi surfaces , gauge fields , quantum order , topological order and so on .
further , inspired by certain recent theoretical development @xcite there is a debate and search for quantum spin liquids in honeycomb lattice hubbard model @xcite .
we believe that silicene and germanene will fit well into the discussion as real candidate materials , albeit with some additional novel features . several ab - initio calculations@xcite , many body theory @xcite and quantum chemical calculations - insights @xcite are available for silicene .
interestingly there is a differing view , which has not been well appreciated .
existing solid state many - body calculations predict stable free standing puckered silicene that is a semi - metal similar to graphene .
however , quantum chemical methods and insights doubt existence of a stable free standing silicene @xcite because of radicalization / reactivity and reduced aromaticity , arising from a weakened p-@xmath0 bond .
it is clear that theory of silicene is challenging and less understood compared to graphene , because of a growing importance of electron electron interaction and a soft c - axis displacement ( puckering ) degree of freedom , arising from possibility of sp@xmath9 mixing . to make progress ,
known phenomenology and various insights need to be used .
our model and theory is aimed to initiate new discussions , theoretical and experimental studies .
the present article is organized as follows .
we interpret certain existing theoretical results , mott s argument for mott transition and quantum chemical insights as providing support for our proposal of a narrow gap mott insulating state for neutral silicene .
we also identify a set of anomalous experimental results that point to a mott insulating state : i ) a significant band narrowing of the p-@xmath0 bands below a robust ` dirac mass gap ' , around k and k points in the brillouin zone , seen in arpes of silicene grown on zrb@xmath3 @xcite , ii ) observation of a large _ superconducting _ gap at fermi level with an anomalous @xmath10 20 , in sts study of silicene on ag ( 111 ) surface , below 35 k @xcite , iii ) absence of landau level splitting in an sts study up to 7 tesla @xcite , iv ) ubiquitous and temperature dependent structural reconstructions @xcite and v ) appearance of _ fermi arc _ like pockets at m points on electron doping by alkali metal @xcite .
we suggest a heisenberg model , also containing multi - spin interaction to describe spin dynamics of the narrow band mott insulator , and a tj model for the spin - charge dynamics of doped mott insulator .
then we discuss the aforementioned experimental anomalies in the light of our model .
prospects for high tc superconductivity , within our model , is discussed next . in view of a larger and favorable tj parameters , in comparison to layered cuprates ,
there is a distinct possibility of _ a tc approaching room temperature scale _ provided competing interactions are taken care of .
using existing works we come to the conclusion that superconductivity is likely to be a chiral spin singlet d + i d superconductivity .
superstructures , not common in graphene but ubiquitous in silicene on substrates , are interpreted as arising from a strong response of c - axis deformable mott localized p-@xmath0 electrons , to substrate perturbations , through site dependent sp@xmath9 mixing .
a model is also presented .
the present paper assumes that tj model in 2d describes kosterlitz - thouless superconductivity for a range of doping . while there is no rigorous proof for this assumption , it is well certified by a body of analytical , numerical and experimental efforts available for the square lattice cuprates , ever since bednorz muller s discovery and the beginning of rvb theory .
existing study on honeycomb lattice tj model indicate that spin singlet superconductivity continues to be a dominating phase over a range of doping , but with d + i d order parameter symmetry .
our present work is a natural extension of our earlier work on graphene @xcite .
it has been our view that graphite and graphene are semi - metals having an unusually broad band , should show striking electron correlation effects , even though electron - electron interaction strengths are moderate compared to the band width . according to us , _ reduced 2-dimensionality of graphene underlies the amplification of electron correlation effects_. our earlier prediction of spin-1 collective mode in neutral graphene @xcite and very high tc superconductivity in doped graphene @xcite are based on use of a moderate electron repulsion strength .
our thesis in the present article is that 2d silicene and germanene , having @xmath11 of graphene band width , but @xmath12 of graphene coulomb interaction strength should exhibit even more striking electron correlation effects . in a recent article
we have suggested a _ five fold way _ to new high tc superconductors @xcite .
one of the ways is _
graphene route_. silicene and germanene are most likely to lie in the graphene route and might help us achieve room temperature superconductivity .
starting with hexasilabenzene , si@xmath13h@xmath13 , a silicon analogue of benzene , chemists have wondered about existence of stable planar p-@xmath0 bonded si based molecules @xcite .
their concern is that an increased si - bond length , a simple consequence of a 60 % increase in atomic radius , will weaken the p@xmath14 bond , leading to reduced aromaticity and increased chemically reactivity .
it is also an experimental fact that free space si@xmath13h@xmath13 molecule has not been synthesized so far .
silicene is an infinitely extended planar p-@xmath0 bonded network of si atoms .
dft calculations and beyond argue for a stable free standing graphene .
on the other hand , sheka in 2009 @xcite and hoffman @xcite very recently , have questioned stability and very existence of free standing silicene because of reduced aromaticity and enhanced radicalization .
radicalization in quantum chemistry is creation of lone electron states in certain molecular orbitals in the ground state ; the loners are generically weakly coupled to other loners , if present . in the context of periodic systems such as a crystalline solid we interpret it to mean an extreme mott localization and formation of nearly decoupled spins . in the narrow gap mott insulating state
we are envisaging for silicene , we do nt expect weakly coupled lone spins - there is a strong super ( kinetic ) exchange coupling and multiple ring exchange interactions among spins , as we will see .
free standing silicene has not been seen experimentally . as mentioned earlier stable silicene on metallic substrates , ag , ir and zrb@xmath3
have been synthesized . a possible key role of metallic substrate in stabilizing a silicene mono layer seen in experiments is an important issue that needs to be addressed further .
in the following discussion we refer to a _ stable mott insulating single layer silicene _ in the following sense .
we assume that metallic substrate stabilizes a single layer silicene and at the same time does not significantly modify the mott insulating character that we are after .
in this section , using recent estimates of hubbard u , neighbor repulsions and mott s argument for metal insulator transition invoking screened long range coulomb , we will show that silicene is a mott insulator .
we begin with a summary of basic quantum chemistry and band theory results for silicene .
silicon , located just below carbon in the periodic table has a larger atomic radius @xmath1 1.17 @xmath2a , in contrast to a smaller value @xmath1 0.77 @xmath2 for carbon .
this leads to a _ significant _
3p-@xmath0 _ bond stretching _ : si - si distance is @xmath1 2.3 @xmath2 , while c - c distance is only 1.4 @xmath2 in graphene . unlike graphene , si allows a small sp@xmath9 admixture to sp@xmath15 bonding , resulting in puckered sigma bonds .
that is , si atoms of the two triangular sublattices of the honeycomb net undergo a small and opposite displacements , leading to a net c - axis stretching @xmath16 au .
electronic structure calculations @xcite predict the above and a semi metallic band structure qualitatively similar to graphene , containing two dirac cones at k and k points . a major differences from graphene
is a 3 fold reduction in the @xmath0/@xmath17 band widths .
following a pioneering work of sorella and tosatti @xcite a recent estimate@xcite gives an accurate critical value @xmath18 3.8 , for metal to mott transition in hubbad model on honeycomb lattice . in a very recent many body theory schuler et al .
@xcite estimate a value @xmath18 4.1 for silicene .
this puts silicene on the mott insulator side , close to the phase transition point . however , schuler et al .
argue that inclusion of nearest neighbor repulsion @xmath19 2.31 , will reduce u to an effective u@xmath20 = u - v and bring silicene back to a semi metallic state like graphene .
in what follows we will show that on the contrary presence of finite v will hasten mott localization and reinforce a mott state at a first order transition .
we use mott s argument @xcite for instability of the metal to the insulating state .
the argument begins by asking whether _ screened coulomb interaction _ present in the metallic state is sufficient to form a quantum mechanical _ bound state of an electron to the hole it left behind at its home site_. within a hubbard model idealization this happens when band width becomes comparable to on site u ; it also means that at very large u every site binds a lone daughter electron . in mott s
argument the transition is preempted by ( a growing ) less screened long range interaction , leading to a feed back and first order phase transition .
this aspect is not contained in the simple hubbard model .
let us assume that in addition to u we have a non - zero nearest neighbor coulomb interaction v , which is a leading component of the long range part of screened coulomb interaction .
if we use mott s arguments , v will add to the already existing on site attraction u between an electron and the hole it left behind .
this additional attraction ( widening the potential well ) will hasten bound state formation .
very crudely , the on site hubbard u gets enhanced : u@xmath21 u + @xmath22v , where @xmath23 1
. thus we conclude , by what we believe to be a correct use of important estimates of schuler et al .
@xcite that silicene is in the mott insulating side of the metal insulator transition point .
we will discuss this point in some detail in a separate paper . in this section we briefly review experimental results in silicene which are anomalous , from the point of view of a semi metal or dirac metal .
but , as we will discuss latter , they are normal from mott insulator point of view .
@xmath24 : in a work that has received a lot of attention , vogt et al .
@xcite using arpes show presence of a dirac cone dispersion over a wide energy range below fermi level , with a v@xmath25 comparable to that in graphene .
there is an ongoing debate @xcite on whether this is primarily a silicene p-@xmath0 band or primarily a ag metallic bands or a strong hybrid .
we believe that a strong metallic screening from ag has caused an insulator to metal transition in silicene through _ self doping _ - we will address this issue in a later section . on the other hand ,
recent arpes study of silicene grown on zrb@xmath3 by fleurence et al .
@xcite very different behavior .
zrb@xmath3 is a low carrier density metal with small fermi pockets .
it is expected to have less electronic influence on silicene .
fleurence et al .
find a remarkably narrow band around around k and k points ( marked x@xmath3 in figure 1b ) , lying below a finite gap @xmath1 0.3 ev .
its spectral weight vanishes in two - third of the brillouin zone .
further 3 silicene bands denoted by x@xmath26 , x@xmath3 and x@xmath3 all lie in a narrow energy interval of 0.5 ev .
an extrapolation from the shape of the visible silicene bands give us a total band width @xmath1 1 ev .
this is to be contrasted with the total @xmath27 band width of 6 ev as given by lda calculations .
thus there is a band width renormalization by a factor of 6 .
this spectrum is reminiscent of hole spectrum in mott insulating .
@xmath28 : chen et al . have observed @xcite superconductor like gap structure in their sts study of silicene on ag and a tc @xmath1 35 to 40 k. they give convincing arguments in support of a gap arising from superconductivity .
a large @xmath29 makes it anomalous .
further experiments are needed to substantiate this important result .
@xmath30 : an stm work @xcite in the presence of a magnetic field as large as 7 tesla , do not find an expected landau level splitting . in our view , a doped hole in a mott insulator looses nearly all its quasi particle weight because of the strong electron - electron interaction . we find that quasi particle line broadening , as inferred from experimental result is larger than the expected landau level splitting , making it invisible in sts measurements . in graphene , because of quasi particle coherence , one sees landau level structures for similar magnetic fields in the sts measurements . @xmath31 recent experiments on silicene grown on metal surfaces exhibit @xcite lattice reconstructions , @xmath32 , @xmath33 , @xmath34 , 4 @xmath35 4 etc .
they are temperature dependent and some of them exhibit phase transitions .
further , reconstructions are accompanied by _
space dependent sp@xmath9 mixing and consequent bond length modulation_. even within band theory there is no fermi surface .
so a standard route for density wave instabilities namely fermi surface nesting is absent .
this explains why reconstructions or cdw order is not seen in graphene .
graphene is stiff , in view of a @xmath1 9 ev wide filled @xmath0-band and an associated large _
aromaticity_. we suggest that p-@xmath0 electrons in silicene , because of mott localization , responds strongly to substrate perturbations by making use of the soft sp@xmath9 hybridization option and corresponding c - axis displacements .
@xmath36 : electron doping in silicene deposited on znb@xmath3 by alkali metal leads to appearance of small fermi arc or pockets like features at the m points @xcite .
this is not easily explained within lda band structure result .
a doped mott insulator on the other hand , does not obey luttinger theorem in the usual fashion and can have unusual emergent fermi pockets and fermi arcs - a striking example is the under doped cuprates .
thus we have some theoretical arguments and a few phenomenological facts in support of a mott insulator picture for neutral silicene . in the following
we will build a model , address the above experimental observations from the mott insulator point of view .
let us first discuss nature of the hypothesized mott insulating state of neutral silicene . from arpes
experiment @xcite we estimate a mott hubbard gap of @xmath1 0.6 ev . in transition metal oxide and organic mott insulators , mott gap is often comparable to band widths .
as the inferred mott gap , 0,6 ev is only a tenth of the total @xmath27 band width @xmath1 6 ev we call silicene as a _ a narrow gap mott insulator_. in a mott insulator low energy degree of freedom are spins and spin - spin interaction arise from super ( kinetic ) exchange processes .
one can estimate j , using the standard expression , j @xmath37 .
the values , t @xmath1 1.14 ev and an effective u @xmath1 5 ev , discussed earlier for silicene , gives us a j @xmath38 1 ev . as j is comparable to the mott hubbard gap of 0.6 ev , a strong virtual charge density and charge current fluctuations will renormalize j to lower values . further , higher order cyclic exchanges will be also present . ignoring spin - orbit coupling for the moment ,
our effective spin hamiltonian in the mott insulating state is a spin - half heisenberg hamiltonian h_s = j _ ij ( * s*_i _ j - ) + 4 & 6 spin terms if the nearest neighbor j dominates we will have long range antiferromagnetic order in the ground state . as recent results show , within nearest neighbor repulsive hubbard model one is unlikely to stabilize a spin liquid phase @xcite . in the present case relevant model
is an extended hubbard model also containing effective non nearest neighbor hopping terms ; this is because of the extra charge and current fluctuations arising from unscreened coulomb interaction and narrow mott hubbard gap .
this model is likely to support a spin liquid phase .
however , as in cuprates , this may not be an important practical issue .
this is because we expect that even a small density of dopants through their dynamics will destroy long range afm order and stabilize some kind of spin liquid state ( pseudo gap phase ) containing incoherent dopant charges .
it is this tiny doping induced spin liquid state that will determine the nature of superconductivity over a range of doping .
it is likely that a variety of spin liquids are around the corner .
for dopant dynamics in a mott insulator , up to certain range of doping , a projective constraint arising from upper and lower hubbard band formation and surviving superexchange are important . thus the relevant model for doped mott insulator is the tj model : h_tj = - t _ ij ( c^_ic^_j + h.c . )
+ j _ ij ( * s*_i _ j - n_i n_j ) with the local constraint @xmath39 2 or 0 , for hole and electron doping respectively .
the value of j , t @xmath1 1 ev .
it is known from the study of single hole dynamics in mott insulating cuprates that a free and coherent propagation of a doped hole or electron ( carrying its charge and spin ) is frustrated by the strongly quantum entangled spin background .
this leads to a significant band narrowing and loss of spectral wight over a large part of the brillouin zone , as seen in arpes experiments @xcite and tj model calculations @xcite . for cuprates
band theory gives a width of 8 t @xmath38 3 to 4 ev , while arpes results in the mott insulating cuprates give a band width of 2j @xmath1 0.3 ev ( essentially spin wave band width ) for holes .
this is a substantial , ten fold reduction of hole band width .
as we saw earlier arpes study on silicene grown on zrb@xmath3 gives a parabolic hole band at k point , about 0.3 ev below the fermi level ( band x@xmath3 in figure 1 b ) with a band width @xmath1 1 ev .
comparing results with cuprates mentioned above , we get a j @xmath1 0.5 ev .
this is in the right ball park , as our estimate of j @xmath1 1 ev .
we start with tj model for our doped mott insulator in a honeycomb lattice and discuss superconductivity .
interestingly , the above model ( equation 2 ) was studied in a mean field approach first in @xcite for graphite .
later the present author independently studied the same model ignoring on site constraints @xcite as a semi microscopic way of incorporating pauling s spin singlet ( resonating valence bond ) correlations @xcite in planar _
graphitic systems _ , within a band theory approach .
it lead to a prediction of very high tc superconductivity around an optimal doping .
our mean field theory result of high tc superconductivity for graphite was reanalyzed and confirmed by annica black - shaefer and doniach @xcite .
they also found a remarkable spin singlet chiral superconducting state , namely a state with d + i d symmetry as the most stable mean field solution for the same range of doping .
a more conservative repulsive hubbard model has been studied from superconductivity point of view using variational monte carlo method in reference @xcite , by us and collaborators for graphene .
there are other important works addressing the same issue@xcite by different methods .
they all support possibility of d + i d chiral spin singlet superconductivity .
a very recent work @xcite studies tj model on a honeycomb lattice using a new and powerful variational approach using grassman tensor network states .
it confirms d + i d superconductivity for tj model on honeycomb lattice for a range of doping .
using available theory and insights we suggest a phase diagram ( figure 1 ) , qualitatively similar to cuprates , including the _ pseudo gap phase_. a new aspect for the honey comb lattice is presence of a van hove singularity and an associated nesting ( figure 1 ) at a doping of x = 0.25 .
it has been suggested that this nesting might stabilize a complex magnetic order @xcite or charge density wave order .
if our variational calculation @xcite performed for graphene , a doped semi - metal described by intermediate repulsive u hubbard model , is any guidance , maximum superconducting tc occurs already around 15 percent , somewhat similar to cuprates .
the issue of scale of superconducting tc is very exciting and it holds some promise . as mentioned earlier we have an unusually large j @xmath1 1 ev and t @xmath1 1 ev for our honeycomb lattice .
this is to be contrasted with a value of j @xmath1 0.15 ev and t @xmath1 0.25 ev for the square lattice cuprates , known for their record tc @xmath1 90@xmath40 k , for for single layer cuprates .
we have an average 4 fold fold increase in t and j for silicene .
does silicene offer a 4 fold increase in tc ?
taking care of lattice structure difference between cuprate and silicene crudely , a scaling gives at least 3 fold increase of tc .
thus room temperature scales for tc seems within reach , provided competing orders are taken care of .
as in cuprates , we expect competing orders , charge and spin stripes to challenge the high tc superconducting states .
further , as we will see , electron lattice interaction can cause patterns of vertical displacements of si atoms , arising from sp@xmath9 mixing , it is conceivable that there will be stripes and 2d patterns , corresponding to various valence bond orders .
these valence bond localization tendencies will compete and reduce tc significantly .
the only available , but remarkable indicator for superconductivity in silicene is an sts study .
they find a tc in the range 30 - 40 k and a large gap of 35 mev ; that is , an anomalous value , @xmath41 .
if we take the maximal gap value from this experiment and extract a mean field tc , using weak coupling bcs result , we get a number close to 200 k !
we interpret this anomaly in @xmath42 , as due to a superconducting state with a large local pairing gap but with tc reduced by a strong phase fluctuations arising from a static or low frequency local competing orders .
this needs to be investigated theoretically and experimentally further .
there are recent suggestions of superconductivity in silicene based on different mechanisms @xcite .
we wish to point out that liu et al .
@xcite have theoretically possibility of spin fluctuation induced chiral d + i d superconductivity in _ neutral _ bilayer silicene . in their mechanism ,
intra layer hopping between the two dirac metallic layers produce small electron and hole pockets around k and k points . in our theory
such a bilayer will remain insulating with an additional feature of a spin gap induced by a strong interlayer exchange coupling ( spin singlet bond formation between si atoms along c - axis ) .
external doping or gate doping by very strong electric fields applied perpendicular to the layers will be needed to create superconductivity in the mott insulating bilayers .
as discussed earlier silicene differs from graphene through presence of a soft sp@xmath9 mixing and mott localized p-@xmath0 electrons .
both these features make p-@xmath0 electrons react strongly to substrate or other perturbations .
the sp@xmath9 mixing is accompanied by puckering of sigma bonds and c - axis displacements .
this is the reason behind a variety of silicene layer reconstructions and patterns of sigma - bond puckerings observed in experiments on silicene .
experimentally @xcite , to a good approximation , there are three types of c - axis displacements of si sites , with reference to an unpuckered flat honeycomb lattice of si - atoms : i ) zero , ii ) + a@xmath43 and iii ) - a@xmath43 .
we represent them as @xmath44 by a _ pseudo spin-1 _ ising variable @xmath45 for a given site i. experimentally the displacement a@xmath46 . in the absence of other perturbations
ideal equilibrium puckered structure has a value @xmath47 for sub lattice a and b respectively , or symmetry related opposite displacements .
however , in stm measurements , at low temperatures one sees modification of the above ideal equilibrium structure , where we have different patterns of c - axis displacements or corresponding pseudo spin patterns .
for example , in the most common @xmath48 reconstructed state sub lattice a has a value @xmath49 , two third of sub lattice b sites have @xmath50 and rest of the b sub lattice sites have a value @xmath51 .
interestingly , in this case , to leading order in c - axis displacement there is no change in si - si bond length . to this extent
the electronic spin state of the mott insulator is not affected . however , in 4 @xmath35 4 reconstructions , a sub lattice of flat hexagons coexist with flat dimers and puckered bonds . in the context of a spin liquid state , 4
@xmath35 4 view can be viewed as a quantum spin liquid containing a valence bond density wave and a hexagonal plaquette resonances density wave . to discuss structural distortions
we introduce a simplified model , a tj model coupled to pseudo spin degree of freedom representing c - axis displacement , arising from sp@xmath9 admixture : h & = & k a_0 ^ 2 _ ij _ i _ j - k_0 a_0 ^ 2 _ i _ i^2 - + & - & _ ij t ( 1 +
_ t a_0 ^ 2 ( _ i - _ j)^2 ) ( c^_ic^_j + h.c . ) + + & + & j _ ij ( 1 + _ j a_0 ^ 2 ( _ i - _ j)^2)(*s*_i _ j - n_i n_j ) in the above model we treat @xmath52 s as independent variables .
this is an approximation .
we have ignored local ( plaquette type ) constraints which maintain overall planarity and connectedness of the structure .
first two terms in equation 3 are lattice energies arising from site dependent @xmath53 hybridization and c - axis displacements a@xmath54 . as an approximation
we assume sharp c - axis displacements a@xmath54 and assume them to be classical as si atoms are heavy .
experimentally and theoretically a@xmath55 0.4 au . the parameters and k@xmath43a@xmath56 are k a@xmath56 are positive @xmath1 100 mev , as we estimate using lda results .
c - axis displacements modulate t and j. we ignore , as a first approximation , ab - plane displacements resulting from c - axis displacements . from symmetry ,
leading change in hopping matrix element is quadratic in relative displacement along c - axis @xmath57 .
our estimate of the parameter @xmath58 5 au .
it is straight forward to perform rvb mean field theory and get solutions corresponding to various distortions seen in experiments .
we will not go into details of the results in the present paper .
the repeated observation of dirac cone feature in silicene and silicene nanoribbon on ag surface might indicate that silicene has undergone a mott insulator to metal transition in view of a large screening coming from the metallic ag substrate .
we will view silicene deposited on ag as a self - doped mott insulator .
earlier we introduced this notion @xcite to describe mott insulator to superconductor transition in certain organic salts , driven by pressures . in the present case
, a strong metallic screening by ag creates an _ electronic pressure _ on silicene causes metalization . in what follows
we discuss self - doped mott insulator briefly .
mott insulator to superconductor transitions in real systems generically occur as a strong first order phase transition , as a function of physical or chemical pressure .
as mentioned earlier , first order transition ows its origin to the often ignored long range coulomb interaction and madelung energy considerations , not contained in simple hubbard model .
interestingly , the metallic state that exists close to first order transition point is best viewed as a mott insulator that has
_ doped itself_. that is , it has spontaneously generated _ a small and equal density of holons and doublons _ to gain delocalization and madelung energy .
density of self doped carriers are generically small , so that the upper hubbard band survives in the optical conductivity , @xmath59 , for example . further superexchange also survives .
metalicity appears as a small drude peak , in the background of a mott insulator like @xmath59 .
it is a projected state where , in the background of a mott insulator , a small and equal density of holons and doublons delocalize and coexist as a single phase , but _ without annihilating each other_. we developed a 2-species tj model to describe the physics of such a metallic and superconducting state .
when self - doping is increased , for example by pressure , the system can go into a over doped and less correlated fermi liquid like state and loose superconductivity .
what makes self - doped mott insulators interesting is that they behave like an externally doped mott insulator exhibiting all rich physics such as high tc superconductivity , pseudo gap phase etc .
we can not exclude the possibility that silicene , grown on metallic substrate is a self - doped mott insulator .
this should be kept in mind in future discussions .
germanium lies below silicon in the periodic table .
carbon , silicon and germanium have atomic radii 0.77 , 1.17 and 1.22 @xmath2 respectively .
correspondingly the 4p-@xmath0 bond is weakened further .
ab - initio calculations for germanene leads to a few percent enhancement of ge - ge bond length , but a substantial increase of puckering and c - axis stretching from @xmath1 0.4 @xmath2 in silicene to @xmath1 0.6 @xmath2 in germanene .
there is a small reduction in @xmath0 and @xmath17 band widths from 3 to 2.5 ev .
all things being quantitatively nearly equal at the level of model , germanene should be a narrow band mott insulator .
consequences we have discussed so for , including possibility of quantum spin liquid and high tc superconductivity on doping should be anticipated . on the experimental front , only fully hydrated geh , germanane has been synthesized and exfoliated .
germanene has not been synthesized so far .
following the example of silicene , it is important to look for suitable metallic substrates and try synthesize a stable germanene .
to summarize , we have hypothesized that silicene and germanene are narrow gap mott insulators , unlike the wide belief that they are dirac metals , like graphene . in making our proposal
we have relied heavily on a synthesis of existing theoretical work and phenomenology . we have used arpes and stm results .
mott s arguments , quantum chemical insights and extensive theoretical study of graphene and silicene has helped us in our proposal . using known results we discussed various solutions for our model , including spectral function of a hole in a mott insulator , absence of landau level splitting , superconductivity and so on .
unconventional superconducting order namely d + i d and high tc s was also discussed .
we also introduced a model to discuss structural reconstructions , that is also ubiquitous in silicene and nearly absent in graphene .
an important support for mott localization comes from our translation of the quantum chemical results of sheka to solid state terminology .
sheka studies nano silicene clusters and finds an extensive radicalization .
radicalization is extracted using a procedure that uses spin polarized ( antiferromagnetic ) ab - initio solutions .
this interesting quantum chemical approximation is less known in solid state context .
as indicated earlier we equate _ radicalization in finite systems to mott localization in extended solid state context_. thus sheka s finding of extensive radicalization and alleged unstable silicene is a conservative evidence for narrow gap mott insulator formation and strongly exchange coupled spins .
we have ignored spin - orbit coupling in the present paper , simply to focus on the key physics of strong correlations .
spin orbit coupling will play its own , important unique role , some what different from standard band insulator or semi metal . within the context of semimetallic graphene interesting consequences of spin orbit coupling
are being studied @xcite .
direct and indirect methods should be used to unravel an underlying mott insulator .
optical conductivity , @xmath59 measurement is an urgent one .
it should focus on finding signatures on hubbard band features
. it will be interesting to confirm the existing claim of superconducting gap in sts measurements .
reconstructions should be studied carefully to distinguish valence bond density wave and plaquette resonance density waves .
pseudo gap physics needs to be explored using nmr and nqr measurement
. our tantalizing prediction of very high tc superconductivity reaching room temperature scales needs to be explored .
this needs ways of understanding and overcoming unavoidable competing instabilities such as spin and charge stripes .
next major experimental and theoretical challenge is to see whether a free standing silicene exists and if it can be synthesized and studied .
can one stabilize germanene in the lab at least on a substrate ?
i am not aware of any real material for which tj model parameters are as large as we have suggested for silicene and germanene , t and j @xmath1 1 ev .
so we consider silicene and germanene as forefront materials in the race for room temperature superconductivity . * acknowledgement * i thank prof .
yamada - takamura for giving permission to reproduce a figure from @xcite and dr ayan datta for a discussion .
i thank dae ( india ) for a rajaramanna fellowship .
this research was supported by the perimeter institute for theoretical physics .
k. s. novoselov , a. k. geim , s. v. morozov , d. jiang , m. i. katsnelson , i. v. grigorieva , s. v. dubonos and a. a. firsov , two - dimensional gas of massless dirac fermions in graphene , nature ( london ) , * 438 * , 197 ( 2005 ) p. vogt , paola de padova , c. quaresima , j. avila , e. frantzeskakis , maria carmen asensio , a. resta , b. ealet , and guy le lay , silicene : compelling experimental evidence for graphenelike two - dimensional silicon , phys .
* 108 * , 155501 ( 2012 ) a. grassi , g. m. lombardo , r. pucci , g. g. n. angilella , f. bartha , and n. h. march , stretched chemical bonds in si6 h6 : a transition from ring currents to localized -electrons ?
chemical physics , * 297 * , 1319 ( 2004 ) n. h. march and a. rubio , structural and electronic properties of low - dimensional c - nanoassemblies and possible analogues for si ( and ge ) , journal of nanomaterials , article i d 932350 , doi:10.1155/2011/932350 ( 2011 ) z majzik , m. r. tchalala , m svec , p hapala , h enriquez , a kara , a j mayne , g dujardin , p jel nek and h oughaddou , combined afm and stm measurements of a silicene sheet grown on the ag(111 ) surface , j. phys .
condens . matter .
, * 25 * 225301 ( 2013 ) b. o. wells , z. -x .
shen , a. matsuura , d. m. king , m. a. kastner , m. greven , and r. j. birgeneau , e versus k relations and many body effects in the model insulating copper oxide sr2cuo2cl2 , phys .
* 74 * , 964967 ( 1995 ) t. c. choy and b. a. mckinnon , significance of nonorthogonality in tight - binding models .
ii . the possibility of high - tc superconductivity in intercalation compounds , phys . rev .
* b 52 * , 14539 - 14543 ( 1995 ) w. s. wang , y.y .
xiang , q.h .
wang , f. wang , f. yang , and d.h .
lee , functional renormalization group and variational monte carlo studies of the electronic instabilities in graphene near 1/4 doping , phys . rev . * 85 * , 035414 ( 2012 ) | combining theoretical considerations and certain striking phenomenology we hypothesize silicene and germanene to be _ narrow gap mott insulators _ and abode of quantum spin liquids and unconventional superconductivity . a weaker @xmath0-bond , because of a 60 % bond stretching , from @xmath1 1.4 @xmath2 in graphene to @xmath1 2.3 @xmath2 in silicene renders mott localization .
a supporting theoretical analysis is presented .
recent experimental results are invoked to provide additional support for our mott insulator modeling : i ) a narrow band of width @xmath1 1 ev lying below k points in arpes of silicene on zrb@xmath3 , ii ) a superconducting gap seen below 35 k with a large @xmath4 20 in silicene on ag , ii ) emergence of electron like pockets at m points on electron doping by na adsorbents and iii ) absence of landau level splitting up to 7 tesla and v ) superstructures , not common in graphene but ubiquitous in silicene . a synthesis of the above results using theory of mott insulator , with and without doping , is attempted .
we surmise that if competing orders are taken care of and optimal doping achieved , superconductivity in silicene and germanene could reach room temperature scales , as our estimates of tj model parameters , t and j @xmath1 1 ev , are encouragingly high . 2 cuo@xmath5 4@xmath6 2o@xmath7 3@xmath8 |
during the past decade , exact renormalization group equations @xcite , in particular in the context of the effective average action @xcite , have become a powerful tool for the investigation of nonperturbative phenomena in both quantum field theory and in statistical physics .
those renormalization group ( rg ) , or flow equations may be regarded the counterpart for the continuum of wilson s renormalization group of iterated kadanoff block spin transformations which had been formulated for discrete spin systems originally @xcite . in both cases
the central idea is to `` integrate out '' all fluctuations with momenta larger than some cutoff @xmath0 , and to take account of them by means of a modified dynamics for the remaining fluctuation modes with momenta smaller than @xmath0 .
this modified dynamics is governed by a scale dependent effective hamiltonian or effective action , @xmath1 , whose @xmath0-dependence is described by a functional differential equation , the exact rg equation . in quantum field theory
this general strategy can be applied to both `` effective '' and `` fundamental '' theories . by definition ,
an effective theory is valid only if all relevant momenta of the process under consideration are close to some specific scale @xmath0 which characterizes the theory . if @xmath1 is the action of an effective theory at scale @xmath0 we can compute cross sections for the scattering of particles with momenta ( or relevant momentum transfers ) of the order of @xmath0 , with all quantum effects included , by simply evaluating the _ tree _ diagrams of @xmath1 .
exact rg equations can be used in order to evolve @xmath1 to a smaller scale @xmath2 by further `` coarse graining '' .
flow equations may also be used for a complete quantization of fundamental theories . if the latter has the classical action @xmath3 one imposes the initial condition @xmath4 at the ultraviolet ( uv ) cutoff scale @xmath5 , uses the exact rg equation to compute @xmath1 for all @xmath6 , and then sends @xmath7 and @xmath8 . loosely speaking
, the defining property of a fundamental theory is that the `` continuum limit '' @xmath8 actually exists after the `` renormalization '' - in the traditional sense of the word - of finitely many parameters in the action ; only a finite number of generalized couplings in @xmath9 is undetermined and has to be taken from the experiment .
this is the case for perturbatively renormalizable theories @xcite , but there are also perturbatively nonrenormalizable theories which admit a limit @xmath10 .
the `` continuum '' limit of those nonperturbatively renormalizable theories is taken at a non - gaussian fixed point of the rg flow .
it replaces the gaussian fixed point which , at least implicitly , underlies the construction of perturbatively renormalizable theories @xcite . thus knowing its fixed point structure is crucial if one wants to assess whether a given model qualifies as a _ fundamental _ theory .
in this paper we shall use a formulation where @xmath1 is the `` effective average action '' @xcite .
it is a coarse grained free energy functional which is constructed in close analogy with the standard effective action @xmath11 to which it reduces in the limit of a vanishing infrared ( ir ) cutoff , @xmath7 .
the euclidean functional integral for the generating functional @xmath12 is modified by adding an ir cutoff term @xmath13 to the classical action .
it supplies a momentum dependent @xmath14-term @xmath15 for a mode of the quantum field with momentum @xmath16 .
the cutoff function @xmath15 vanishes for @xmath17 ; hence the high - momentum modes get integrated out in the usual way .
for @xmath18 it behaves as @xmath19 so that the small - momentum modes get suppressed in the path integral by a mass term @xmath20 @xcite . the scale dependent action @xmath1 is closely related to the legendre transform of the modified generating functional @xmath21 . when regarded as a function of @xmath0
, @xmath1 runs along a rg trajectory in the space of all actions which starts at @xmath4 and ends at @xmath22 . in the simplest case , the exact rg equation which describes this trajectory has the following structure : @xmath23\;.\end{aligned}\ ] ] here @xmath24 denotes the infinite dimensional matrix of second functional derivatives of @xmath1 with respect to all dynamical fields .
this construction is fairly straightforward for matter fields , the inclusion of gauge fields introduces additional complications though . using background gauge techniques , a solution to this problem was given in refs .
@xcite for yang - mills theory and in refs .
@xcite for gravity . leaving the faddeev - popov ghosts aside ,
the effective average action for gravity , @xmath25 $ ] , depends not only on the `` ordinary '' dynamical metric @xmath26 but also on a background metric @xmath27 .
the conventional effective action @xmath28 $ ] is obtained as the @xmath7 limit of the functional @xmath29\equiv\gamma_k[g,\bar{g}=g]$ ] with the two metrics identified @xcite .
the motivation for this construction is that in this manner @xmath29 $ ] becomes invariant under general coordinate transformations .
nonperturbative solutions to the above rg equation which do not require a small expansion parameter can be obtained by the method of `` truncations '' .
this means that one projects the rg flow from the infinite dimensional space of all actions onto some finite dimensional subspace which is particularly relevant for the problem at hand . in this manner
the functional rg equation becomes an ordinary differential equation for a finite set of generalized couplings which serve as coordinates on this subspace . in ref .
@xcite the rg flow of quantum general relativity was projected on the 2-dimensional subspace spanned by the invariants @xmath30 and @xmath31 .
this so - called einstein - hilbert truncation amounts to considering only functionals of the form @xmath32=\left(16\pi
g_k\right)^{-1}\int d^dx\,\sqrt{g}\left\ { -r(g)+2\bar{\lambda}_k\right\}+\mbox{classical gauge fixing}.\end{aligned}\ ] ] here @xmath33 and @xmath34 are the running newton constant and cosmological constant , respectively .
more general and , therefore , more precise truncations would include higher powers of the curvature tensor as well as nonlocal terms @xcite which are not present classically .
quantum gravity is certainly a particularly interesting topic where exact rg equations can lead to important new insights . as
quantized einstein gravity is perturbatively nonrenormalizable a natural option is to consider it an effective field theory @xcite .
already within this setting quantum effects can be studied in a consistent and predictive way .
in fact , in refs .
@xcite the running couplings @xmath33 and @xmath34 obtained in @xcite were used to investigate how quantum gravity effects modify the structure of black holes , and in @xcite the implications for the cosmology of the planck era in the very early universe were studied . along a different line of research
it has been proposed @xcite that there are strong quantum gravitational effects also in the later stages of the cosmological evolution which even might drive the cosmological constant to zero dynamically ; the effective average action would be an ideal tool for exploring such infrared effects .
an even more intriguing possibility is that , despite its perturbative nonrenormalizability , quantized gravity exists _ nonperturbatively _ as a fundamental theory .
it would then be mathematically consistent down to arbitrarily small length scales .
a proposal along these lines is weinberg s `` asymptotic safety '' scenario @xcite .
it assumes that there exists a non - gaussian rg fixed point at which the @xmath8 limit can be taken , i.e. that the theory is `` nonperturbatively renormalizable '' in wilson s sense .
asymptotic safety requires that the non - gaussian fixed point is uv attractive ( i.e. attractive for @xmath35 ) for finitely many parameters in the action , i.e. that its uv critical hypersurface is finite dimensional .
this means that the rg trajectories along which the theory can flow as we send the cutoff @xmath5 to infinity are labeled by only finitely many parameters .
therefore the theory is as predictive as any conventionally renormalizable theory ; it is not plagued by the notorious increase of free parameters which is typical of effective theories .
the set of generalized couplings for which the non - gaussian fixed point is uv attractive should include the dimensionless newton constant , @xmath36 , and cosmological constant , @xmath37 . using the @xmath38-expansion , weinberg showed already long ago that gravity in @xmath39 dimensions @xmath40 is indeed asymptotically safe @xcite .
further progress in this direction , in particular for @xmath41 , was hampered by the lack of an efficient calculational scheme which could be used to search for nonperturbative fixed points .
as a solution to this problem which does not rely on the @xmath38-expansion we propose to use the effective average action in order to find nontrivial fixed points @xmath42 of the gravitational rg flow .
( the dots stand for the infinitely many other couplings which parametrize a generic action functional . ) using this approach , the case @xmath43 was reanalyzed in a more general setting and , more importantly , it was shown that the einstein - hilbert truncation predicts the existence of a non - gaussian fixed point @xmath44 also in dimensions @xmath45 , in particular for @xmath41 @xcite . the crucial question which arises
is whether this result is an artifact of the truncation used , or if it correctly reflects a property of the full theory .
it is clear that in order to answer this question one would like to include further invariants into the truncation and to check whether the predictions stabilize . from the technical point of view such calculations
are extremely complicated so that in the present paper we shall use a different method in order to get a first idea about the reliability of the nontrivial fixed point .
we are going to analyze to what extent its location in the @xmath37-@xmath36 plane and its attractivity properties ( critical exponents ) are scheme dependent . here
`` scheme dependence '' refers to the dependence on the cutoff operator @xmath46 used in the derivation of the rg equation .
first of all , @xmath46 is a matrix in the space of irreducible component fields ( see below ) which is not uniquely determined by the general principles .
hence we can vary it to some extent .
in fact , in the present paper we shall introduce a new cutoff ( `` cutoff of type b '' ) whose matrix structure is different from the original one of ref .
@xcite ( `` cutoff of type a '' ) .
either of these cutoffs is proportional to a `` shape function '' @xmath47 which describes the `` thinning out '' of degrees of freedom as we pass the threshold @xmath48 . also this function can be varied in order to assess the scheme dependence of the fixed point properties .
while in general only the critical exponents but not the location @xmath42 of the fixed point are expected to be universal , i.e. scheme independent @xcite , we shall argue that the product @xmath49 is an observable quantity as well . for observables
the @xmath46-dependence is a pure truncation artifact ; in an exact treatment all @xmath46-dependencies cancel
. the status of the einstein - hilbert truncation would be rather questionable if the fixed point was present for some cutoffs but absent for others .
instead , we find that it is actually there for all admissible cutoffs , and moreover that the observable @xmath49 is scheme independent with a quite unexpected precision .
our results strongly support the conjecture that the non - gaussian fixed point is present in the exact theory and is not a truncation artifact .
also another prerequisite of asymptotic safety turns out to be satisfied : we find that , for any cutoff , the fixed point is uv attractive in both directions of the @xmath37-@xmath36 plane .
ultimately one would like to use more general truncations than ( [ in2 ] ) in order to study the rg flow in a larger subspace .
typically this requires computations whose algebraic complexity is quite formidable .
assume we make an ansatz @xmath50 containing @xmath51 diffeomorphism invariant functionals @xmath52 $ ] . in order to project
the rg flow on the @xmath51-dimensional space with coordinates @xmath53 we must insert the ansatz into the rhs of the flow equation ( [ in1 ] ) . at this point the nontrivial problem , both conceptually and computationally , is to expand the trace with respect to a complete set of actions , @xmath54\}$ ] , in such a way that the @xmath55 s retained in the ansatz are a subset of the @xmath56 s .
the coefficients of the remaining @xmath56 s , those not present in @xmath1 , are set to zero by the truncation . in practice
the projection on the @xmath57-subspace is done by inserting a set of metrics @xmath58 on both sides of eq .
@xmath59([in1 ] ) which give a nonzero value only to specific linear combinations of the @xmath55 s .
provided one manages to compute the functional trace for sufficiently many @xmath60-pairs one can then deduce the ordinary differential equations for the generalized couplings @xmath61 . for the einstein - hilbert truncation
this procedure is fairly simple since ( ignoring the running of the gauge parameter ) it is sufficient to insert for @xmath62 the metric of a family of spheres @xmath63 parametrized by their radius @xmath64 .
their maximal symmetry facilitates the calculations considerably . with @xmath64
kept as a free parameter , these metrics are general enough to disentangle @xmath65 and @xmath66 .
but already when we include invariants with four derivatives of the metric this method fails : the spheres can not distinguish @xmath67 from @xmath68 , for instance .
these remarks hint at ( at least ) two major problems which one faces in generalizations of the exact rg approach to gravity .
( i ) the momentum dependent `` mass '' term @xmath69 $ ] depends quadratically on the metric fluctuation @xmath70 , but also , via @xmath71 $ ] on the background @xmath72 . in general
it is a quite nontrivial task to construct a cutoff operator @xmath73 $ ] which has the desired properties mentioned above for a class of background metrics @xmath72 general enough for the projection on the truncation subspace .
( ii ) assume we found an appropriate @xmath73 $ ] .
then there arises the computational problem of evaluating the trace on the rhs of the rg equation for various @xmath26 s and @xmath72 s .
these metrics do not coincide when we allow for an evolution of the gauge fixing sector .
even if we ignore this complication , the hessian @xmath24 under the trace is an extremely complicated nonminimal covariant matrix differential operator constructed from the curvature tensor and covariant derivatives @xmath74 .
a priori , even for maximally symmetric backgrounds , not all derivatives @xmath74 are contracted to form powers of the covariant laplacian @xmath75 , and @xmath24 is not diagonal in the space of fields with a definite helicity therefore .
hence standard heat kernel techniques or perhaps information about the spectrum of @xmath76 are of no help at this point .
in this paper we outline a general strategy for tackling these problems .
it is based upon york s `` tt - decomposition '' @xcite which is available on ( almost ) every spacetime manifold needed for our projection method .
the idea is to decompose the fluctuation @xmath70 into a transverse , traceless tensor @xmath77 , a longitudinal - transverse tensor ( parametrized by a transverse vector @xmath78 ) , a longitudinal - longitudinal tensor ( parametrized by a scalar @xmath79 ) , and a trace part ( parametrized by another scalar @xmath80 ) . in the basis of
the component fields @xmath81 , all @xmath74 s appear in powers of the laplacian only , at least for the class of maximally symmetric backgrounds .
the important point is that this decomposition can be used in order to simplify the structure of @xmath24 on essentially all backgrounds , not just on spheres .
( on @xmath63 the tt - decomposition boils down to the familiar decomposition of @xmath70 with respect to pieces which are irreducible under the isometry group so(@xmath82 ) . in some of the work following the original paper @xcite this decomposition on @xmath63 had been used already @xcite . )
compared to @xmath63 a certain complication arises , however , because the tt - decomposition is nonorthogonal in general .
the tt - decomposition also helps in solving the first problem , the construction of @xmath46 , because @xmath13 has a much simpler structure when expressed in terms of the component fields rather than the original @xmath70 .
this paper is organized as follows . in the first part ( sections [ s2 ] and [ s3 ] ) we describe the construction of a new rg equation where the component fields @xmath83 are used from the outset . along the way we discuss the problems related to a proper identification of @xmath13 .
this part of the paper is meant to supply a set of tools which will become indispensable in future investigations when one includes further invariants into the truncation ( @xmath84-terms @xcite , for instance ) , if one adds matter fields , or if one allows for a running gauge fixing . as a first application
, we revisit the einstein - hilbert truncation in the second part of the paper ( sections [ s4 ] and [ s5 ] ) .
we introduce a new cutoff @xmath46 which is natural in the tt - language , and we use an arbitrary gauge parameter .
this allows for a nontrivial comparison of the resulting rg equations and their fixed point properties with those of @xcite whose cutoff operator @xmath46 has a rather different structure .
we find both a gaussian and a non - gaussian fixed point in the @xmath85-system and we perform a detailed analysis of their properties , in particular of their scheme dependence .
the chances for realizing the asymptotic safety scenario in 4 dimensions will be discussed in detail . in section [ s6 ]
we investigate the implications of the non - gaussian fixed point for the effective graviton propagator at large momenta .
a kind of dimensional reduction from 4 to 2 dimensions takes place in the vicinity of this fixed point .
the asymptotic form of the propagator suggests that when 4-dimensional spacetime is probed by a very high - energetic graviton it appears to be effectively 2-dimensional .
various technical results , needed in the present paper but presented also with an eye towards future applications @xcite , are relegated to a set of appendices . at this point
the reader who is mostly interested in the results rather than their derivation can proceed directly to section [ s5 ] .
[ s2 ] following @xcite we define a scale dependent modification of the euclidean functional integral for the generating functional @xmath86 by using the background gauge fixing technique @xcite . for this purpose
we decompose the integration variable in the functional integral over all metrics , @xmath87 , into a fixed background metric @xmath72 and a fluctuation field @xmath70 , @xmath88 then we replace the integration over @xmath87 by an integration over @xmath70 . with the faddeev - popov ghosts
@xmath89 and @xmath90 the generating functional @xmath86 may be written as @xmath91=\int{\cal d}h_{\mu\nu}\ , { \cal d}c^\mu\,{\cal d}\bar{c}_\mu } \nonumber\\ & & \times\exp\left[-s[\bar{g}+h]-s_{\rm gf}[h;\bar{g}]-s_{\rm gh } [ h , c,\bar{c};\bar{g}]-\delta_k s[h , c,\bar{c};\bar{g}]-s_{\rm source}\right]\;.\end{aligned}\ ] ] the first term in the exponential , @xmath92=s[\bar{g}+h]$ ] , is the classical action which , for the moment , is assumed to be positive definite .
it is invariant under arbitrary general coordinate transformations .
@xmath93 denotes the gauge fixing term @xmath94=\frac{1}{2\alpha}\int d^dx\,\sqrt{\bar{g}}\ , \bar{g}^{\mu\nu}\,f_\mu[\bar{g},h]\,f_\nu[\bar{g},h]\;.\end{aligned}\ ] ] it corresponds to the gauge condition @xmath95=0 $ ] .
linear gauge conditions , @xmath96=\sqrt{2}\kappa\,{\cal f}^{\alpha\beta}_\mu[\bar{g}]\ , h_{\alpha\beta}\;,\end{aligned}\ ] ] are particularly convenient . in the present paper
we use the harmonic gauge for which @xmath97=\delta^\beta_\mu\bar{g}^{\alpha\gamma } \bar{d}_\gamma-\frac{1}{2}\bar{g}^{\alpha\beta}\bar{d}_\mu\;.\end{aligned}\ ] ] here @xmath98 denotes the covariant derivative constructed from the background metric @xmath72 , while we shall write @xmath74 for the covariant derivative involving the quantum metric @xmath87 . in eq .
( [ 03 ] ) we introduced the constant @xmath99 where @xmath100 denotes the bare newton constant .
the faddeev - popov operator associated with the gauge fixing ( [ 03 ] ) with ( [ 04 ] ) takes the form @xmath101^\mu_{\;\;\;\nu}=\bar{g}^{\mu\rho}\bar{g}^{\sigma \lambda}\bar{d}_\lambda\left(\gamma_{\rho\nu}d_\sigma+\gamma_{\sigma\nu } d_\rho\right)-\bar{g}^{\rho\sigma}\bar{g}^{\mu\lambda}\bar{d}_\lambda \gamma_{\sigma\nu}d_\rho\;.\end{aligned}\ ] ] it enters the functional integral ( [ a ] ) via the ghost action @xmath102=-\sqrt{2}\int d^dx\,\sqrt{\bar{g}}\ , \bar{c}_\mu\,{\cal m}[\bar{g}+h,\bar{g}]^\mu_{\;\;\;\nu}\,c^\nu\;.\end{aligned}\ ] ] furthermore , @xmath13 and @xmath103 are the cutoff and the source action , respectively .
@xmath13 provides an appropriate infrared cutoff for the integration variables and @xmath103 introduces sources for the fields @xmath70 , @xmath89 and @xmath90 .
their explicit structure will be discussed later on . for the calculations in the following sections it turns out to be convenient to decompose the gravitational field @xmath70 according to ( see e.g. @xcite ) @xmath104 to obtain this `` tt - decomposition ''
one starts by splitting off the trace part @xmath105 from @xmath70 .
it involves a scalar field @xmath80 .
the remaining symmetric traceless tensor may be decomposed further into a transverse component @xmath106 and a longitudinal component @xmath107 . introducing a transverse vector field @xmath108 and another scalar @xmath79 ,
the longitudinal tensor can be expressed by @xmath109 with @xmath110 and @xmath111 thereby ending up with eq .
( [ f ] ) .
thus the components of @xmath70 introduced by this transverse - traceless ( tt-)decomposition obey the relations @xmath112 this decomposition is valid for complete , closed riemannian @xmath113-spaces ( i.e. compact riemannian manifolds without boundary ) .
as argued in @xcite , its domain of validity can be extended to open , asymptotically flat @xmath113-spaces , certain assumptions concerning the asymptotic behavior of the fields being made . from now on we assume that the gravitational background belongs to one of these classes of spaces . obviously @xmath70 receives no contribution from those @xmath108- and @xmath79-modes which satisfy the killing equation @xmath114 and the scalar equation @xmath115 respectively .
therefore such modes , referred to as unphysical @xmath108- and @xmath79-modes , have to be excluded from the functional integral . considering the conformal killing equation @xmath116 we recognize that the unphysical @xmath79-modes correspond to constants or are related via @xmath117 to proper conformal killing vectors ( pckv s ) , i.e. solutions of eq .
( [ 011 ] ) which are not at the same time ordinary killing vectors ( kv s ) , @xcite . ) . ] by virtue of the decomposition ( [ f ] ) the inner product on the space of symmetric tensor fields may be decomposed according to @xmath118 from eq .
( [ 08 ] ) we see that , for a general background metric , only @xmath106 , @xmath107 and @xmath119 form an orthogonal set , whereas @xmath120 and @xmath121 are not orthogonal in general .
this nonorthogonality manifests itself in the appearance of terms where the components @xmath108 and @xmath79 mix .
but at least for einstein spaces , where @xmath122 with @xmath123 a constant , we find @xmath124 and therefore @xmath125 .
thus @xmath126 represents an orthogonal set of field components in this case . in order to determine the jacobian @xmath127 which appears in the functional integral ( [ a ] ) after performing the transformation of integration variables
@xmath128 we proceed as follows .
we consider a gaussian integral over @xmath70 and reexpress it in terms of the component fields @xcite : @xmath129 & = & j_1\int{\cal d}h^t_{\mu\nu}\,{\cal d}\widehat{\xi}_\mu\,{\cal d } \widehat{\sigma}\,{\cal d}\phi\,\exp\bigg [ -\frac{1}{2}\int d^dx\,\sqrt{\bar{g}}\nonumber\\ & & \times \bigg\{h^t_{\mu\nu}h^{t\mu\nu}+\frac{1}{d}\phi^2+\left(\widehat{\xi}_\mu , \widehat{\sigma}\right ) m^{(\mu,\nu)}\left(\begin{array}{c}\widehat{\xi}_\nu\\\widehat{\sigma } \end{array}\right)\bigg\}\bigg]\;.\end{aligned}\ ] ] here @xmath130 is a hermitian matrix differential operator . since all functional integrals appearing in eq .
( [ i ] ) are gaussian they are easily evaluated .
this leads to the jacobian @xmath131 here @xmath132 represents an infinite constant which may be absorbed into the normalization of the measure @xmath133 .
the notation adopted in eq .
( [ k ] ) has to be interpreted as follows .
a prime at the determinant or the trace of an operator @xmath134 indicates that all unphysical @xmath108- and @xmath79-eigenmodes of @xmath134 , characterized by eqs .
( [ 09 ] ) and ( [ 010 ] ) , are to be excluded from the calculation .
a subscript at determinants or traces describes on which kind of field the operator @xmath134 acts .
we use the subscripts @xmath135 , @xmath136 and @xmath137 for spin-0 fields @xmath79 , transverse spin-1 fields @xmath108 and symmetric transverse traceless spin-2 fields @xmath138 , repectively .
the subscript @xmath139 appearing in eq .
( [ k ] ) refers to a @xmath140-matrix differential operator whose first @xmath113 columns act on traceless spin-1 fields @xmath141 whereas the last column acts on spin-0 fields @xmath79 .
likewise we decompose the ghost and the antighost into their orthogonal components according to @xmath142 where @xmath143 and @xmath144 are the transverse components of @xmath90 and @xmath89 : @xmath145 , @xmath146 . in order to compute the jacobian @xmath147 induced by the change of variables @xmath148 , @xmath149 we write @xmath150}\nonumber\\ & = & j_2\int{\cal d}c^{t\mu}\,{\cal d}\bar{c}^t_\mu\,{\cal d}\widehat{\eta}\ , { \cal d}\widehat{\bar{\eta}}\,\exp \left[-\int d^dx\,\sqrt{\bar{g}}\,\left\{\bar{c}^t_\mu c^{t\mu } + \widehat{\bar{\eta}}(-\bar{d}^2)\widehat{\eta}\right\}\right]\;.\end{aligned}\ ] ] and perform the grassmann functional integrals .
the result is @xmath151^{-1}\;.\end{aligned}\ ] ] in this case the constant @xmath152-mode represents an unphysical mode which has to be excluded .
it will prove convenient to introduce new variables of integration , @xmath153 , @xmath154 , @xmath155 and @xmath156 , by means of the momentum dependent ( nonlocal ) redefinitions @xmath157 here the operator @xmath158 maps vectors onto vectors according to @xmath159 note that the transformations ( [ y ] ) are well defined and invertible since for any ( physical ) eigenmode the operators under the square roots of ( [ y ] ) have strictly positive eigenvalues .
this is due to the fact that these operators arise from the squares of @xmath160 and @xmath161 and from @xmath162 by shifting all covariant derivatives to the right .
thus they can not assume negative eigenvalues .
for example , @xmath163 furthermore , the spectra of the operators in ( [ y ] ) do not even contain zeros , since the potential zero - modes coincide precisely with the aforementioned unphysical modes which have to be excluded .
for instance , @xmath164 is a zero - mode of @xmath165 if and only if @xmath164 is a killing vector . along the lines outlined in the previous subsection
, we now determine the jacobians for the transformation of integration variables .
we obtain @xmath166)}'\left(-\bar{d}^2 -\overline{{\rm ric}}\right)\right]^{-\frac{1}{2}}\nonumber\\ j_4&=&\left[\det\nolimits_{(0)}'\left((\bar{d}^2)^2 + \frac{d}{d-1}\bar{d}_\mu\bar{r}^{\mu\nu}\bar{d}_\nu\right ) \right]^{-\frac{1}{2}}\nonumber\\ j_5&=&\det\nolimits_{(0)}'\left(-\bar{d}^2\right)\end{aligned}\ ] ] for the transformations @xmath167 , @xmath168 and @xmath169 , @xmath170 , respectively .
( the integration measures have been chosen such that no additional infinite constants occur in the jacobians . )
the square brackets appearing in the subscript @xmath171)$ ] at @xmath172 indicate that the operator under consideration acts on spin - one fields which are transverse only for certain background metrics , because the property of transversality is not necessarily transmitted from @xmath78 to @xmath164 .
however , at least for einstein spaces @xmath164 is transverse as well . after carrying out this change of integration variables , @xmath127 , @xmath172 and @xmath173 are the only jacobians appearing in the generating functional @xmath174 since @xmath147 and @xmath175 cancel . by adding an infrared ( ir ) cutoff @xmath13 to the classical action under the path integral ( [ a ] ) we obtain a scale - dependent generating functional @xmath86 .
the term @xmath13 is chosen to depend on the fluctuation fields in such a way that their eigenmodes with respect to @xmath176 which correspond to large eigenvalues @xmath17 are not influenced , whereas contributions from eigenmodes with small eigenvalues @xmath18 are suppressed . in this sense @xmath86
describes an effective theory at the scale @xmath6 . for technical simplicity
we implement the suppression of the low - momentum modes by momentum - dependent `` mass '' terms , i.e. by cutoffs which are quadratic in the fluctuation fields : @xmath177 = \frac{1}{2}\left < h,{\cal r}_k^{\rm grav}\,h\right > + \left<\bar{c},{\cal r}_k^{\rm gh}\,c\right>\;.\end{aligned}\ ] ] here the operators @xmath178 and @xmath179 are constructed from the covariant derivative with respect to the background metric , @xmath98 .
note that @xmath178 and @xmath179 must not depend on the quantum metric @xmath87 but only on the background metric @xmath72 since otherwise the cutoff can not be quadratic . in order to provide the desired behavior
these operators must vanish for @xmath180 ( in particular for @xmath7 ) and must behave as @xmath181 for @xmath182 .
( the meaning of the constant @xmath183 will be explained later . ) as a consequence , all modes with @xmath18 acquire a mass @xmath184 . at this stage of the discussion it is not necessary to specify the explicit structure of the cutoff operators .
we only mention the following point . according to appendix [ dec ] , @xmath178 and @xmath179 can be chosen such that , at the level of the component fields , @xmath185 = \frac{1}{2}\sum\limits_{\zeta_1,\zeta_2\in i_1 } \left<\zeta_1,\left({\cal r}_k\right)_{\zeta_1\zeta_2}\ , \zeta_2\right>+\frac{1}{2 } \sum\limits_{\psi_1,\psi_2\in i_2 } \left<\psi_1,\left({\cal r}_k\right)_{\psi_1\psi_2}\psi_2\right>\end{aligned}\ ] ] with the index sets @xmath186 , @xmath187 .
in contrast to generic cutoffs which are defined in terms of the component fields from the outset , the structure ( [ 013 ] ) allows us to return to the formulation in terms of the fundamental fields , eq .
( [ v ] ) , in a straightforward way .
( see appendix [ dec ] . )
the set of operators @xmath188 , @xmath189 introduced by this realization of the cutoff may be fixed later on .
hermiticity demands that they satisfy @xmath190 and @xmath191 .
furthermore , @xmath192 if both @xmath193 and @xmath194 , or if both @xmath195 and @xmath196 .
a similar decomposition is applied to @xmath103 .
the source terms are defined as @xmath197=-\left < j , h\right > -\left<\bar{k},c\right>-\left < k,\bar{c}\right>\end{aligned}\ ] ] with external sources @xmath198 , @xmath199 and @xmath200 for the fundamental fields @xmath70 , @xmath90 and @xmath89 , respectively . proceeding as described in appendix [ dec ] ,
an alternative form of @xmath103 may be derived from eq .
( [ hh ] ) where each component field is coupled to a certain component of the fundamental sources .
then , in terms of these `` component sources '' , @xmath103 takes the form @xmath201 = -\sum\limits_{\zeta\in i_1}\left < j_\zeta,\zeta\right > -\sum\limits_{\psi\in\{c^t,\eta\}}\left<\bar{k}_{\psi},\psi\right >
-\sum\limits_{\psi\in\{\bar{c}^t,\bar{\eta}\}}\left < k_{\psi},\psi\right>\;.\end{aligned}\ ] ] as a consequence , the functional @xmath202=j_1j_3j_4 \int{\cal d}h^{t}_{\mu\nu}\,{\cal d}\xi_\mu\ , { \cal d}\sigma\,{\cal d}\phi\,{\cal d}c^{t\mu}\,{\cal d}\bar{c}_\mu^t\ , { \cal d}\eta\,{\cal d}\bar{\eta}\,\exp\bigg[-s[h;\bar{g } ] } \nonumber\\ & & -s_{\rm gf}[h;\bar{g}]-s_{\rm gh } [ h , c,\bar{c};\bar{g}]-\delta_k s\left[h , c,\bar{c};\bar{g}\right ] -s_{\rm source}[h , c,\bar{c},j , k,\bar{k};\bar{g } ] \bigg]\end{aligned}\ ] ] as well as the scale - dependent generating functional for the connected green s functions , @xmath203=\ln { \bf z}_k \left[j , k,\bar{k};\bar{g}\right]\;,\end{aligned}\ ] ] may be viewed as functionals of either the fundamental or the component sources .
furthermore , we may derive @xmath0-dependent classical fields for both fundamental and component fields in terms of functional derivatives of @xmath21 .
in either case the @xmath0-dependent classical fields represent expectation values @xmath204 of quantum fields @xmath205 , in the sense that all degrees of freedom corresponding to momenta with @xmath206 have been averaged out .
the classical fundamental fields are given by @xmath207 and the classical component fields are obtained as @xmath208 here we are making use of the shorthand notation @xmath209 for the quantum component fields , @xmath210 for their sources and @xmath211 for the classical component fields .
we may reconstruct the classical fundamental fields from @xmath212 according to @xmath213^{-\frac{1}{2}}\bar{\xi}_\nu + \bar{d}_\nu\left[-\bar{d}^2- \overline{\rm ric}\right]^{-\frac{1}{2}}\bar{\xi}_\mu + \frac{1}{d}\bar{g}_{\mu\nu}\bar{\phi}\nonumber\\ & & + \bar{d}_\mu\bar{d}_\nu\left[(\bar{d}^2)^2+\frac{d-1}{d}\bar{d}_\rho \bar{r}^{\rho\lambda}\bar{d}_\lambda\right]^{-\frac{1}{2}}\bar{\sigma } -\frac{1}{d}\bar{g}_{\mu\nu}\bar{d}^2 \left[(\bar{d}^2)^2+\frac{d-1}{d}\bar{d}_\rho \bar{r}^{\rho\lambda}\bar{d}_\lambda\right]^{-\frac{1}{2}}\bar{\sigma}\ ; , \nonumber\\ \bar{v}_\mu&=&\bar{v}_\mu^t+\bar{d}_\mu\left(-\bar{d}^2\right)^{-\frac{1}{2 } } \bar{\varrho}\;,\;\;\ ; v^\mu = v^{t\mu}+\bar{d}^\mu\left(-\bar{d}^2\right)^{-\frac{1}{2}}\varrho\;.\end{aligned}\ ] ] performing a legendre - transformation on @xmath21 with respect to @xmath214 , @xmath199 and @xmath200 leads to the following scale - dependent modification of the effective action : @xmath215 = \left < j,\bar{h}\right>+\left<\bar{k},v\right>+\left
< k,\bar{v}\right > -w_k\left[j , k,\bar{k};\bar{g}\right]\end{aligned}\ ] ] since eqs .
( [ hh ] ) , ( [ 071 ] ) imply @xmath216 it is clear that eq .
( [ mm ] ) also would result from legendre - transforming @xmath21 with respect to the component sources . denoting the corresponding legendre - transform by @xmath217 we have @xmath218\equiv\widetilde{\gamma}_k[\bar{h } , v,\bar{v};\bar{g}]$ ] where the arguments of @xmath217 and @xmath219 are related by ( [ 07 ] ) .
the effective average action proper , @xmath1 , is defined as the difference between @xmath219 and the cutoff action with the classical fields inserted @xcite : @xmath220 \equiv\widetilde{\gamma}_k\left[g-\bar{g},v,\bar{v};\bar{g}\right ] -\delta_k s[g-\bar{g},v,\bar{v};\bar{g}]\;.\end{aligned}\ ] ] here we expressed @xmath221 in terms of the classical counterpart @xmath26 of the quantum metric @xmath222 which , by definition , is given by @xmath223 the main advantage of the background gauge is that it makes @xmath1 a gauge invariant functional of its agruments @xcite .
it is invariant under general coordinate transformations of the form @xmath224=\gamma_k[\phi+{\cal l}_u\phi]\;\;,\;\;\;\phi\equiv \left(g_{\mu\nu},\bar{g}_{\mu\nu},v^\mu,\bar{v}_\mu\right)\end{aligned}\ ] ] where @xmath225 is the lie derivative with respect to the generating vector field @xmath226 . since general coordinate
invariance ensures that no symmetry violating terms occur in the course of the evolution of @xmath1 the class of consistent truncations is restricted to those which involve only invariant field combinations .
this is important for practical applications of the evolution equation .
we are mainly interested in the exclusively @xmath26-dependent functional @xmath227\equiv\gamma_k[g , g,0,0]\;.\end{aligned}\ ] ] in the limit @xmath7 it coincides with the conventional effective action @xmath228 $ ] , the generator of the 1pi graviton green s functions @xcite : @xmath28= \lim\limits_{k\rightarrow 0}\bar{\gamma}_k[g]$ ] . however , in order to derive an exact evolution equation it is necessary to retain the dependence on the ghost fields and @xmath72 . from the definition of the effective average action it follows that @xmath1 satisfies the integro - differential equation @xmath229\right\ } = \int{\cal d}h_{\mu\nu}\,{\cal d}c^\mu\,{\cal d}\bar{c}_\mu\ ,
\exp\right[-\widetilde{s}[h , c,\bar{c};\bar{g}]}\nonumber\\ & & \left.+\int d^dx\left\{\left(h_{\mu\nu } -g_{\mu\nu}+\bar{g}_{\mu\nu}\right)\frac{\delta\gamma_k}{\delta \bar{h}_{\mu\nu } } + \left(c^\mu - v^\mu\right)\frac{\delta\gamma_k}{\delta v^\mu } + \left(\bar{c}_\mu-\bar{v}_\mu\right)\frac{\delta\gamma_k } { \delta\bar{v}_\mu}\right\}\right]\nonumber\\ & & \times\exp\left\{-\delta_k s\left[h - g+\bar{g},c - v,\bar{c}-\bar{v};\bar{g } \right]\right\}\end{aligned}\ ] ] with @xmath230\equiv s[\bar{g}+h]+s_{\rm gf}[h;\bar{g } ] + s_{\rm gh}[h , c,\bar{c};\bar{g}]\;.\end{aligned}\ ] ] eq .
( [ 022 ] ) may be derived by inserting the definition of @xmath1 into ( [ a ] ) and replacing the sources according to @xmath231 the exact renormalization group equation describes the change of the action functional @xmath1 induced by a change in the scale @xmath0 .
it may be obtained as follows .
differentiating the functional integral ( [ ee ] ) with respect to @xmath232 leads to @xmath233 + \frac{1}{2}{\rm tr}'\left[\sum\limits_{\psi_1,\psi_2\in i_2}\!\ !
\left<\psi_1\otimes\psi_2\right > \partial_t\left({\cal r}_k\right)_{\psi_1\psi_2}\right]\end{aligned}\ ] ] here we used eq .
( [ jj ] ) and adopted the matrix notation on the rhs of ( [ 050 ] ) which , in turn , can be expressed in terms of the hessian @xmath234 } \frac{1}{\sqrt{\bar{g}(x ) \bar{g}(y)}}\frac{\delta^2\widetilde{\gamma}_k}{\delta\varphi_i(x ) \delta\varphi_j(y)}\end{aligned}\ ] ] with @xmath235=0 $ ] for commuting fields @xmath236 and @xmath235=1 $ ] for grassmann fields @xmath236 . since the connected two point function @xmath237 and @xmath238 are inverse matrices in the sense that @xmath239 we may replace the expectation values @xmath240 appearing in eq . ( [ 050 ] ) with @xmath241 . then performing a legendre - transformation according to eq .
( [ mm ] ) and subtracting the cutoff action @xmath242 $ ] yields the desired exact renormalization group equation : @xmath243 & = & \frac{1}{2}{\rm tr}'\left[\sum\limits_{\zeta_1,\zeta_2\in \bar{i}_1 } \left(\gamma_k^{(2)}[g,\bar{g},v,\bar{v } ] + { \cal r}_k\right)^{-1}_{\zeta_1\zeta_2 } \partial_t\left({\cal r}_k\right)_{\zeta_2\zeta_1}\right ] \nonumber\\ & & + \frac{1}{2}{\rm tr}'\left[\sum\limits_{\psi_1,\psi_2\in \bar{i}_2 } \left(\gamma_k^{(2)}[g,\bar{g},v,\bar{v } ] + { \cal r}_k\right)^{-1}_{\psi_1\psi_2 } \partial_t\left({\cal r}_k\right)_{\psi_2\psi_1 } \right]\;.\end{aligned}\ ] ] here we wrote @xmath244 , @xmath245 and introduced the index sets @xmath246 in a position space representation , the operators appearing on the rhs of the flow equation are given by matrix elements whose traces are evaluated according to @xmath247 for instance .
the notation adopted for the matrix elements is similar to eq .
( [ 052 ] ) ; for example , @xmath248 by virtue of the properties of @xmath46 discussed above the traces appearing in the flow equation ( [ oo ] ) are perfectly convergent for all values of @xmath249 . provided we impose the correct initial condition at the uv scale @xmath250 we can , in principle ,
determine the functional integral ( [ a ] ) by integrating the flow equation from @xmath5 down to @xmath0 and letting @xmath7 , @xmath8 after appropriate renormalizations .
the initial condition @xmath251 can be obtained from the integro - differential equation ( [ 022 ] ) . for sufficiently large values of @xmath0 , the cutoff term in eq .
( [ 022 ] ) strongly suppresses fluctuations with @xmath252 so that the main contribution to the functional integral results from small fluctuations about @xmath253 .
this field configuration corresponds to the global minimum of the total action in the exponential of eq .
( [ 022 ] ) . performing a saddle point expansion of the functional integral about this minimum leads to @xmath254=\widetilde{s}[g-\bar{g},v,\bar{v};\bar{g } ] -\frac{1}{2}\ln{\rm det}'\left(\widetilde{s}^{(2)}+{\cal r}_k\right)\end{aligned}\ ] ] where the second term contains one - loop effects .
for @xmath255 they amount to an often unimportant shift in the bare parameters of @xmath256 which can be ignored usually . for finite @xmath5 ,
additional contributions from the determinant occur which are suppressed by inverse powers of @xmath5 @xcite .
therefore we obtain the initial value for @xmath8 @xmath257=s[g]+s_{\rm gf}[g-\bar{g } ; \bar{g}]+s_{\rm gh}[g-\bar{g},v,\bar{v};\bar{g}]\;.\end{aligned}\ ] ] at the level of the functional @xmath258 $ ] this initial condition boils down to @xmath259=s[g]\;.\end{aligned}\ ] ] so far we assumed the fundamental action to be positive definite .
however , the einstein - hilbert action , for instance , does not have this property which is due to the appearance of a `` wrong - sign '' kinetic term associated with the conformal factor . in such cases it is nevertheless possible to formulate a well - defined evolution equation if the signs of the cutoff operators @xmath46 are properly adjusted @xcite .
we will return to this point in the next section . before continuing we
summarize the simplifications that occur for einstein backgrounds , for which @xmath122 with @xmath123 a constant . in this case the decomposition ( [ f ] ) of @xmath70 is completely orthogonal .
in fact , thanks to the einstein condition , the @xmath108-@xmath79 mixing terms in the inner product ( [ 08 ] ) vanish so that @xmath77 , @xmath160 , @xmath161 and @xmath119 form an orthogonal set .
furthermore , for einstein spaces the jacobians appearing in the path integral ( [ ee ] ) cancel , at least up to an ( infinite ) constant which can be absorbed into the normalization of the integration measure .
this can be seen as follows : @xmath260\bigg)^{-1}\nonumber\\ & = & n_1\sqrt{\det\nolimits_{(1t)}'\left(-\bar{d}^2-c\right ) } \sqrt{\det\nolimits_{(0)}'\left(-\bar{d}^2\right)}\sqrt{\det\nolimits_{(0 ) } ' \left(-\bar{d}^2-\frac{d\,c}{d-1}\right)}=n_2\,j_3^{-1}\,j_4^{-1}\end{aligned}\ ] ] here @xmath261 and @xmath262 are unimportant constants so that @xmath263 is indeed field independent . finally , for einstein spaces the field redefinitions in the gravitational sector take the form @xmath264 as a consequence , we find that @xmath265 .
thus , transversality of @xmath78 implies that @xmath164 is transverse as well .
[ s3 ] in practical applications of the exact evolution equation one encounters the problem of dealing with an infinite system of coupled differential equations since the evolution equation describes trajectories in an infinite dimensional space of action functionals . in general
it is impossible to find an exact solution so that we are forced to rely on approximations .
a powerful nonperturbative approximation scheme is the truncation of the parameter space , i.e. only a finite number of couplings is considered . in this manner
the renormalization group flow of @xmath1 is projected onto a finite - dimensional subspace of action functionals . in practice one
makes an ansatz for @xmath1 that comprises only a few couplings and inserts it on both sides of eq . (
[ oo ] ) , thereby obtaining a truncated evolution equation . by projecting the rhs of this equation onto the space of operators appearing on the lhs one arrives at a set of coupled differential equations for the couplings taken into account . as discussed in refs .
@xcite , ward identities provide an important tool for judging the admissability and quantitative reliability of a given truncation ; approximate solutions of the flow equation are not necessarily consistent with the ward identities , in contrast to the exact solution .
therefore , only those truncations which are indeed consistent with the ward identities , at least up to a certain degree of accuracy , will yield reliable results .
the ward identities to be considered here are modified by additional terms coming from the cutoff which are not present in the ordinary identities . since @xmath13 vanishes as @xmath7
the ordinary ward identities are recovered in this limit . in ref .
@xcite the modified ward identities were derived for the gravitational effective average action @xmath266 $ ] where @xmath267 and @xmath268 are auxiliary sources for the brs variations of the graviton and the ghosts which are needed in order to formulate the ward identities .
setting @xmath269 in the argument of this more general functional we obtain the action @xmath270 $ ] discussed in the present paper .
( it would be straightforward to include the @xmath267,@xmath268-sources also in the new formulation of the flow equation , but we will not need them in the following . ) in @xcite the ward identities were used to test the consistency of truncations of the form @xmath271=\bar{\gamma}_k[g]+\widehat{\gamma}_k[g,\bar{g } ] + s_{\rm gf}[g-\bar{g};\bar{g}]+s_{\rm gh}[g-\bar{g},v,\bar{v};\bar{g}]\end{aligned}\ ] ] with @xmath272 $ ] defined as in eq .
( [ 024 ] ) .
the term @xmath273 $ ] encodes the quantum corrections to the gauge fixing term .
this interpretation of @xmath273 $ ] is obvious because for @xmath274 the purely gravitational part of eq .
( [ 030 ] ) implies @xmath275-\gamma_k[g , g,0,0 ] = \widehat{\gamma}_k[g,\bar{g}]+s_{\rm gf}[g-\bar{g};\bar{g}]$ ] . by definition ,
@xmath276=0 $ ] . in the ansatz ( [ 030 ] )
the ghost dependence has been extracted in terms of the classical @xmath277 , thereby neglecting the evolution of the ghost action .
this guarantees that the initial condition ( [ 023 ] ) is satisfied automatically in the ghost sector . in the gravitational sector
it requires @xmath278 , @xmath279 . for truncations of the type ( [ 030 ] )
the ward identities demand that @xmath272 $ ] is a gauge invariant functional of @xmath26 and they yield a constraint equation for @xmath273 $ ] .
to lowest order , this equation is solved by @xmath280 @xmath281 . in the einstein - hilbert truncation
we go beyond this approximation and set @xmath282 with a constant of proportionality which vanishes at @xmath250 ; it takes the running of the graviton s wave function normalization into account ( see below ) . inserting the ansatz ( [ 030 ] ) into the exact evolution equation ( [ oo ] ) leads to a truncated renormalization group equation which describes the evolution of @xmath1 in the subspace of action functionals spanned by ( [ 030 ] ) . the equation governing the evolution of the purely gravitational action @xmath283\equiv\gamma_k[g,\bar{g},0,0]=\bar{\gamma}_k[g]+s_{\rm gf } [ g-\bar{g};\bar{g}]+\widehat{\gamma}_k[g,\bar{g}]\end{aligned}\ ] ] reads @xmath284 & = & \frac{1}{2}{\rm tr}'\left[\sum\limits_{\zeta_1,\zeta_2\in\bar{i}_1 } \left(\gamma_k^{(2)}[g,\bar{g } ] + { \cal r}_k\right)^{-1}_{\zeta_1\zeta_2 } \partial_t\left({\cal r}_k\right)_{\zeta_2\zeta_1}\right]\nonumber\\ & & + \frac{1}{2}{\rm tr}'\left[\sum\limits_{\psi_1,\psi_2\in\bar{i}_2 } \left(s_{\rm gh}^{(2)}[g,\bar{g } ] + { \cal r}_k\right)^{-1}_{\psi_1\psi_2 } \partial_t\left({\cal r}_k\right)_{\psi_2\psi_1 } \right]\;.\end{aligned}\ ] ] here @xmath285 and @xmath286 are the hessians of @xmath287 $ ] and @xmath288 $ ] with respect to the gravitational and the ghost component fields , respectively .
they are taken at fixed @xmath72 . in order to obtain a tractable evolution equation for a given truncation
it is convenient to use a cutoff which is adapted to this truncation but still has the general suppression properties described in subsection [ 2d ] .
it is desirable to start from a definition of @xmath289 that brings about the correct suppression of low - momentum modes for a class of truncations and of gravitational backgrounds which is as large as possible . a convenient , adapted cutoff can be found by the following rule @xcite . given a truncation
, we assume that for @xmath290 the kinetic operators of all modes with a definite helicity are of the form @xmath291 where @xmath292 is a set of c - number functions and the indices @xmath293 refer to the different types of fields .
( the difficulty of bringing @xmath24 to this form is one of the main reasons for using the tt - decomposition .
at least for maximally symmetric spaces it allows us to eliminate all covariant derivatives which do not appear as a laplacian @xmath294 . )
then we choose the cutoff in such a way that the structure @xmath295 is achieved . here
the function @xmath296 , @xmath297 , describes the details of the mode suppression ; it is required to satisfy the boundary conditions @xmath298 and @xmath299 , but is arbitrary otherwise . by virtue of eq .
( [ 090 ] ) , the inverse propagator of a field mode with covariant momentum square @xmath300 is given by @xmath301 which equals @xmath302 for @xmath17 and @xmath303 for @xmath18 .
this means that the small-@xmath302 modes , and only those , have acquired a mass @xmath184 which leads to the desired suppression . in the next section
we shall see in detail that for the truncations used in the present paper we can comply with the above rule by using the following cutoff operator : @xmath304\;,\nonumber\\ \left({\cal r}_k\right)_{\bar{\phi}\bar{\phi}}&=&{\cal z}_k^{\bar{\phi } \bar{\phi}}\kappa^2\,k^2 r^{(0)}\left(-\bar{d}^2/k^2\right)\;,\nonumber\\ \left({\cal r}_k\right)^{\mu\nu}_{\bar{v}^tv^t}&=&-\left({\cal r}_k \right)^{\mu\nu}_{v^t\bar{v}^t}={\cal z}_k^{\bar{v}^t v^t}\ , \bar{g}^{\mu\nu}\,k^2 r^{(0)}\left(-\bar{d}^2/k^2\right)\;,\nonumber\\ \left({\cal r}_k\right)_{\bar{\varrho}\varrho } & = & -\left({\cal r}_k\right)_{\varrho\bar{\varrho } } = { \cal z}_k^{\bar{\varrho}\varrho}\,k^2 r^{(0)}\left(-\bar{d}^2/k^2\right)\;.\end{aligned}\ ] ] here @xmath305 is defined as @xmath306 the remaining cutoff operators not listed in eq .
( [ uu ] ) are set to zero .
the @xmath183 s are constants which , again by using ( [ 090 ] ) , will be fixed in terms of the generalized couplings appearing in the ansatz for @xmath1 . the cutoff ( [ uu ] ) is inspired by the @xmath46 used in @xcite for @xmath63 . if eq .
( [ 090 ] ) allows us to choose @xmath307 for all @xmath308 and @xmath309 one obtains a positive definite @xmath13 in the gravitational sector . in this case @xmath310 is a damped exponential which indeed suppresses the contributions from the low - momentum modes . in the following sections we shall focus on the einstein - hilbert truncation for @xmath1 which suffers from the conformal factor problem :
its kinetic term for @xmath311 is negative definite . as a consequence , eq .
( [ 090 ] ) forces us to work with a @xmath312 .
hence , in the @xmath80-sector , @xmath13 is negative definite and , at least at a naive level , @xmath310 seems to enhance rather than suppress the low - momentum modes . as we discussed in detail in ref .
@xcite we nevertheless believe that the rule ( [ 090 ] ) , i.e. allowing for @xmath313 , is correct also in this case .
we emphasize that the rhs of the flow equation , contrary to the euclidean path integral , is perfectly well - defined even if @xmath3 and @xmath1 are not positive definite .
at this point it should be mentioned that the situation with respect to the positivity of the action improves considerably by including higher - derivative terms in @xmath3 and the truncated @xmath1 since these actions are bounded below , provided we choose the correct sign in front of these higher - derivative terms .
furthermore , their quadratic forms are positive definite at least for sufficiently large momenta , and so is the cutoff . for a study of the evolution equation for @xmath84-gravity
we refer to @xcite .
as compared to the original paper @xcite , the cutoff ( [ uu ] ) has a rather different structure which is due to the fact that it is formulated in terms of the component fields arising form the tt - decomposition .
contrary to the original one of ref .
@xcite , the new cutoff ( [ uu ] ) is defined for _ all _ values of @xmath314 .
this is one of the main advantages of the new approach .
note that in refs .
@xcite where the tt - decomposition was used on @xmath63 the actual construction of the effective average action and its rg equation was omitted and has been replaced by an ad hoc modification of the standard one - loop determinants .
no @xmath13 has been specified at the component field level .
hence the scale dependent action constructed in this manner has no reason to respect the general properties of an effective average action @xcite . despite the use of the component fields in @xcite
their cutoff seems to be more similar to the original one in @xcite than to the new one of the present paper .
in fact , it represents an @xmath314-dependent generalization of the cutoff in @xcite , in the sense that the latter is recovered from the one of refs .
@xcite by setting @xmath315 .
from now on we will refer to the cutoff used in the original paper @xcite and in @xcite as the cutoff of * type a*. however , one has to keep in mind that the existence of a corresponding @xmath13 is guaranteed only for @xmath315 , i.e. the case considered in @xcite .
furthermore the cutoff ( [ uu ] ) of the present paper will be referred to as the cutoff * type b * ; it is defined for all values of @xmath314 .
each cutoff type contains the shape function @xmath316 .
a particularly suitable choice is the exponential shape function @xmath317^{-1}\;.\end{aligned}\ ] ] in order to check the scheme independence of universal quantities we employ a one - parameter `` deformation '' of ( [ expcut ] ) , the * class of exponential shape functions * , @xmath318^{-1}\;,\end{aligned}\ ] ] with @xmath319 parametrizing the profile of @xmath316 @xcite .
another admissible choice is the following * class of shape functions with compact support * : @xmath320\right ] & \;\;\;b < y<1.5\\0 & \;\;\;y\ge 1.5 \end{array}\right.\;.\end{aligned}\ ] ] here @xmath321 parametrizes the profile of @xmath316 . for our analysis of the flow equation in section [ s5 ] we shall use both cutoff types with both classes of shape functions .
[ s4 ] in this section we use a simple truncation to derive the renormalization group flow of the newton and the cosmological `` constant '' by means of the truncated flow equation ( [ 032 ] ) . in our example
we assume that , at the uv scale @xmath8 , gravity is described by the classical einstein - hilbert action in @xmath113 dimensions , @xmath322=s[g]=\frac{1}{16\pi\bar{g}}\int d^dx\,\sqrt{g } \left\{-r(g)+2\bar{\lambda}\right\}\;.\end{aligned}\ ] ] for the investigation of the evolution of @xmath323 $ ] towards smaller scales @xmath6 we consider a truncated action functional of the following form : @xmath324&=&2\kappa^2 z_{nk}\int d^dx\,\sqrt{g}\left\ { -r(g)+2\bar{\lambda}_k\right\}\nonumber\\ & & + \kappa^2 \frac{z_{nk}}{\alpha}\int d^dx\,\sqrt{\bar{g}}\,\bar{g}^{\mu\nu } \left({\cal f}_\mu^{\alpha\beta } g_{\alpha\beta}\right)\left({\cal f}_\nu^{\rho\sigma}g_{\rho\sigma}\right)\;.\end{aligned}\ ] ] eq .
( [ 37 ] ) is obtained from @xmath325 by replacing @xmath326 so that its form agrees with that of the gravitational sector of the ansatz ( [ 030 ] ) with @xmath327=\kappa^2 \frac{z_{nk}-1}{\alpha}\int d^dx\ , \sqrt{\bar{g}}\,\bar{g}^{\mu\nu}\left({\cal f}_\mu^{\alpha\beta } g_{\alpha\beta}\right)\left({\cal f}_\nu^{\rho\sigma}g_{\rho\sigma}\right)\;.\end{aligned}\ ] ] generally speaking also the gauge fixing parameter @xmath314 should be treated as a running quantity , @xmath328 .
fortunately there is a simple shortcut which avoids an explicit computation of the corresponding @xmath329-function .
in fact , there are general arguments showing that @xmath314 should have a ( ir attractive ) fixed point at @xmath330 .
this means that the initial condition @xmath331 leads to @xmath332 for all @xmath249 .
thus , even using the truncation with a constant @xmath314 , we can take the correct `` flow '' of the gauge fixing term into account simply by setting @xmath333 . in yang - mills theory
the existence of the fixed point @xmath334 has been demonstrated for a truncation containing a covariant gauge fixing @xcite , while for the axial gauge a nonperturbative proof is available @xcite .
the following general argument suggests that this fixed point should exist in any gauge theory , including gravity @xcite . in the ordinary functional integral , the limit
@xmath335 corresponds to a sharp implementation of the gauge fixing condition , i.e. @xmath336 becomes proportional to @xmath337 $ ] .
the domain of the @xmath338-integration consists of those @xmath70 s which satisfy the gauge condition exactly , @xmath339 . adding the ir cutoff at @xmath0 amounts to suppressing some of the @xmath70-modes while retaining the others .
but since all of them satisfy @xmath339 , it is clear that a variation of @xmath0 can not change the domain of the @xmath70-integration .
the delta - functional @xmath337 $ ] continues to be present for any value of @xmath0 if it was there originally .
hence @xmath314 vanishes for arbitrary @xmath0 . the @xmath0-dependent couplings in eq .
( [ 042 ] ) satisfy the initial conditions @xmath340 and @xmath341 which implies @xmath342 . here
the uv scale @xmath5 is taken to be large but finite .
the evolution of @xmath343 and @xmath34 towards smaller scales may now be determined as follows . as a first step the ansatz ( [ 37 ] ) is inserted into both sides of the truncated flow equation ( [ 032 ] ) .
then we may set @xmath62 . as a consequence ,
the gauge fixing term drops out from the lhs which then reads @xmath344=2\kappa^2\int d^dx\,\sqrt{\bar{g } } \left[-\bar{r}(\bar{g})\,\partial_t z_{nk}+2\partial_t\left(z_{nk}\bar{\lambda}_k\right)\right]\;.\end{aligned}\ ] ] performing a derivative expansion on the rhs we may extract those contributions which are proportional to the operators spanning the lhs , i.e. @xmath30 and @xmath31 .
then , comparing the coefficients of these operators yields a system of coupled differential equations for @xmath343 and @xmath34 .
it describes the projection of the renormalization group flow onto the two - dimensional subspace of the space of all action functionals which is spanned by @xmath30 and @xmath31 .
it is important to note that during this calculation we may insert any metric @xmath72 that is general enough to allow for a unique identification of the operators @xmath30 and @xmath31 . in practice
it proves particularly convenient to exploit this freedom by choosing the gravitional background to be maximally symmetric .
such spaces form a special class of einstein spaces and are characterized by @xmath345 with the curvature scalar @xmath346 considered a constant number rather than a functional of the metric . in the following
we restrict our considerations to maximally symmetric spaces with positive curvature scalar @xmath347 , i.e. @xmath113-spheres @xmath63 . for
@xmath113 fixed , @xmath63 is parametrized by the radius @xmath64 of the spheres , which is related to the curvature scalar and the volume in the usual way , @xmath348 before continuing with the evaluation of the rhs of the flow equation we have to comment on the properties of fields defined on spherical backgrounds .
according to appendix [ harm ] , we may expand the quantum and classical component fields in terms of spherical harmonics @xmath349 , @xmath350 , @xmath351 , which form complete sets of orthogonal eigenfunctions with respect to the corresponding covariant laplacians .
the expansions of @xmath77 , @xmath80 , @xmath89 , @xmath90 and their classical counterparts can be inferred directly from eq .
( [ 067 ] ) .
the remaining component fields are expanded according to @xmath352 similar expansions hold for the associated classical fields
. note that in eq .
( [ 092 ] ) the summations do _ not _ start at @xmath353 for vectors and at @xmath354 for scalars as in eq .
( [ 067 ] ) , but at @xmath355 for @xmath153 and @xmath154 , and at @xmath353 for the scalar ghost fields .
the modes omitted here are the kv s @xmath356 , the solutions of the scalar equation ( [ 010 ] ) which are in one - to - one correspondence with the pckv s @xmath357 , and the constants @xmath358 .
as discussed in subsection [ 2b ] , the fundamental fields obtain no contribution from these modes .
therefore they have to be excluded from eq .
( [ 092 ] ) .
this exclusion is also of importance for the momentum dependent field redefinitions ( [ y ] ) because they would not be well - defined otherwise , as can be seen e.g. from @xmath359 eq . ( [ 095 ] ) follows from inverting eq .
( [ y ] ) and then inserting eq .
( [ 092 ] ) .
the eigenvalues corresponding to the modes excluded , i.e. @xmath360 and @xmath361 ( see table 1 in appendix [ harm ] ) , would lead to a vanishing denominator in eq .
( [ 095 ] ) .
similar arguments hold for the other fields in ( [ y ] ) .
we may now split the quantum field @xmath80 into a part @xmath362 spanned by the same set of eigenfunctions as @xmath154 , and a part @xmath363 containing the contributions from the remaining modes : @xmath364 the orthogonality of the spherical harmonics implies @xmath365 so that @xmath366 and in particular @xmath367 . as a consequence , decomposing @xmath80 according to eq .
( [ 112 ] ) ensures that any nonzero term bilinear in the scalar fields is of such a form that the scalars involved are spanned by the same set of eigenfunctions .
the same is true for the corresponding classical fields @xmath368 and @xmath369 .
let us now return to the evaluation of the flow equation . on the rhs we need the operator @xmath370 $ ] . for our purposes
it is sufficient to determine this operator at @xmath62 .
it may be derived by expanding @xmath323 $ ] according to @xmath371=\gamma_k[\bar{g},\bar{g}]+{\cal o}(\bar{h } ) + \gamma_k^{\rm quad}[\bar{h};\bar{g}]+{\cal o}(\bar{h}^3)\end{aligned}\ ] ] and retaining only the part quadratic in @xmath221 , i.e. @xmath372 $ ] . for our truncation
it takes the form @xmath373 = \kappa^2 z_{nk}\int d^dx\,\sqrt{\bar{g}}\,\bar{h}_{\mu\nu}\bigg\ { -\left[\frac{1}{2}\delta^\mu_\rho\delta_\sigma^\nu+\frac{1 - 2\alpha}{4\alpha } \bar{g}^{\mu\nu}\bar{g}_{\rho\sigma}\right]\bar{d}^2}\nonumber\\ & & + \frac{1}{4}\left[2\delta^\mu_\rho\delta^\nu_\sigma-\bar{g}^{\mu\nu } \bar{g}_{\rho\sigma}\right]\left(\bar{r}-2\bar{\lambda}_k\right ) + \bar{g}^{\mu\nu}\bar{r}_{\rho\sigma}-\delta_\sigma^\mu\bar{r}^\nu_{\ ; \rho}-\bar{r}^{\nu\;\;\mu}_{\;\rho\;\;\;\sigma}\nonumber\\ & & + \frac{1-\alpha}{\alpha } \left[\bar{g}^{\mu\nu}\bar{d}_\rho\bar{d}_\sigma-\delta^\mu_\sigma \bar{d}^\nu\bar{d}_\rho\right ] \bigg\}\bar{h}^{\rho\sigma}\end{aligned}\ ] ] where @xmath72 is fixed but still arbitrary . in order to ( partially ) diagonalize this quadratic form
we insert the family of spherical background metrics into eq .
( [ 38 ] ) and decompose @xmath221 according to eq .
. then we use the classical analog of eq .
( [ 112 ] ) to decompose @xmath311 as well .
this leads to @xmath374 = \kappa^2 z_{nk}\int d^dx\,\sqrt{\bar{g}}\,\frac{1}{2}\bigg\ { \bar{h}^t_{\mu\nu}\left[-\bar{d}^2+a_t(d)\bar{r}-2\bar{\lambda}_k \right]\bar{h}^{t\mu\nu}}\nonumber\\ & & + \frac{2}{\alpha}\,\bar{\xi}_\mu\left[-\bar{d}^2+a_v(d,\alpha)\bar{r } -2\alpha\bar{\lambda}_k\right]\bar{\xi}^\mu + c_{s2}(d,\alpha)\bigg ( \bar{\sigma}\left[-\bar{d}^2+a_{s2}(d,\alpha)\bar{r } + b_{s2}(d,\alpha)\bar{\lambda}_k\right]\bar{\sigma}\nonumber\\ & & + 2c_{s3}(d,\alpha)\ , \bar{\phi}_1\sqrt{-\bar{d}^2}\sqrt{-\bar{d}^2-\frac{\bar{r}}{d-1}}\bar{\sigma } \nonumber\\ & & + c_{s1}(d,\alpha)\,\sum\limits_{\bar{\phi}\in\{\bar{\phi}_0,\bar{\phi}_1\ } } \bar{\phi}\left[-\bar{d}^2+a_{s1}(d,\alpha)\bar{r}+b_{s1}(d,\alpha ) \bar{\lambda}_k\right]\bar{\phi}\bigg)\bigg\}\;.\end{aligned}\ ] ] here the @xmath375 s , @xmath376 s , and @xmath123 s are functions of the dimensionality @xmath113 and the gauge parameter @xmath314 .
the explicit expressions for these coefficients can be found in appendix [ coeff ] .
note that this partial diagonalization simplifies further calculations considerably , and this is the main reason for using the decomposition ( [ 07 ] ) and specifying a concrete background .
in contrast to the case @xmath315 considered in @xcite , a complete diagonalization can not be achieved by merely splitting off the trace part from @xmath221 since eq .
( [ 38 ] ) contains additional terms proportional to @xmath377 which introduce mixings between the traceless part and @xmath80 . to be more precise ,
it is the term @xmath378\bar{h}^{\rho\sigma}$ ] that gives rise to such cross terms . in terms of the component fields these cross terms boil down to a purely scalar @xmath379-@xmath311 mixing term that vanishes for the spherical harmonics @xmath380 and @xmath381 .
since these modes contribute to @xmath311 ( but not to @xmath379 ) we can not directly invert the associated matrix differential operator @xmath382 . as a way out
, we split @xmath311 according to eq .
( [ 112 ] ) into @xmath368 and @xmath369 .
this has the effect that only mixings between the scalars @xmath379 and @xmath369 survive , which have the same set of eigenfunctions @xmath351 starting at @xmath355 .
hence the resulting matrix differential operator @xmath383 is invertible , but since this additional split of @xmath80 affects the matrix structure of this operator it leads to a slightly modified flow equation .
in fact , on the rhs of eq .
( [ 032 ] ) the summation in the gravitational sector now runs over the set of fields @xmath384 , with @xmath385 . in the context of the einstein - hilbert truncation
it is only the @xmath314-dependence that introduces mixings of @xmath311 and the traceless part of @xmath221 and therefore necessitates the decompositions ( [ f ] ) , ( [ 112 ] ) .
in general the inclusion of higher derivative terms like @xmath386 and of matter fields leads to similar mixings . in order to determine the contributions from the ghosts appearing on the rhs of eq .
( [ 032 ] ) we set @xmath62 in @xmath277 and assume that @xmath72 corresponds to a spherical background .
then we decompose the ghost fields according to eq .
( [ 07 ] ) which leads to @xmath387=\sqrt{2}\int d^dx\,\sqrt{g}\left\{\bar{v}_\mu^t\left[-d^2-\frac{r}{d}\right ] v^{t\mu}+\bar{\varrho}\left[-d^2 - 2\frac{r}{d}\right]\varrho\right\}\;.\end{aligned}\ ] ] from now on the bars are omitted form the metric , the curvature and the operators @xmath76 and @xmath388 .
note that the decomposition of the ghosts is not really necessary , but it allows for a comparison with the results obtained in @xcite . at this point
we can continue with the adaption of the cutoff to the operators @xmath24 and @xmath389 of eqs .
( [ 44 ] ) , ( [ 48 ] ) .
according to the rule ( [ 090 ] ) the @xmath183 s have to be chosen as @xmath390 thus , for @xmath391 the nonvanishing entries of the matrix differential operators @xmath392 and @xmath393 take the form @xmath394+{\cal r}_k\right)_{\bar{h}^t\bar{h}^t } & = & z_{nk}\kappa^2\left[p_k+a_t(d)r-2\bar{\lambda}_k\right]\;,\nonumber\\ \left(\gamma^{(2)}_k[g , g]+{\cal r}_k\right)_{\bar{\xi}\bar{\xi } } & = & z_{nk}\kappa^2\,\frac{2}{\alpha}\left[p_k+a_v(d,\alpha)r-2\alpha \bar{\lambda}_k\right]\;,\nonumber\\ \left(\gamma^{(2)}_k[g , g]+{\cal r}_k\right)_{\bar{\sigma}\bar{\sigma } } & = & z_{nk}\kappa^2\,c_{s2}(d,\alpha)\left[p_k+a_{s2}(d,\alpha)r + b_{s2}(d,\alpha)\bar{\lambda}_k\right]\;,\nonumber\\ \left(\gamma^{(2)}_k[g , g]+{\cal r}_k\right)_{\bar{\phi}_1\bar{\sigma } } & = & \left(\gamma^{(2)}_k[g , g]+{\cal r}_k\right)_{\bar{\sigma}\bar{\phi}_1 } \nonumber\\ & = & z_{nk}\kappa^2\,c_{s2}(d,\alpha)c_{s3}(d,\alpha)\,\sqrt{p_k } \sqrt{p_k-\frac{r}{d-1}}\;,\nonumber\\ \left(\gamma^{(2)}_k[g , g]+{\cal r}_k\right)_{\bar{\phi}_0\bar{\phi}_0 } & = & \left(\gamma^{(2)}_k[g , g]+{\cal r}_k\right)_{\bar{\phi}_1\bar{\phi}_1 } \nonumber\\ & = & z_{nk}\kappa^2\,c_{s2}(d,\alpha)c_{s1}(d,\alpha)\left[p_k+a_{s1}(d,\alpha ) r+b_{s1}(d,\alpha)\bar{\lambda}_k\right]\;,\nonumber\\ \left(s_{\rm gh}^{(2)}[g , g]+{\cal r}_k\right)_{\bar{v}^tv^t } & = & -\left(s_{\rm gh}^{(2)}[g , g]+{\cal r}_k\right)_{v^t\bar{v}^t } = \sqrt{2}\left[p_k-\frac{r}{d}\right]\;,\nonumber\\ \left(s_{\rm gh}^{(2)}[g , g]+{\cal r}_k\right)_{\bar{\varrho}\varrho } & = & -\left(s_{\rm gh}^{(2)}[g , g]+{\cal r}_k\right)_{\varrho\bar{\varrho } } = \sqrt{2}\left[p_k-2\frac{r}{d}\right]\;.\end{aligned}\ ] ] here we set @xmath395\right)_{\psi_1\psi_2}\equiv \left(s_{\rm gh}^{(2)}[g , g]\right)_{\psi_1\psi_2}$ ] for @xmath396 .
now we are in a position to write down the rhs of the flow equation with @xmath391
. we shall denote it @xmath397 in the following .
in @xmath397 we need the inverse operators @xmath398 and @xmath399 .
the inversion is carried out in appendix [ inv ] .
inserting the inverse operators into @xmath397 leads to @xmath400 + { \rm tr}_{(1t)}'\left[\left(p_k+a_v(d,\alpha)r-2\alpha \bar{\lambda}_k\right)^{-1}{\cal n}\right]\nonumber\\ & & + { \rm tr}_{(0)}''\bigg [ \left(p_k+a_{s3}(d)r-2\bar{\lambda}_k\right)^{-1 } \left(p_k+a_{s4}(d,\alpha)r-2\alpha\bar{\lambda}_k\right)^{-1}\nonumber\\ & & \times\bigg\{\left(f_{s1}(d,\alpha)p_k+a_{s5}(d,\alpha)r-2 ( \alpha+1)\bar{\lambda}_k\right){\cal n}\nonumber\\ & & + f_{s2}(d,\alpha)\sqrt{p_k}\sqrt{p_k-\frac{r}{d-1}}\nonumber\\ & & \times\frac{1}{2z_{nk}}\partial_t\left[z_{nk}\left ( \sqrt{p_k}\sqrt{p_k-\frac{r}{d-1 } } -\sqrt{-d^2}\sqrt{-d^2-\frac{r}{d-1}}\right)\right ] \bigg\}\bigg ] \nonumber\\ & & -2{\rm tr}_{(1t)}\left[\left(p_k-\frac{r}{d}\right)^{-1}{\cal n}_0\right ] -2{\rm tr'}_{(0)}\left[\left(p_k-2\frac{r}{d}\right)^{-1}{\cal n}_0\right ] \nonumber\\ & & + \frac{1}{2z_{nk}}\sum\limits_{l=0}^{1}\left[d_l(d,0)\frac{\partial_t \left[z_{nk } k^2r^{(0)}(\lambda_l(d,0)/k^2)\right]}{\lambda_l(d,0)+k^2r^{(0 ) } ( \lambda_l(d,0)/k^2)+a_{s1}(d,\alpha ) r+b_{s1}(d,\alpha)\bar{\lambda}_k } \right]\;.\end{aligned}\ ] ] here we set @xmath401\nonumber\\ & = & \left[1-\frac{1}{2}\eta_n(k)\right]k^2r^{(0)}(-d^2/k^2 ) + d^2r^{(0)'}(-d^2/k^2)\;,\nonumber\\ { \cal n}_0&=&2^{-1 } \partial_t \left[k^2r^{(0)}(-d^2/k^2)\right ] = k^2r^{(0)}(-d^2/k^2)+d^2r^{(0)'}(-d^2/k^2)\end{aligned}\ ] ] where @xmath402 is the anomalous dimension of the operator @xmath31 and the prime at @xmath316 denotes the derivative with respect to the argument .
furthermore , the new @xmath375 s and @xmath403 s introduced above are again functions of @xmath113 and @xmath314 , tabulated in appendix [ coeff ] . in eq .
( [ 52 ] ) we refined our notation concerning the primes at the traces . from now on one prime indicates the subtraction of the contribution from the lowest eigenvalue , while two primes indicate that the modes corresponding to the lowest two eigenvalues have to be excluded . the next step is to extract the contributions proportional to @xmath30 and @xmath31 by expanding @xmath397 with respect to @xmath404 or @xmath64 , respectively . since @xmath65 , @xmath405 , only terms of order @xmath406 and @xmath407
are needed .
this leads to @xmath408 + { \rm tr}_{(1t)}'\left[\left(p_k-2\alpha\bar{\lambda}_k \right)^{-1}{\cal n}\right]\nonumber\\ & & + { \rm tr}_{(0)}''\left[\left(p_k-2\bar{\lambda}_k\right)^{-1}{\cal
n } \right ] + { \rm tr}_{(0)}''\left[\left(p_k-2\alpha\bar{\lambda}_k \right)^{-1}{\cal n}\right]\nonumber\\ & & -2{\rm tr}_{(1t)}\left[p_k^{-1}{\cal n}_0\right ] -2{\rm tr}_{(0)}'\left[p_k^{-1}{\cal n}_0\right ] -r\bigg\{a_t(d){\rm
tr}_{(2st^2)}\left[\left(p_k-2\bar{\lambda}_k \right)^{-2}{\cal n}\right]\nonumber\\ & & + a_v(d,\alpha){\rm tr}_{(1t)}'\left[\left(p_k-2\alpha \bar{\lambda}_k\right)^{-2}{\cal n}\right ] + a_{s3}(d){\rm tr}_{(0)}''\left[\left(p_k-2\bar{\lambda}_k \right)^{-2}{\cal n}\right ] \nonumber\\ & & + a_{s4}(d,\alpha){\rm tr}_{(0)}''\left[\left(p_k-2\alpha \bar{\lambda}_k\right)^{-2}{\cal n}\right ] + \frac{2}{d}{\rm tr}_{(1t)}\left[p_k^{-2}{\cal n}_0\right]\nonumber\\ & & + \frac{4}{d}{\rm tr}_{(0)}'\left[p_k^{-2}{\cal n}_0\right]-\frac{\delta_{d,2 } } { 4\pi}\int d^2x\,\sqrt{g}\frac{\partial_t\left(z_{nk}k^2\right ) } { z_{nk}\left(k^2 - 2\bar{\lambda}_k\right ) } \bigg\}+{\cal o}(r^ { < d-2})\;.\end{aligned}\ ] ] here @xmath409 means that terms @xmath410 with powers @xmath411 are neglected . the term in eq .
( [ 55 ] ) proportional to @xmath412 arises from the last term in eq .
( [ 52 ] ) .
contrary to the other terms of eq .
( [ 52 ] ) , its expansion does not contain @xmath113-dependent powers of @xmath64 , but is of the form @xmath413 with @xmath414 a set of @xmath64-independent coefficients . as for comparing powers of @xmath64 ,
this has the following consequence . since
, for all @xmath415 and @xmath416 , @xmath417 is satisfied only if @xmath418 and @xmath419 , and since @xmath420 can not be satisfied at all , this term contributes to the evolution equation only in the two - dimensional case . using eq .
( [ 035 ] ) the piece contributing , i.e. @xmath421 , may be expressed in terms of the operator @xmath31 which yields the last term in eq .
( [ 55 ] ) .
the traces appearing in eq .
( [ 55 ] ) are evaluated in appendix [ trace ] using heat kernel techniques . then combining the result with the lhs of the flow equation , eq .
( [ 040 ] ) , and comparing the coefficients of the invariants @xmath30 and @xmath422 leads to the desired system of coupled differential equations for @xmath423 and @xmath34 .
we obtain @xmath424 - 2d\,\phi^1_{d/2}(0)\bigg\}\;,\nonumber\end{aligned}\ ] ] @xmath425 \nonumber\\ & & -2c_2(d)\,\phi^1_{d/2 - 1}(0)+c_5(d)\,\phi^2_{d/2}(0)\nonumber\\ & & + 3\delta_{d,2}\left(1-\frac{1}{2}\eta_n(k)\right)\left[\frac{1}{1 -2\bar{\lambda}_k / k^2}-\frac{1}{1 -2\alpha\bar{\lambda}_k / k^2}\right]\bigg\}\;.\end{aligned}\ ] ] here @xmath426 , @xmath427 are cutoff - dependent `` threshold '' functions defined as @xmath428 the coefficients @xmath429 are given by @xmath430 in eq .
( [ 75 ] ) the terms proportional to @xmath412 arise not only from the last term of eq .
( [ 55 ] ) , but also by evaluating the `` primed '' traces , i.e. by subtracting the contributions coming from unphysical modes , see appendix [ trace ] for details .
all these contributions are obtained by expanding various functions @xmath431 with respect to @xmath404 and retaining only the term of zeroth order , @xmath432 . as we argued above , these are the only pieces of @xmath433 which may contribute to the evolution in the truncated parameter space .
furthermore , the heat kernel expansions of the traces corresponding to differentially constrained fields introduce additional contributions @xmath434 into eq .
( [ 75 ] ) . in appendix [ d=4 ]
we concentrate on the 4-dimensional case and compare our result for @xmath397 and for the corresponding rg flow of @xmath343 and @xmath34 with the one of ref .
@xcite where a cutoff of type b is used , too .
now we introduce the dimensionless , renormalized newton constant @xmath437 and the dimensionless , renormalized cosmological constant @xmath438 where @xmath33 denotes the corresponding dimensionful , renormalized newton constant at scale @xmath0 . inserting eq .
( [ 007 ] ) into @xmath439 leads to the relation @xmath440 then , by using eq .
( [ 74 ] ) , we obtain the following differential equation for the dimensionless cosmological constant : @xmath441 the @xmath442-function @xmath443 contains the quantities @xmath444 and @xmath445 which are defined as @xmath446 the corresponding @xmath442-function for @xmath435 may be determined as follows .
taking the scale derivative of eq .
( [ 003 ] ) leads to @xmath447g_k$}\;.\end{aligned}\ ] ] for the anomalous dimension @xmath448 we obtain from eq .
( [ 75 ] ) @xmath449 with @xmath450 , @xmath451 functions of @xmath436 , @xmath113 and @xmath314 given by @xmath452\bigg\}\;,\nonumber\\ b_2(\lambda_k;\alpha , d)&\equiv & -2(4\pi)^{1-d/2}\bigg\ { c_1(d)\,\widetilde{\phi}^1_{d/2 - 1}(-2\lambda_k ) + c_2(d)\,\widetilde{\phi}^1_{d/2 - 1}(-2\alpha\lambda_k)\nonumber\\ & & + c_3(d)\,\widetilde{\phi}^2_{d/2}(-2\lambda_k ) + c_4(d,\alpha)\,\widetilde{\phi}^2_{d/2}(-2\alpha\lambda_k)\nonumber\\ & & + 3\delta_{d,2}\left[\frac{1}{1 - 2\lambda_k}-\frac{1}{1 -2\alpha\lambda_k}\right]\bigg\}\;.\end{aligned}\ ] ] eq .
( [ 005 ] ) may now be solved for the anomalous dimension in terms of @xmath436 , @xmath435 , @xmath314 and @xmath113 : @xmath453 the system of coupled flow equations ( [ 001 ] ) , ( [ 004 ] ) is the main result of this section .
let us compare the flow equations ( [ 001 ] ) , ( [ 004 ] ) of the present paper with those obtained in ref .
@xcite . ref .
@xcite covers the case @xmath315 only , and the cutoff used there has a different structure than the present one . in ref .
@xcite , the @xmath13 for the cutoff a is formulated at the level of the _ complete _ field @xmath70 , i.e. , symbolically , @xmath454 , while the cutoff b of the present paper is based upon a similar action for the component fields : @xmath455 .
for @xmath315 the new @xmath442-function for @xmath436 , @xmath443 , agrees perfectly with the result in @xcite , whereas the coefficients @xmath450 , @xmath451 in the @xmath456-function for @xmath435 , @xmath457 , do not coincide with the corresponding results derived there .
however , with both cutoffs these coefficients are of the form @xmath458 in the present paper ( cutoff b ) the coefficients @xmath459 are obtained as @xmath460 while in @xcite ( cutoff a ) they are given by @xmath461 upon subtracting the coefficients in eq .
( [ 065 ] ) from those in eq .
( [ 066 ] ) we obtain @xmath462 and @xmath463 .
quite remarkably , the sum of the deviations @xmath464 vanishes not only in total but also separately for the gravitational contributions , involving @xmath465 and @xmath466 , and the contributions from the ghost , which contain @xmath467 and @xmath468 .
obviously , this amounts to a shift from the gravitational @xmath469-sector to the @xmath470-sector as well as to a shift between the corresponding ghost - sectors .
the simplicity of this result is somewhat mysterious .
because of its complexity it is impossible to solve the system of flow equations for @xmath435 and @xmath436 , eqs .
( [ 001 ] ) and ( [ 004 ] ) , exactly .
even a numerical solution would be a formidable task .
however , it is possible to gain important information about the general structure of the rg flow by looking at its fixed point structure . given a set of @xmath329-functions corresponding to an arbitrary set of dimensionless essential couplings @xmath61 , it is often possible to predict the scale dependence of the couplings for very small and/or very large scales @xmath0 by investigating their fixed points .
the fixed points are those points in the space spanned by the @xmath53 where all @xmath329-functions vanish .
( the essential couplings are those combinations of the couplings appearing in the action functional that are invariant under point transformations of the fields . )
fixed points are characterized by their stability properties .
a given eigendirection of the linearized flow is said to be uv or ir attractive ( or stable ) if , for @xmath35 or @xmath7 , respectively , the trajectories are attracted towards the fixed point along this direction .
the uv critical hypersurface in the space of all couplings is defined to consist of all trajectories that run into the fixed point for @xmath35 . in quantum field theory ,
fixed points play an important role in the modern approach to renormalization theory@xcite . at a uv fixed point
the infinite cutoff limit can be taken in a controlled way .
as for gravity , weinberg @xcite argued that a theory described by a trajectory lying on a _ finite - dimensional _ uv critical hypersurface of some fixed point is presumably free from unphysical singularities .
it is predictive since it depends only on a _ finite _ number of free ( essential ) parameters . in weinberg s words ,
such a theory is _
asymptotically safe_. asymptotic safety has to be regarded as a generalized , nonperturbative version of renormalizability .
it covers the class of perturbatively renormalizable theories , whose infinite cutoff limit is taken at the gaussian fixed point @xmath471 , as well as those perturbatively nonrenormalizable theories which are described by a rg trajectory on a finite - dimensional uv critical hypersurface of a non - gaussian fixed point @xmath472 and are nonperturbatively renormalizable therefore @xcite .
let us now consider the system of differential equations @xmath473 for a set of dimensionless essential couplings @xmath474 .
we assume that @xmath475 is a fixed point of eq .
( [ fp0 ] ) , i.e. @xmath476 for all @xmath477 .
we linearize the rg flow about @xmath475 which leads to @xmath478 where @xmath479 are the entries of the stability matrix @xmath480 .
diagonalizing @xmath481 according to @xmath482 , @xmath483 , where @xmath484 is the right - eigenvector of @xmath481 with eigenvalue @xmath485 we have @xmath486 the general solution to eq .
( [ gfp4 ] ) may be written as @xmath487 here @xmath488 are arbitrary real parameters and @xmath489 is a reference scale .
obviously the fixed point @xmath490 is uv attractive ( i.e. attractive for @xmath35 ) only if all @xmath491 corresponding to negative @xmath492 are set to zero .
therefore the dimensionality of the uv critical hypersurface equals the number of positive @xmath493 .
conversely , setting to zero all @xmath491 corresponding to positive @xmath494 , @xmath475 becomes an ir attractive fixed point ( approached in the limit @xmath7 ) with an ir critical hypersurface whose dimensionality equals the number of negative @xmath494 . in a slight abuse of language
we shall refer to the @xmath494 s as the _
critical exponents_. strictly speaking , the solution ( [ gfp5 ] ) and its above interpretation is valid only in such cases where all eigenvalues @xmath485 are real , which is not guaranteed since the matrix @xmath495 is not symmetric in general . if complex eigenvalues occur one has to consider complex @xmath491 s and to take the real part of eq .
( [ gfp5 ] ) , see below .
then the real parts of the critical exponents determine which directions in coupling constant space are attractive or repulsive . at this point
it is necessary to discuss the impact a change of the cutoff scheme has on the scaling behavior .
since the path integral for @xmath1 depends on the cutoff scheme , i.e. on the @xmath13 chosen , it is clear that the couplings and their fixed point values are scheme dependent .
hence a variation of the cutoff scheme , i.e. of @xmath46 , induces a change in the corresponding @xmath481-matrix .
so one might naively expect that also its eigenvalues , the critical exponents , are scheme dependent .
in fact , this is not the case . according to the general theory of critical phenomena and a recent reanalysis in the framework of the exact rg equations @xcite any variation of the cutoff scheme
can be generated by a specific coordinate transformation in the space of couplings with the cutoff held fixed .
such transformations leave the eigenvalues of the @xmath481-matrix invariant , so that the critical behavior near the corresponding fixed point is universal .
the positions of fixed points are scheme dependent but their ( non)existence and the qualitative structure of the rg flow are universal features . therefore a truncation can be considered reliable only if it predicts the same fixed point structure for all admissible choices of @xmath46 . in the context of the einstein - hilbert truncation
the space of couplings is parametrized by @xmath496 and @xmath497 .
the @xmath498-functions occuring in the two flow equations @xmath499 are given in eqs .
( [ 001 ] ) and ( [ 004 ] ) , respectively .
as we shall see in subsection b , they have a trivial zero at @xmath500 , referred to as the * gaussian fixed point .
* the analysis of subsection c reveals that there exists also a * non - gaussian fixed point * at @xmath501 , @xmath502 . in subsection c
we study its cutoff dependence and the cutoff dependence of the associated critical exponents employing the above @xmath442-functions of type b as well as those of refs .
@xcite based on the cutoff type a , with the families of shape functions ( [ h6 ] ) or ( [ supp ] ) inserted . in this subsection
we discuss the features of the gaussian fixed point @xmath503 . in order to investigate the rg flow in its vicinity
we expand the @xmath442-functions in powers of @xmath37 and @xmath36 according to eq .
( [ gfp2 ] ) of appendix [ approx ] and read off the @xmath481-matrix .
it takes the form @xmath504 here @xmath505 is a @xmath113-dependent parameter defined as @xmath506 diagonalizing the matrix ( [ gfp6 ] ) yields the ( obviously universal ) critical exponents @xmath507 and @xmath508 which are associated with the eigenvectors @xmath509 and @xmath510 . hence , for the linearized system obtained from ( [ gfp2 ] ) the solution ( [ gfp5 ] ) assumes the following form : @xmath511 since the expanded @xmath442-function @xmath457 of eq .
( [ gfp2 ] ) is @xmath436-independent up to terms of third order in the couplings we can easily calculate also the next - to - leading approximation for @xmath435 near the fixed point . in terms of the dimensionful quantity @xmath33
this improved solution reads @xmath512^{-1}\end{aligned}\ ] ] with @xmath513 a @xmath113- and @xmath314-dependent parameter . for @xmath514 and with the reference scale @xmath515 ( which is admissible only for trajectories lying on the ir critical hypersurface of the fixed point ) eq .
( [ gfp9 ] ) yields @xmath516 \;.\end{aligned}\ ] ] for the dimensionful cosmological constant we obtain from eq .
( [ gfp8 ] ) @xmath517 apart from the different expression of @xmath518 due to the new cutoff , the solutions ( [ gfp10 ] ) and ( [ gfp11 ] ) coincide in @xmath41 dimensions with those derived in ref .
@xcite by using a similar approximation scheme , see eqs .
( 5.18 ) and ( 5.25 ) of this reference .
let us now analyze the scaling behavior near @xmath503 . since @xmath519 the @xmath520-eigendirection , which coincides with the @xmath37-direction , is ir repulsive ( and thus uv attractive ) .
for @xmath521 , @xmath522 is positive which implies that the gaussian fixed point is uv attractive for any direction in the 2-dimensional parameter space . for @xmath45
we have @xmath523 so that the @xmath524-eigendirection is ir attractive ( and uv repulsive ) .
hence , in this case both the uv and the ir critical hypersurface of the gaussian fixed point are one - dimensional , i.e. they consist of a single trajectory . for the ir critical trajectory that hits the fixed point in the limit @xmath525 we have @xmath526 for sufficiently small values of @xmath0 @xcite , with @xmath33 given by eq .
( [ gfp10 ] ) .
since @xmath527 depends on the shape function @xmath316 , @xmath505 is not a universal quantity .
therefore the slope of the distinguished trajectory ( [ gfp12 ] ) is not fixed in a universal manner .
this is in accordance with the general expectation that the eigenvalues of @xmath481 should be universal , but not its eigenvectors . for @xmath528
the parameters @xmath529 and @xmath530 appearing in @xmath518 are scheme dependent as well .
furthermore , @xmath518 is a function of the gauge parameter @xmath314 .
hence @xmath518 is a nonuniversal quantity , too .
in the most interesting case of @xmath41 dimensions it takes the form @xmath531\;.\end{aligned}\ ] ] since @xmath532 and @xmath533 are positive for any admissible shape function we can infer from eq .
( [ gfp14 ] ) that @xmath534 is positive for all @xmath535 and negative for all @xmath536 . here
$ ] is a negative number of order unity .
thus , if we identify einstein gravity with the theory described by the ir critical trajectory of the gaussian fixed point , eq .
( [ gfp10 ] ) implies that einstein gravity is antiscreening for all @xmath535 , i.e. @xmath33 decreases as @xmath0 increases . on the other hand if @xmath536 gravity would exhibit a screening behavior .
as argued in subsection [ 4a ] , the gauge parameter should be regarded as a scale dependent parameter in an exact treatment where it is expected to approach the fixed point value @xmath334 .
setting @xmath333 from the outset we may conclude that the physical @xmath33 displays the antiscreening behavior found for @xmath535 . in ref .
@xcite a similar result was obtained with a cutoff of type a , while the above calculation employed the cutoff b. the only difference between our result for the behavior of @xmath33 and the one obtained in @xcite lies in the slightly differing value of @xmath538 which is a scheme dependent parameter . for the cutoff type a , @xmath539 is given by ( see @xcite ) @xmath540\end{aligned}\ ] ] so that @xmath541/6 $ ] in this case , while @xmath542 is the same with both cutoffs .
for @xmath315 , eq .
( [ gfp15 ] ) boils down to @xmath543/(6\pi)$ ] which equals the result obtained in the original paper @xcite .
this is because for @xmath315 the cutoff type a coincides with the one used in ref .
for a comparison of this result with the one for the cutoff type b we insert @xmath315 into eq .
( [ gfp14 ] ) which yields @xmath544/(24\pi)$ ] . using the exponential shape function @xmath316 with @xmath545 we have @xmath546 ,
@xmath547 so that @xmath548 for the cutoff type b , which lies rather close to the value @xmath549 obtained in @xcite for the cutoff type a. furthermore , we have @xmath550 where @xmath551 denotes the zeta function , and thus @xmath552 with both cutoffs .
now we turn to the nontrivial zeros of the set of @xmath442-functions @xmath553 given by eqs .
( [ 001 ] ) , ( [ 004 ] ) .
such non - gaussian fixed points @xmath554 satisfy the condition @xmath555 which follows immediately from eq .
( [ 004 ] ) . as a warm up we consider the case of @xmath43 dimensions with @xmath556 which can be dealt with analytically and for which the existence of a non - gaussian fixed point has already been shown @xcite . in this case
the condition ( [ ngfp0 ] ) takes the form @xmath557 with @xmath558 solving eq .
( [ ngfp6 ] ) for @xmath559 and expanding the result with respect to @xmath38 leads to @xmath560^{-1}\,\varepsilon + { \cal o}\left(\varepsilon^2\right)\;.\end{aligned}\ ] ] furthermore , expanding also @xmath561 with respect to @xmath38 and equating equal powers of @xmath38 yields @xmath562^{-1 } \phi^1_1(0)\,\varepsilon+{\cal o}\left(\varepsilon^2\right)\;.\end{aligned}\ ] ] in particular we obtain @xmath563 which implies @xmath564^{-1}\,\varepsilon+{\cal o}\left(\varepsilon^2 \right)$ ] .
the parameters @xmath565 , @xmath566 , may be obtained from the zeroth order terms of the expansions @xmath567 which take the form @xmath568 inserting @xmath569 into eq .
( [ uni1 ] ) and using that @xmath570 is scheme independent @xcite yields @xmath571 in contrast to the universal quantity @xmath572 , @xmath573 depends on the shape of @xmath316 via @xmath574 .
however , @xmath573 does not enter the leading order term of @xmath575 and @xmath559 , which may now be written as @xmath576 the leading order term of @xmath575 is nonuniversal since it contains the scheme dependent parameter @xmath532 .
this is not the case for @xmath559 whose leading order contribution has a universal meaning .
let us now analyze the scaling behavior near the non - gaussian fixed point ( [ ngfp9 ] ) .
one finds that the associated @xmath481-matrix is of the form @xmath577 from ( [ ngfp10 ] ) we obtain the critical exponents @xmath578 and @xmath579 . @xmath580 and @xmath522 are scheme independent up to terms of @xmath581 .
@xmath580 depends on the gauge parameter . for @xmath315
the critical exponents coincide with those following from the @xmath442-functions of ref .
@xcite , where the cutoff type a is used .
these findings nicely confirm that , to lowest order , _ the critical exponents are the same for the cutoffs a and b and are independent of @xmath316 . _ for @xmath582 both critical exponents are positive .
hence the non - gaussian fixed point ( [ ngfp9 ] ) is uv attractive for all trajectories so that the condition for the asymptotic safety scenario is met .
it is interesting to investigate whether this result stabilizes in the sense that more general truncations including higher powers of the curvature tensor reproduce this fixed point and lead to a finite - dimensional uv critical hypersurface .
in @xcite we will discuss this point in detail . in @xmath41 dimensions , and for the cutoff
a , the non - gaussian fixed point of the einstein - hilbert truncation was first discussed in @xcite , and in ref .
@xcite the @xmath314- and @xmath316-dependence of its projection @xmath584 onto the @xmath36-direction has been investigated . however , since for @xmath585 the cutoff of type a is introduced by an ad hoc modification of the standard one - loop determinants it is not clear whether it may be expressed in terms of an action @xmath13 , except for the case @xmath315 @xcite .
since a specification of @xmath13 is indispensable for the construction of @xmath1 , the status of the results derived in @xcite is somewhat unclear . in the following
we determine the fixed point properties using different cutoffs of type b , for which a @xmath13 is known to exist , and compare them to the analogous results for the cutoff a. in a first attempt to determine the non - gaussian fixed point we neglect the cosmological constant and set @xmath586 , thereby projecting the renormalization group flow onto the one - dimensional space parametrized by @xmath36 . in this case
the non - gaussian fixed point is obtained as the nontrivial solution of @xmath587 .
it is determined in appendix [ approx ] with the result given by eq .
( [ h1 ] ) . in order to get a first impression of the position of @xmath490 we insert the exponential shape function with @xmath545 into eq .
( [ h1 ] ) and set @xmath41 , @xmath315 .
we obtain @xmath588 . assuming that for the combined @xmath37-@xmath36 system both @xmath490 and @xmath589 are of the same order of magnitude as @xmath490 above we expand the @xmath442-functions about @xmath590 and neglect terms of higher orders in the couplings . again in appendix [ approx ] we determine the non - gaussian fixed point for the corresponding system of differential equations . inserting the shape function ( [ expcut ] ) and setting @xmath41 , @xmath315 , we find @xmath591 .
= 7.7 cm ( a ) = 7.7 cm ( b ) = 7.7 cm ( a ) = 7.7 cm ( b ) = 7.7 cm ( a ) = 7.7 cm ( b ) in order to determine the _ exact _ position of the non - gaussian fixed point @xmath44 we have to resort to numerical methods . given a starting value for the fixed point , e.g. one of the approximate solutions above
, the program we use determines a numerical solution which is exact up to an arbitrary degree of accuracy . under the same conditions as above ,
i.e. @xmath545 , @xmath41 , we obtain @xmath592 next we study the gauge and scheme dependence of the non - gaussian fixed point .
the scheme dependence is investigated by looking at the @xmath319-dependence introduced via the family of exponential shape functions ( [ h6 ] ) where @xmath319 parametrizes the profile of @xmath316 .
[ fi1 ] shows @xmath593 obtained from the approximation @xmath586 , while figs .
[ fi2 ] and [ fi3 ] display the ( exact ) functions @xmath593 and @xmath594 resulting from the combined @xmath37-@xmath36 system . in each of these figures
the plot on the lhs ( i.e. figs .
1,2,3(a ) ) is obtained from the cutoff type a and the one on the rhs ( i.e. figs .
1,2,3(b ) ) is obtained from the cutoff type b used in the present paper .
our results establish the existence of the non - gaussian fixed point in a wide range of @xmath314- and @xmath319-values .
as expected , the position of the fixed point turns out to be @xmath319- , i.e. scheme dependent , but the crucial point is that it exists for any of the cutoffs employed .
this is one of the important results of our analysis because it gives a first hint at the reliability of the einstein - hilbert truncation .
as for the @xmath314-dependence , @xmath333 is , in principle , the only relevant case since according to subsection [ 4a ] , @xmath333 is assumed to be the physical value of the gauge parameter . in practical calculations @xmath315 is often used instead , because this simplifies the evaluation of the flow equation considerably . in @xcite , for instance , all calculations are performed with @xmath315 for this reason .
therefore it is necessary to compare the gauges @xmath333 and @xmath315 in order to judge whether the results obtained by using @xmath315 are a sensible approximation to the physical case @xmath333 .
here we see that this is indeed the case . as for comparing different types of cutoffs , we recognize from fig .
[ fi1 ] that , in the approximation @xmath586 , the @xmath319-dependence of @xmath490 is much weaker for type b than for type a. contrary to this , both cutoffs yield nearly the same results for @xmath490 and @xmath589 if we consider the combined @xmath37-@xmath36 system , see figs .
[ fi2 ] and [ fi3 ] .
furthermore , the scheme dependence of @xmath490 in fig .
[ fi2 ] is stronger than in fig .
[ fi1](b ) , but much weaker than in fig .
[ fi1](a ) .
[ fi1](a ) reproduces the result of ref .
@xcite obtained from the cutoff a , see fig . 2 of this reference .
it should be noted that we are forced to restrict our considerations to shape functions ( [ h6 ] ) with @xmath595 .
this is because for @xmath596 the numerical integrations are plagued by convergence problems which is due to the fact that in @xmath41 dimensions the threshold functions in @xmath597 and @xmath457 diverge in the limit @xmath598 , see also @xcite .
because the scale @xmath0 enters the flow equation via @xmath46 as a purely mathematical device it is clear that the functions @xmath599 and their uv limits @xmath600 are scheme dependent and not directly observable therefore .
it can be argued that the product @xmath49 must be scheme independent , however . while @xmath0 and , at a fixed value of @xmath0 , @xmath33 and @xmath34 can not be measured separately , we may invert the function @xmath601 and insert the result @xmath602 into @xmath34 .
this leads to a relationship between newton s constant and the cosmological constant which , at least in principle , could be tested experimentally : @xmath603 .
in general this relation depends on the rg trajectory chosen ( specified by its ir values @xmath604 and @xmath605 , for instance ) , but in the fixed point regime all trajectories approach @xmath606 and @xmath607 which gives rise to @xmath608 eq .
( [ ngfp4 ] ) is valid if @xmath609 and @xmath610 .
( we define the planck mass in terms of the ir limit of @xmath33 , @xmath611 . ) assuming that @xmath612 and @xmath613 have the status of observable quantities , eq .
( [ ngfp4 ] ) shows that @xmath49 must be observable , and hence scheme independent , too .
( for a related discussion see @xcite . )
below the planck regime the function @xmath614 becomes much more complicated than @xmath615 ( which follows already from dimensional analysis ) because the dimensionful quantities @xmath604 and @xmath605 enter explicitly there .
= 7.9 cm = 5.2 cm ( a ) = 7.9 cm = 5.2 cm ( b ) as for the universality of @xmath49 , it is also interesting to note that , for any @xmath0 , the product @xmath616 is essentially the inverse of the on shell value of @xmath1 .
the stationary points of ( [ in2 ] ) with @xmath617 satisfy einstein s equation @xmath618 .
hence @xmath619 , so that from ( [ in2 ] ) in four dimensions , @xmath620=-\frac{v}{8\pi g_k\bar{\lambda}_k } = -\frac{v}{8\pi g_k\lambda_k}\;.\end{aligned}\ ] ] here we used that , for dimensional reasons , @xmath621 where @xmath622 is a finite , positive constant for any solution with a finite four - volume .
= 7.9 cm = 5.2 cm ( a ) = 7.9 cm = 5.2 cm ( b ) = 7.9 cm = 5.2 cm ( c ) = 7.9 cm = 5.2 cm ( d ) = 7.9 cm = 5.2 cm ( a ) = 7.9 cm = 5.2 cm ( b ) quite remarkably , the universality of the product @xmath49 is confirmed by our results in a rather impressive manner , as is illustrated in figs .
[ fi4]-[fi7 ] .
[ fi4](a ) contains several parametric plots of @xmath623 for various values of @xmath314 , obtained from the @xmath442-functions ( [ 001 ] ) and ( [ 004 ] ) which are based on the cutoff type b. the hyperbolic shape of these plots is a first hint at the @xmath319-independence of the product @xmath49 .
its direct confirmation is supplied by fig .
[ fi5 ] which shows @xmath490 , @xmath589 , and @xmath49 as functions of @xmath319 for @xmath333 ( fig . [ fi5](a),(b ) ) and @xmath315 ( fig . [ fi5](c),(d ) ) , again using the cutoff type b. in fig .
[ fi5](a),(c ) these functions are plotted in the range of values @xmath624 while fig .
[ fi5](b),(d ) contains the sector corresponding to @xmath625 where the largest changes in @xmath589 and @xmath490 occur . in any of these figures
the product of @xmath589 and @xmath490 is almost constant for the whole range of @xmath319-values considered .
its universal value is @xmath626 obviously the difference between the physical case @xmath333 and the case preferred for technical reasons , @xmath315 , is rather small .
it is reassuring to see that employing the @xmath442-functions of refs .
@xcite which are based on the cutoff a we obtain almost identical results .
they are illustrated by means of fig .
[ fi6 ] which shows @xmath490 , @xmath589 , and @xmath49 as functions of @xmath319 , @xmath625 , for @xmath333 ( fig . [ fi6](a ) ) and @xmath315 ( fig . [ fi6](b ) ) .
= 7.9 cm = 5.2 cm ( a ) = 7.9 cm = 5.2 cm ( b ) it is also interesting to compare the above results with those obtained from a different shape function . fig .
[ fi7 ] displays the behavior of @xmath490 , @xmath589 , and their product resulting from the @xmath442-functions ( [ 001 ] ) and ( [ 004 ] ) of the cutoff b , with the family of shape functions @xmath627 with compact support , eq . ( [ supp ] ) , inserted . here
@xmath628 parametrizes the profile of these shape functions . in fig .
[ fi7](a ) we present a parametric plot of @xmath629 for @xmath333 starting at @xmath630 and ending at @xmath631 .
furthermore , fig .
[ fi7](b ) shows the product @xmath49 as a function of @xmath632 for @xmath333 and @xmath315 . for @xmath633 , the parametric plot in fig .
[ fi7](a ) exhibits an approximately linear behavior which leads to a @xmath49-plateau in fig . [
fi7](b ) where @xmath49 is nearly constant .
remarkably , the position of these plateaus coincides quite precisely with those of the corresponding plateaus in figs .
[ fi5 ] and [ fi6 ] obtained with the other cutoffs . as for the quality of the einstein - hilbert truncation this result is rather encouraging .
for @xmath634 the curves in fig .
[ fi7 ] have a rather strong and erratic @xmath632-dependence .
this is because @xmath627 approaches a sharp cutoff as @xmath635 , which introduces discontinuities into the integrands of the threshold functions @xmath426 and @xmath636 .
already for @xmath637 the @xmath442-functions start to `` feel '' the sharp cutoff limit so that the results can not be trusted beyond this point .
as for the @xmath314-dependence of the fixed point , in fig .
[ fi4](b ) we present parametric plots of @xmath638 for various fixed values of @xmath319 .
here we used the cutoff type b with the shape function ( [ h6 ] ) inserted .
these plots start at positions in the @xmath37-@xmath36 plane which correspond to @xmath333 and which are different for the distinct @xmath319-values . as @xmath639
all curves run into the gaussian fixed point . * to summarize : * + a ) in 4 dimensions
, the einstein - hilbert truncation leads to a non - gaussian fixed point with positive values of @xmath589 and @xmath490 for all admissible cutoffs , both of type a and type b. the scheme independence of this prediction is a nontrivial result .
( to emphasize this point we mention that in higher dimensions , where the einstein - hilbert truncation is less reliable , the fixed point exists or does not exist depending on the cutoff chosen @xcite . ) + b ) universal quantities are strictly cutoff independent only in an exact treatment .
any truncation leads to a scheme dependence of these quantities .
the extent of this scheme dependence is a measure for the reliability of the truncation .
the product @xmath49 is an example of a universal quantity .
while we find a considerable scheme dependence of @xmath490 and @xmath589 separately , their product is scheme independent at a quite amazing level of accuracy , see fig .
[ fi5 ] . as for the reliability of the einstein - hilbert truncation
, we consider this result a highly nontrivial confirmation of our assumption that the region of parameter space where the fixed point occurs is well described by this truncation ansatz so that the fixed point also exists in the _ exact _ theory and is not a truncation artifact . before
continuing our analysis of the @xmath41 dimensional case we study the @xmath113-dependence of the non - gaussian fixed point .
this is done by means of the parametric plots in fig .
[ fi8 ] which are obtained from the @xmath640-functions of type b , eqs .
( [ 001 ] ) and ( [ 004 ] ) , with the shape function ( [ h6 ] ) with @xmath545 inserted .
[ fi8](b ) shows @xmath641 in @xmath642 for @xmath315 . remarkably , this plot is almost identical with that in fig . 4 of ref .
@xcite which was derived from the @xmath442-functions based on the cutoff type a. = 7.9 cm = 5.2 cm ( a ) = 7.9 cm = 5.2 cm ( b ) in order to gain information about the behavior of the fixed point in @xmath643 we also plotted @xmath641 in @xmath644 , but this time for @xmath333 , see fig .
[ fi8](a ) . for @xmath113 beyond @xmath645 ,
the value of @xmath490 increases significantly , whereas @xmath589 seems to approach a constant value .
however , this result might be a truncation artifact since it is plausible to assume that in higher dimensions the einstein - hilbert truncation becomes less reliable .
this is because with increasing dimensionality @xmath113 the number of terms which are relevant at the non - gaussian fixed point and which are neglected in the einstein - hilbert truncation increases most probably . at the non - gaussian fixed point
the scaling dimensions of local invariants such as @xmath646 , @xmath647 , @xmath648 , @xmath649 , etc . are not known a priori .
we only know that _ with respect to the gaussian fixed point _
all local monomials @xmath650 are relevant or marginal if their canonical mass dimension does not exceed @xmath113
. a sensible truncation should retain at least the relevant terms , whence it is clear that the number of terms needed increases with the dimensionality @xmath113 .
( in 4 dimensions , @xmath646 and @xmath647 are relevant and the @xmath651-invariants are marginal . ) by analogy we expect that the description of the non - gaussian fixed point , too , requires increasingly high powers of the curvature when @xmath113 is increased @xcite .
let us now return to the 4-dimensional case and analyze the critical behavior near the non - gaussian fixed point . in order to get a first impression of its features we restrict our considerations to the cutoff type b with the exponential shape function ( [ h6 ] ) with @xmath545 and to the gauge @xmath315 .
in this case we have @xmath652 , see above .
the corresponding @xmath495-matrix assumes the form @xmath653 it leads to a pair of _ complex _ critical exponents @xmath654 and @xmath655 . for the real quantities
@xmath656 and @xmath657 we find @xmath658 and @xmath659 .
( in general we define @xmath580 as the critical exponent with the positive imaginary part so that @xmath660 . ) the behavior of @xmath436 and @xmath435 near the fixed point is described by the real part of eq .
( [ gfp5 ] ) in this case . using that @xmath661 , and setting @xmath662 and @xmath663 , the general solution to the linearized flow equation
may then be written as @xmath664 { \rm re}\,v\nonumber\\ & & + \left[{\rm re}\,c\,\sin\left(\theta''\,t\right)-{\rm im}\,c \,\cos\left(\theta''\,t\right)\right]{\rm im}\,v\bigg\}e^{-\theta ' t}\;.\end{aligned}\ ] ] here @xmath665 .
obviously the non - gaussian fixed point is uv ( ir ) attractive if @xmath666 @xmath667 .
the imaginary parts @xmath668 of the critical exponents do not influence the stability of the fixed point .
they only give rise to a rotation of the vector @xmath669 about the fixed point . in the case under consideration
we have @xmath670 which implies that the non - gaussian fixed point is uv attractive in both directions of @xmath671-space .
all rg trajectories which reach its basin of attraction spiral into the fixed point for @xmath35 .
thus , the einstein - hilbert truncation predicts all the ingredients which are necessary for the asymptotic safety scenario and the nonperturbative renormalizability of 4-dimensional quantum gravity .
clearly the dimensionality of the uv critical hypersurface can not be determined within the present approach .
we shall come back to this question in the framework of a more general truncation including higher derivative terms @xcite .
= 7.9 cm = 5.2 cm ( a ) = 7.9 cm = 5.2 cm ( b ) = 7.9 cm = 5.2 cm ( a ) = 7.9 cm = 5.2 cm ( b ) = 7.9 cm = 5.2 cm ( a ) = 7.9 cm = 5.2 cm ( b ) as discussed in subsection [ 5a ] the critical exponents are universal in an exact treatment , in contrast to @xmath490 and @xmath589 .
however , in a truncated parameter space a scheme dependence is expected to occur as an artifact of the truncation .
therefore we may use this scheme dependence of the critical exponents to judge the quality of our truncation .
also in this respect the einstein - hilbert truncation yields satisfactory results , which we display in figs .
[ fi9]-[fi11 ] .
first of all it should be noted that the critical exponents obtained from our numerical analysis have a nonzero imaginary part in any of the cases considered
[ fi9]-[fi11](a ) show their real part @xmath656 while their imaginary part @xmath657 is depicted in figs . [
fi9]-[fi11](b ) . figs .
[ fi9 ] and [ fi10 ] are derived from the @xmath442-functions ( [ 001 ] ) and ( [ 004 ] ) based on the cutoff b of the present paper , with the family of shape functions ( [ h6 ] ) ( in fig .
[ fi9 ] ) and ( [ supp ] ) ( in fig . [ fi10 ] ) inserted .
for comparison , fig .
[ fi11 ] is obtained from the @xmath442-functions of refs .
@xcite derived from the cutoff type a , with the family of shape functions ( [ h6 ] ) inserted .
the figures contain various plots for distinct values of @xmath314 , which describe the @xmath319- or @xmath632-dependence of @xmath656 and @xmath657 .
the @xmath656-plots in figs . [ fi9](a ) and [ fi11](a ) for the cutoffs b and a , respectively , exhibit a similar @xmath319-dependence , and the same holds true for the @xmath657-plots in figs . [ fi9](b ) and [ fi11](b ) .
moreover , in the range @xmath625 one recognizes a total variation of both @xmath656 and @xmath657 which is approximately of the order of magnitude of @xmath656 and @xmath657 , respectively . for @xmath672
there remains only a rather weak dependence on @xmath319 , such that the functions @xmath673 and @xmath674 develop a plateau - like shape .
the @xmath319-dependence is slightly stronger for the cutoff type b than for type a , and the positions of the `` plateaus '' are different for both cutoff types . for @xmath315 , for instance , we have @xmath675 and @xmath676 using type a , while employing type b yields @xmath677 and @xmath678 .
these differences have to be interpreted as truncation artifacts .
the above results may now be compared with those for the cutoff b with the shape functions with compact support , eq .
( [ supp ] ) , inserted .
our results are shown in fig .
remarkably , @xmath656 is almost constant in the range @xmath679 .
furthermore , the change in @xmath657 is only very weak in this region and occurs mainly for @xmath680 . for @xmath681 @xmath657 is almost constant as well . in the region @xmath682 both @xmath656 and @xmath657 strongly vary with @xmath632 which is caused by the sharp cutoff limit as @xmath635 , see above .
the position of the plateaus is nearly identical with those obtained above by using the same cutoff b , but the shape function ( [ h6 ] ) . for @xmath315
we have @xmath683 and @xmath684 .
as discussed above , @xmath333 is assumed to be the physical value of the gauge parameter while in practical applications @xmath315 is often preferred . for a further justification of this approximation
we compare the critical exponents obtained from @xmath315 with those obtained from the ( physical ) gauge @xmath333 .
the results are qualitatively the same , but quantitatively we find a relative deviation of about 10 to 20 per - cent .
this has to be kept in mind when calculations are performed with @xmath315 . * to summarize : * for all admissible cutoffs , both of type a and type b , the non - gaussian fixed point is uv attractive in both directions of parameter space .
it is characterized by a pair of complex conjugate critical exponents which leads to spiral - type trajectories in its vicinity .
the exact critical exponents are universal .
those obtained from the einstein - hilbert truncation are approximately scheme independent , but their scheme dependence is stronger than that of @xmath49 .
this might be related to the fact that the @xmath648-term neglected by the einstein - hilbert truncation is ( very weakly ) relevant at the non - gaussian fixed point .
we shall come back to this point in ref .
[ s6 ] the uv fixed point is characterized by an anomalous dimension @xmath685 .
we can use this information in order to determine the effective momentum dependence of the dressed graviton propagator for momenta in the fixed point regime , i.e. for @xmath686 . expanding the truncated @xmath1 about flat space and omitting the standard tensor structures we find the inverse propagator @xmath687 .
the conventional dressed propagator @xmath688 contained in @xmath689 obtains from @xmath690 in the limit @xmath7 . assuming that the actual physical cutoff scale is the momentum @xmath302 itself ( which is indeed true for @xmath691 )
, the @xmath0-evolution of @xmath692 stops at the threshold @xmath693 .
therefore @xmath694 because @xmath695 when @xmath696 is ( approximately ) constant
. in @xmath113 dimensions , and for @xmath697 , the fourier transform of @xmath698 yields the following euclidean propagator in position space : @xmath699 this is a standard result well known from the theory of critical phenomena , for instance . in the latter case it applies to large distances , while in quantum gravity we are interested in the extreme short distance regime governed by the uv fixed point .
however , as it stands eq .
( [ gp2 ] ) is not valid for the case of interest , @xmath41 and @xmath700 .
for @xmath700 the dressed propagator is @xmath701 which , in @xmath41 dimensions , has the following representation in position space : @xmath702 here @xmath703 is an arbitrary constant with the dimension of a mass . usually in ordinary matter field theories on flat space a @xmath704-propagator is considered problematic because it is incompatible with a positive spectral density .
it is by no means clear , however , that the familiar notions of positivity , analyticity , and causality which are appropriate for minkowski space are of relevance to the situation under consideration . for @xmath35
the `` on shell '' spacetimes of @xmath1 have a large curvature @xmath20 , after all .
so let us assume that the improvement ( [ gp1 ] ) is indeed correct and that in the fixed point regime , i.e. for distances much smaller than the planck length , the effective graviton propagator has a logarithmic dependence on the distance .
this result is quite remarkable because it implies a kind of dimensional reduction from 4 to 2 dimensions .
in fact , eq . ( [ gp3 ] ) is precisely what one obtains from a standard @xmath705-propagator in @xmath419 dimensions .
this means that , in a certain sense , spacetime appears to be two - dimensional when it is probed by a very high energetic graviton . since , symbolically , @xmath706 , the propagator ( [ gp3 ] ) yields the curvature - curvature correlation function @xmath707 its short - distance singularity has to be contrasted with the @xmath708 behavior one finds at tree level .
here @xmath709 stands for the curvature scalar or for any component of the riemann or ricci tensor .
switching for a moment to spacetimes with a lorentzian signature , it is interesting to look at the linearized gravitational field produced very close to a _
static _ source .
decomposing @xmath710 , the relevant green s function for static problems reads @xmath711 in our case this is the 3-dimensional fourier transform of @xmath712 provided @xmath713 . in an ,
admittedly somewhat naive , newtonian language this result would mean that a point mass located at @xmath714 creates a gravitational potential which behaves as @xmath715 as long as @xmath716 is much smaller than the planck length .
probably this linear potential is related to a similar phenomenon shown by the renormalization group improved schwarzschild black hole which has been constructed recently @xcite .
the radial geodesics in this spacetime , in a fully relativistic treatment , experience a linear .
this corresponds to the case @xmath717 which ignores the impact the background curvature has on the cutoff identification , see eq .
( 4.35 ) of @xcite .
this is consistent with the fact that eq .
( [ gp3 ] ) is valid for a flat background . ]
repulsive `` potential '' close to the core of the black hole .
[ s7 ] on the formal side , the main result of this paper is the construction of a new exact rg equation for the gravitational effective average action .
it is formulated in terms of the component fields appearing in the tt - decomposition of the metric .
it is defined on a sufficiently large class of background spacetimes so as to facilitate the projection of the rg flow onto very general truncated parameter spaces .
it also helps in finding admissible ir cutoffs .
a formalism of this kind is mandatory for truncations including higher - derivative invariants , matter fields , or a running gauge fixing , for instance . in a forthcoming paper
@xcite we shall use this technology in order to explore a more general truncation with a @xmath84-term . in the present paper we analyzed the einstein - hilbert truncation with the new equation and a new cutoff . after deriving the exact functional equation in section [ s2 ] , we discussed in section [ s3 ] a general strategy for constructing an appropriate cutoff operator in the context of a class of truncations which is still very general . in section [ s4 ] we specialized for the einstein - hilbert truncation and derived the nonperturbative @xmath442-functions which govern the rg evolution of @xmath435 and @xmath436 . on the applied side ,
most of our results concern the non - gaussian fixed point of the @xmath37-@xmath36 system in 4 dimensions which we analyzed in section [ s5 ] .
if this fixed point is actually present in the exact theory , its importance can hardly be overestimated .
its existence would imply that in spite of its notorious perturbative nonrenormalizability quantum einstein gravity is most probably renormalizable at the nonperturbative level and thus qualifies as a fundamental ( microscopic ) quantum theory of gravity .
clearly the crucial question is whether the fixed point of the einstein - hilbert truncation is indeed genuine or merely a truncation artifact . in order to get an impression of the reliability of the einstein - hilbert truncation we investigated how its predictions change when we vary the cutoff which is built into the rg equations .
we used both the original cutoff of type a , formulated in terms of @xmath70 , and the new cutoff of type b which is natural in the tt - approach ; the cutoffs were equipped with two different one - parameter families of shape functions . in 4 dimensions
, we found that the einstein - hilbert truncation leads to a non - gaussian fixed point for all admissible cutoffs , both of type a and b. the robustness of this prediction is a nontrivial result since in higher dimensions , for instance , where this truncation is less reliable , the fixed point is present for some cutoffs but absent for others .
another consistency test successfully passed by the truncation is that all cutoffs agree on _ positive _ values of @xmath490 and @xmath589 .
a negative @xmath490 would probably be unacceptable for stability reasons , but there is no mechanism in the flow equation which would exclude it on general grounds .
in fact , @xmath718 is realized for @xmath521 . for all cutoffs of type a and b
the non - gaussian fixed point is found to be uv attractive in both directions of the @xmath37-@xmath36 plane .
the linearized flow in its vicinity is always characterized by a pair of complex conjugate critical exponents leading to spiral - type trajectories which hit the fixed point for @xmath35 .
this is precisely the stability property needed for asymptotic safety . by definition ,
universal quantities are scheme- , or cutoff - independent in an exact calculation .
truncations lead to a scheme dependence , however .
we can use the degree of the scheme dependence as a measure for the reliability of the truncation .
the critical exponents and , as we argued , the product @xmath49 are universal quantities .
the existence of fixed points is a universal feature of the rg flow , but not their precise location in parameter space .
the critical exponents were indeed found to be reasonably constant for a wide range of shape parameters .
the universality properties of @xmath49 are much more impressive though . while we found a considerable scheme dependence of @xmath490 and @xmath589 separately ,
their product is scheme independent at a rather amazing level of accuracy .
we believe that these results hardly can be a mathematical accident , and we consider them a very nontrivial confirmation of the hypothesis that the region of parameter space where the non - gaussian fixed point is situated is well described by the einstein - hilbert truncation . as a consequence ,
the fixed point should exist in the exact theory , too .
apart from the renormalizability issue the nontrivial fixed point is also intriguing from a `` phenomenological '' point of view .
its relevance for the structure of black holes @xcite and the cosmology of the planck era @xcite has been pointed out already .
moreover , we saw in section [ s6 ] that the rg improvement of the graviton propagator suggests a kind of dimensional reduction from 4 to 2 dimensions when spacetime is probed at sub - planckian length scales .
+ acknowledgement : we would like to thank c. wetterich for many helpful discussions .
[ dec ] in subsection [ 2b ] we introduced the tt - decomposition @xmath719 valid for arbitrary symmetric rank 2 tensors defined on either closed or open , asymptotically flat riemannian @xmath113-spaces . here
@xmath77 , @xmath720 and @xmath119 represent the transverse traceless , longitudinal traceless and pure trace part , respectively , which are mutually orthogonal . according to eq .
( [ f ] ) these parts may be expressed in terms of pure spin-2 , spin-1 and spin-0 component fields @xmath77 , @xmath108 , @xmath79 , @xmath80 . in this subsection
we show that the component fields can be obtained by applying certain operators @xmath721 to the full field @xmath70 . in the following these operators
will be termed pseudo - projectors . as a first step
we express the longitudinal - transverse and the pure trace part as @xmath722 here the operator @xmath723 maps vectors onto longitudinal traceless tensors . given a tensor @xmath720 , the equation @xmath724 is solved by @xmath725 where @xmath108 is a transverse vector and @xmath79 a scalar .
this solution is unique up to the addition of conformal killing vectors ( ckv s ) , as discussed in subsection [ 2b ] .
we recover eq .
( [ f ] ) by inserting eqs .
( [ ap2 ] ) , ( [ 099 ] ) into eq .
( [ ap1 ] ) .
contrary to @xmath726 , the scalar @xmath80 is uniquely determined by @xmath70 . now taking the covariant divergence of eq .
( [ ap1 ] ) with eq .
( [ ap2 ] ) inserted , and using the transversality requirement @xmath727 leads to @xmath728 with the operator @xmath729 defined by @xmath730 as shown in ref .
@xcite , @xmath729 is a positive definite , hermitian operator mapping vectors onto vectors .
moreover , the equation @xmath731 with an arbitrary given vector @xmath732 always possesses solutions @xmath726 which are unique up to ckv s .
however , even if these ckv s exist they cause no problems in solving eq .
( [ ap3 ] ) for @xmath726 , see ref .
@xcite for details . in order to determine this solution we have to invert @xmath729 . for this purpose
we assume that @xmath729 ( and any additional operator that needs to be inverted in the course of this discussion ) has a complete set of orthogonal eigenfunctions and that the corresponding eigenvalues do not have zero as an accumulation point
. then @xmath733 exists and the solution to eq .
( [ ap3 ] ) may be written as @xmath734 here the operator @xmath735 maps tensors onto vectors according to @xmath736 with @xmath737 being the operator that projects symmetric tensors onto their traceless part : @xmath738 taking the covariant divergence of the solution ( [ ap9 ] ) , inserting eq .
( [ 099 ] ) , and using @xmath739 yields @xmath740 this leads to the final result for @xmath108 and @xmath79 in terms of @xmath70 : @xmath741 hence the pseudo - projectors @xmath742 which map @xmath70 onto the individual component fields are obtained as @xmath743 here we defined @xmath744 and @xmath745 in terms of the redefined fields @xmath153 and @xmath154 which are related to @xmath108 and @xmath79 by eq .
( [ y ] ) .
furthermore , the pseudo - projectors for the transverse decomposition of arbitrary vector fields can be inferred from eq .
( [ ap11 ] ) . for @xmath746 with @xmath747
they are determined by @xmath748 obviously @xmath749 maps tensors onto vectors , @xmath750 and @xmath751 map tensors onto scalars , and @xmath752 maps vectors onto scalars . hence these operators can not be projection operators in the usual sense .
however , projection operators @xmath753 mapping arbitrary @xmath70 onto @xmath720 or @xmath119 , or arbitrary @xmath726 onto their longitudinal component can be constructed from them @xcite . since the @xmath754 s map vectors and symmetric tensors onto their component fields they generate a kind of projection in a wider sense of the word .
therefore we call the @xmath754 s pseudo - projectors .
contrary to @xmath749 , @xmath750 , @xmath751 and @xmath752 , the operators @xmath755 and @xmath756 are genuine projection operators mapping symmetric tensors and vectors onto their @xmath757 and @xmath758 component , respectively . in the present paper
we need a formulation for @xmath1 which allows for a description in terms of the fundamental as well as the component fields .
the translation between the two descriptions can be achieved by using the pseudo - projectors for the construction of the cutoff and for an appropriate decomposition of the source terms .
starting from the definition of the cutoff in terms of the fundamental fields , eq .
( [ v ] ) , we choose the cutoff operators @xmath178 and @xmath179 as @xmath759 here @xmath760 represents a block of the more general matrix operator @xmath761 for which @xmath762 if both @xmath193 and @xmath763 , or if both @xmath195 and @xmath764 .
the operators @xmath765 and @xmath766 are required to satisfy the hermiticity conditions @xmath767 and @xmath191 .
furthermore , we set @xmath768 , @xmath769 , @xmath770 , @xmath771 , @xmath772 , @xmath773 , and @xmath774 , @xmath775 denote the hermitian conjugates of @xmath776 , @xmath777 .
hence the operators @xmath754 appearing in eq .
( [ ap15 ] ) can be inferred from eqs .
( [ ap13 ] ) , ( [ ap7 ] ) . inserting eq .
( [ ap15 ] ) into eq .
( [ v ] ) leads to @xmath778&=&\frac{1}{2}\sum\limits_{\zeta_1,\zeta_2 \in\{h^t,\xi,\sigma,\phi\}}\left<{\bf\pi}_{\zeta_1 } h,({\cal r}_k)_{\zeta_1 \zeta_2}\,{\bf\pi}_{\zeta_2}\,h\right>\nonumber\\ & & + \sum\limits_{\bar{\vartheta}_1\in\{\bar{c}^t,\bar{\eta}\}}\ ; \sum\limits_{\vartheta_2\in\{c^t,\eta\}}\left<{\bf\pi}_{\bar{\vartheta}_1 } \bar{c},({\cal r}_k)_{\bar{\vartheta}_1\vartheta_2}\,{\bf\pi}_{\vartheta_2}\,c \right>\end{aligned}\ ] ] which can be rewritten in terms of the component fields .
in fact , applying eqs .
( [ ap13 ] ) and ( [ ap7 ] ) and using the relation between the matrix operators @xmath779 and @xmath780 stated above , we end up with eq .
( [ 013 ] ) .
let us now consider the source terms in eq .
. decomposing the sources for the ghost fields according to @xmath781 with @xmath782 yields @xmath783 the decomposition of @xmath784 is more involved . in analogy with @xmath70 in eq .
( [ ap1 ] ) we decompose the source @xmath214 into its orthogonal parts : @xmath785 according to eqs .
( [ ap2 ] ) , ( [ 099 ] ) we write @xmath786 as @xmath787 with @xmath788 analogous to eq .
( [ ap2 ] ) , the vector @xmath789 with @xmath790 is unique up to ckv s . according to eq .
( [ 08 ] ) the inner product @xmath784 may now be decomposed as follows : @xmath791 it is important to note that , for arbitrary nonvanishing @xmath77 , @xmath160 , @xmath161 , @xmath119 , we have @xmath792 this implies that the operators @xmath793 , @xmath744 , @xmath745 , @xmath751 are invertible if their action is restricted to fields of the type @xmath77 , @xmath160 , @xmath161 , @xmath119 , respectively .
therefore the inner product of eq .
( [ new ] ) may be written as @xmath794 here we used eq .
( [ ap13 ] ) and introduced the source components @xmath795 then combining the results of eqs .
( [ ap18 ] ) and ( [ ap19 ] ) we eventually arrive at eq .
( [ 071 ] ) .
in this section we present several rather lengthy calculations needed for the discussion of the einstein - hilbert truncation in section [ s4 ] . in the following , all calculations are performed with @xmath62 where @xmath72 is assumed to correspond to a spherical background and the bars are omitted from the metric , the curvature and the operators .
[ eval ] in section [ s4 ] we derived explicit expressions for the kinetic operators @xmath798 and @xmath799 .
they may be represented as matrix differential operators acting on the column vectors @xmath800 and @xmath801 , respectively . in this representation
they take the form @xmath802= \left(\begin{array}{cccc } \left(\widetilde{\gamma}_k^{(2)}[g , g]\right)_{\bar{h}^t\bar{h}^t } & 0 & 0 & 0_{1\times 2}\\0 & \left(\widetilde{\gamma}_k^{(2)}[g , g]\right)_{\bar{\xi } \bar{\xi } } & 0 & 0_{1\times 2}\\ 0 & 0 & \left(\widetilde{\gamma}_k^{(2)}[g , g]\right)_{\bar{\phi}_0 \bar{\phi}_0 } & 0_{1\times 2}\\ 0_{2\times 1 } & 0_{2\times 1 } & 0_{2\times 1 } & { \cal q}_k \end{array}\right)\end{aligned}\ ] ] and @xmath803= \left(\begin{array}{cccc } 0 & \left(\widetilde{s}_{\rm gh}^{(2)}[g , g]\right)_{\bar{v}^tv^t } & 0 & 0\\ \left(\widetilde{s}_{\rm gh}^{(2)}[g , g]\right)_{v^t\bar{v}^t } & 0 & 0 & 0\\ 0 & 0 & 0 & \left(\widetilde{s}_{\rm gh}^{(2)}[g , g]\right)_{\bar{\varrho } \varrho}\\ 0 & 0 & \left(\widetilde{s}_{\rm gh}^{(2)}[g , g]\right)_{\varrho\bar{\varrho } } & 0 \end{array}\right ) \end{aligned}\ ] ] where @xmath804\right)_{\bar{\sigma } \bar{\sigma } } & \left(\widetilde{\gamma}_k^{(2)}[g , g ] \right)_{\bar{\phi}_1\bar{\sigma}}\\ \left(\widetilde{\gamma}_k^{(2)}[g , g]\right)_{\bar{\phi}_1\bar{\sigma } } & \left(\widetilde{\gamma}_k^{(2)}[g , g]\right)_{\bar{\phi}\bar{\phi } } \end{array}\right)\;.\end{aligned}\ ] ] the entries of these matrices are given in eq .
( [ 50 ] ) . on the rhs of the flow equation ( [ 032 ] ) these operators appear in terms of their inverses , which are determined in the following . at this point
it is important to note that , because of the maximally symmetric background , all covariant derivatives contained in the operators ( [ 83 ] ) and ( [ 84 ] ) appear as covariant laplacians and that the various entries are @xmath805-independent otherwise .
this implies that these entries are _ commuting _ differential operators which allows for particularly simple manipulations .
therefore it is not difficult to verify that the inverse operators assume the form @xmath806\right)^{-1}= \left(\begin{array}{cccc } \left[\left(\widetilde{\gamma}_k^{(2)}[g , g]\right)_{\bar{h}^t\bar{h}^t } \right]^{-1 } & 0 & 0 & 0_{1\times 2}\\ 0 & \left[\left(\widetilde{\gamma}_k^{(2)}[g , g]\right)_{\bar{\xi}\bar{\xi } } \right]^{-1 } & 0 & 0_{1\times 2}\\ 0 & 0 & \left[\left(\widetilde{\gamma}_k^{(2)}[g , g]\right)_{\bar{\phi}_0 \bar{\phi}_0}\right]^{-1 } & 0_{1\times 2}\\ 0_{2\times 1 } & 0_{2\times 1 } & 0_{2\times 1 } & { \cal q}_k^{-1 } \end{array}\right)\end{aligned}\ ] ] and @xmath807\right)^{-1}=}\nonumber\\ & & \left(\begin{array}{cccc } 0 & \left[\left(\widetilde{s}_{\rm gh}^{(2)}[g , g]\right)_{\bar{v}^tv^t } \right]^{-1 } & 0 & 0\\ \left[\left(\widetilde{s}_{\rm gh}^{(2)}[g , g]\right)_{v^t\bar{v}^t } \right]^{-1 } & 0 & 0 & 0\\ 0 & 0 & 0 & \left[\left(\widetilde{s}_{\rm gh}^{(2)}[g , g]\right ) _ { \bar{\varrho}\varrho}\right]^{-1}\\ 0 & 0 & \left[\left(\widetilde{s}_{\rm gh}^{(2)}[g , g]\right)_{\varrho \bar{\varrho}}\right]^{-1 } & 0 \end{array}\right ) \end{aligned}\ ] ] with @xmath808\right)_{\bar{\sigma } \bar{\sigma } } \left(\widetilde{\gamma}_k^{(2)}[g , g]\right)_{\bar{\phi}_1\bar{\phi}_1 } -\left(\widetilde{\gamma}_k^{(2)}[g , g]\right)^2_{\bar{\phi}_1\bar{\sigma } } \right]^{-1}\nonumber\\ & & \times\left(\begin{array}{cc } \left(\widetilde{\gamma}_k^{(2)}[g , g]\right)_{\bar{\phi}_1\bar{\phi}_1 } & -\left(\widetilde{\gamma}_k^{(2)}[g , g ] \right)_{\bar{\phi}_1\bar{\sigma}}\\ -\left(\widetilde{\gamma}_k^{(2)}[g , g]\right)_{\bar{\phi}_1\bar{\sigma } } & \left(\widetilde{\gamma}_k^{(2)}[g , g]\right)_{\bar{\sigma}\bar{\sigma } } \end{array}\right)\;.\end{aligned}\ ] ] inserting these expressions into the rhs of the flow equation ( [ 032 ] ) leads to @xmath809 + { \cal r}_k[g]\right)_{\zeta\zeta}\right]^{-1 } \partial_t\left({\cal r}_k[g]\right)_{\zeta\zeta}\right ] } \nonumber\\ & & + \frac{1}{2}{\rm tr}'\left [ \left\{\left(\widetilde{\gamma}_k^{(2)}[g , g]\right)_{\bar{\sigma}\bar{\sigma } } \left(\widetilde{\gamma}_k^{(2)}[g , g]\right)_{\bar{\phi}_1\bar{\phi}_1 } -\left(\widetilde{\gamma}_k^{(2)}[g , g]\right)^2_{\bar{\phi}_1\bar{\sigma } } \right\}^{-1}\right.\nonumber\\ & & \times\left\{\left(\gamma_k^{(2)}[g , g ] + { \cal r}_k[g]\right)_{\bar{\sigma}\bar{\sigma } } \partial_t\left({\cal r}_k[g]\right)_{\bar{\phi}_1\bar{\phi}_1 } + \left(\gamma_k^{(2)}[g , g]+{\cal r}_k[g]\right)_{\bar{\phi}_1\bar{\phi}_1 } \partial_t\left({\cal r}_k[g]\right)_{\bar{\sigma } \bar{\sigma}}\right.\nonumber\\ & & \left.\left.-2\left(\gamma_k^{(2)}[g , g]+{\cal r}_k[g ] \right)_{\bar{\phi}_1\bar{\sigma } } \partial_t\left({\cal r}_k[g]\right)_{\bar{\phi}_1\bar{\sigma } } \right\}\right]\nonumber\\ & & -{\rm tr}'\left[\sum\limits_{\psi\in\{v^t,\varrho\ } } \left[\left(s_{\rm gh}^{(2)}[g , g ] + { \cal r}_k\right)_{\bar{\psi}\psi}\right]^{-1 } \partial_t\left({\cal r}_k\right)_{\bar{\psi}\psi}\right]\end{aligned}\ ] ] where we used the relations @xmath810 the trace of the @xmath363-term appearing in eq . ( [ 061 ] ) may be easily evaluated since only the scalar eigenmodes @xmath811 and @xmath812 contribute .
we obtain @xmath813 + { \cal r}_k[g]\right)_{\phi_0\phi_0}\right]^{-1 } \partial_t\left({\cal r}_k[g]\right)_{\phi_0\phi_0}\right]}\nonumber\\ & = & \frac{1}{2z_{nk}}\sum\limits_{l=0}^{1}\sum\limits_{m=1}^{d_l(d,0 ) } \int d^dx\,\sqrt{g(x)}\,t^{lm}(x)\left[p_k + a_{s1}(d,\alpha ) r+b_{s1}(d,\alpha)\bar{\lambda}_k\right]^{-1}\nonumber\\ & & \times\partial_t\left[z_{nk}k^2r^{(0)}(-d^2/k^2)\right ] t^{lm}(x)\nonumber\\ & = & \frac{1}{2z_{nk}}\sum\limits_{l=0}^{1}\left[d_l(d,0)\frac{\partial_t \left[z_{nk } k^2r^{(0)}(\lambda_l(d,0)/k^2)\right]}{\lambda_l(d,0)+k^2r^{(0 ) } ( \lambda_l(d,0)/k^2)+a_{s1}(d,\alpha)r+b_{s1}(d,\alpha)\bar{\lambda}_k}\right ] \;.\end{aligned}\ ] ] here @xmath814 is the eigenvalue with respect to @xmath815 corresponding to @xmath351 . inserting also the remaining operators given in eq .
( [ 50 ] ) into eq .
( [ 061 ] ) finally leads to eq .
( [ 52 ] ) .
in this part of the appendix we evaluate the traces appearing in eq .
( [ 55 ] ) by applying the asymptotic heat kernel expansion . in its original form
it has often been used to compute traces of operators acting on unconstrained fields , see e.g. refs .
for our purposes we need the corresponding expansions for operators acting on constrained fields , i.e. fields satisfying appropriate transversality conditions . in appendix
[ heat ] these expansions are derived in detail for laplacians @xmath76 acting on symmetric transverse traceless tensors , on transverse vectors and on scalars , with the following results : @xmath816 & = & \left(\frac{{\rm i}}{4\pi ( s+{\rm i}\varepsilon)}\right)^{d/2}\int d^dx\ , \sqrt{g}\left\ { \frac{1}{2}(d-2)(d+1)\right.\nonumber\\ & & \left.-\frac{(d+1)(d+2)(d-5 + 3\delta_{d,2})}{12(d-1 ) } ( { \rm i } s-\varepsilon ) r+{\cal o}(r^2)\right\}\;,\end{aligned}\ ] ] @xmath817 & = & \left(\frac{{\rm i}}{4\pi ( s+{\rm i}\varepsilon)}\right)^{d/2}\int d^dx\ , \sqrt{g}\bigg\{d-1\nonumber\\ & & -\frac{(d+2)(d-3)+6\delta_{d,2}}{6d}({\rm i}s-\varepsilon ) r + { \cal o}(r^2)\bigg\}\;,\end{aligned}\ ] ] @xmath818 = \left(\frac{{\rm i}}{4\pi ( s+{\rm i}\varepsilon)}\right)^{d/2}\int d^dx\ , \sqrt{g}\left\ { 1-\frac{1}{6}({\rm i } s-\varepsilon ) r+{\cal o}(r^2)\right\}\;.\end{aligned}\ ] ] the next step is to consider an arbitrary function @xmath819 with a fourier transform @xmath820 . for such functions @xmath12 , we may express the trace of the operator @xmath821 that results from replacing the argument of @xmath12 with @xmath815 in terms of @xmath820 : @xmath822=\lim\limits_{\varepsilon\searrow 0 } \int\limits^{\infty}_{-\infty}ds\,\widetilde{w}(s)\,{\rm tr } \left[e^{-({\rm i } s-\varepsilon)d^2}\right]\;.\end{aligned}\ ] ] we obtain the asymptotic expansion of @xmath823 $ ] by inserting the heat kernel expansion for @xmath824 $ ] into eq .
( [ 56 ] ) . for laplacians acting on the constrained fields considered here they read as follows : @xmath825&=&(4\pi)^{-d/2}\bigg\ { \frac{1}{2}(d-2)(d+1)\,q_{d/2}[w]\ , \int d^dx\,\sqrt{g}\nonumber\\ & & + \frac{(d+1)(d+2)(d-5 + 3\delta_{d,2})}{12(d-1 ) } \,q_{d/2 - 1}[w]\,\int d^dx\,\sqrt{g}r\nonumber\\ & & + { \cal o}(r^{<d-2})\bigg\}\;,\end{aligned}\ ] ] @xmath826&=&(4\pi)^{-d/2}\bigg\ { ( d-1)\,q_{d/2}[w]\ , \int d^dx\,\sqrt{g}\nonumber\\ & & + \frac{(d+2)(d-3)+6\delta_{d,2}}{6d } \,q_{d/2 - 1}[w]\,\int d^dx\,\sqrt{g}r+{\cal o}(r^{<d-2})\bigg\}\;,\end{aligned}\ ] ] @xmath827&=&(4\pi)^{-d/2}\bigg\{q_{d/2}[w]\ , \int d^dx\,\sqrt{g}\nonumber\\ & & + \frac{1}{6}\,q_{d/2 - 1}[w]\,\int d^dx\,\sqrt{g}r+{\cal o}(r^{<d-2 } ) \bigg\}\;.\end{aligned}\ ] ] here the set of functionals
@xmath828 $ ] is defined as @xmath829\equiv\lim\limits_{\varepsilon\searrow 0}\int \limits^{\infty}_{-\infty } ds\,(-{\rm i } s+\varepsilon)^{-n}\ , \widetilde{w}(s)\;.\end{aligned}\ ] ] by virtue of the mellin transformation we may now reexpress @xmath830 in terms of @xmath12 so that @xmath831&=&\frac{1}{\gamma(n)}\int\limits_{0}^{\infty}dz\,z^{n-1}\,w(z)\;,\;\ ; n>0\;;\nonumber\\ q_n[w]&=&\frac{(-1)^m}{\gamma(m+n)}\int\limits_{0}^{\infty}dz\,z^{m+n-1}\ , \frac{d^m w(z)}{dz^m}\;,\;\,n\le 0\;,\;\,m>-n\;,\;\,m\in @xmath832=w(0)$ ] .
let us now consider the case where isolated eigenvalues have to be excluded from @xmath823 $ ] . according to appendix [ heat ]
such traces can be expressed as the difference between the complete trace @xmath823 $ ] and a term of the form @xmath833 @xmath834 . here
@xmath835 refer to the modes to be omitted and @xmath836 and @xmath837 denote the corresponding eigenvalues of @xmath815 and their degrees of degeneracy , respectively .
since @xmath838 we may view @xmath839 as a function of @xmath404 .
as outlined in section [ s4 ] such a function contributes to the evolution of @xmath343 and @xmath34 only for @xmath419 , with the contribution given by @xmath840 . using the explicit expressions for @xmath837 ( see table 1 in appendix [ harm ] ) and applying eq .
( [ 035 ] ) we therefore obtain for the traces relevant to the flow equation @xmath841&=&{\rm tr}_{(1t)}[w(-d^2 ) ] -\frac{3\delta_{d,2}}{8\pi}\,w(0)\,\int d^2x\,\sqrt{g}r+{\cal o}(r^{<d-2})\ ; , \nonumber\\ { \rm tr}_{(0)}''[w(-d^2)]&=&{\rm tr}_{(0)}[w(-d^2 ) ] -\frac{\delta_{d,2}}{2\pi}\,w(0)\,\int d^2x\,\sqrt{g}r+{\cal o}(r^{<d-2})\ ; , \nonumber\\ { \rm tr}_{(0)}'[w(-d^2)]&=&{\rm tr}_{(0)}[w(-d^2 ) ] -\frac{\delta_{d,2}}{8\pi}\,w(0)\,\int d^2x\,\sqrt{g}r+{\cal o}(r^{<d-2})\;,\end{aligned}\ ] ] where the primes have to be interpreted as in section [ s4 ] .
the next step is to insert the expansions of the traces into @xmath397 , eq .
( [ 55 ] ) , and to compare the coefficients of the operators @xmath30 and @xmath31 with those on the lhs , eq .
( [ 040 ] ) .
this leads to the following differential equations : @xmath842\nonumber\\ & & + d\,q_{d/2}\left[\left({\cal a}_0 - 2\alpha\bar{\lambda}_k\right)^{-1 } { \cal n}\right]-2d\,q_{d/2}\left[{\cal a}_0^{-1}{\cal n}_0\right]\bigg\ } \;,\end{aligned}\ ] ] @xmath843\nonumber\\ & & + c_2(d)\,q_{d/2 - 1}\left[\left({\cal a}_0 -2\alpha\bar{\lambda}_k\right)^{-1}{\cal n}\right ] + c_3(d)\,q_{d/2}\left[\left({\cal a}_0 - 2\bar{\lambda}_k \right)^{-2}{\cal n}\right]\nonumber\\ & & + c_4(d,\alpha)\,q_{d/2}\left[\left({\cal a}_0 - 2\alpha \bar{\lambda}_k\right)^{-2}{\cal n}\right ] -2c_2(d)\,q_{d/2 - 1 } \left[{\cal a}_0^{-1}{\cal n}_0\right]\nonumber\\ & & + c_5(d)\,q_{d/2}\left[{\cal a}_0^{-2}{\cal n}_0\right ] + 3\delta_{d,2}\left[\frac{\partial_t(z_{nk}k^2)}{2z_{nk}(k^2 -2\bar{\lambda}_k)}-\frac{\partial_t(z_{nk}k^2)}{2z_{nk}(k^2 -2\alpha\bar{\lambda}_k)}\right]\bigg\}\;.\end{aligned}\ ] ] here the coefficients @xmath429 are defined as in eq .
( [ 044 ] ) . in eqs .
( [ 66 ] ) , ( [ 67 ] ) , the various @xmath830 may now be expressed in terms of the cutoff - dependent threshold functions @xmath426 and @xmath427 introduced in eq .
( [ 68 ] ) . using the relations @xmath844 & = & k^{2(n - p+1)}\phi^p_n(c / k^2)-\frac{1}{2}\eta_n(k)\,k^{2(n - p+1 ) } \widetilde{\phi}^p_n(c / k^2)\nonumber\\ q_n\left[\left({\cal a}_0+c\right)^{-p}{\cal n}_0\right ] & = & k^{2(n - p+1)}\phi^p_n(c / k^2)\end{aligned}\ ] ] we arrive at the differential equations ( [ 74 ] ) and ( [ 75 ] ) . at this stage the following point should be mentioned . in order to achieve that the integrals in eq .
( [ 63 ] ) actually converge we have to demand that @xmath296 rapidly decreases as @xmath845
. however , since from now on its form for @xmath846 does not play a role any more we identify @xmath296 with its part for nonnegative arguments and assume that @xmath296 is a smooth function defined only for @xmath847 and endowed with the properties stated in subsection [ 3b ] .
[ d=4 ] in this section we compare our results to those derived in ref .
@xcite for the @xmath41 dimensional case .
inserting @xmath41 into eq .
( [ 52 ] ) we obtain for the rhs @xmath848 of the evolution equation @xmath849 + \frac{1}{2}{\rm tr}_{(1t)}'\left[\frac{\partial_t p_k } { p_k+\frac{2\alpha-1}{4}r-2\alpha\bar{\lambda}_k}\right ] \nonumber\\ & & + \frac{1}{2}{\rm tr}_{(0)}''\left[\frac{\partial_t p_k}{p_k+\frac{\alpha-1 } { 2}r-2\alpha\bar{\lambda}_k}\right ] + \frac{1}{2}{\rm tr}_{(0)}''\left[\frac{\partial_t p_k}{p_k -2\bar{\lambda}_k}\right]\nonumber\\ & & + \frac{1}{2}\sum\limits_{l=0}^{1}\left[d_l(4,0)\frac{\partial_t p_k ( \lambda_l(4,0))}{p_k(\lambda_l(4,0 ) ) -\frac{4\alpha}{3\alpha-1}\bar{\lambda}_k}\right ] -{\rm tr}_{(1t)}\left[\frac{\partial_t p_k } { p_k-\frac{r}{4}}\right ]
-{\rm tr}_{(0)}'\left[\frac{\partial_t p_k } { p_k-\frac{r}{2}}\right]\nonumber\\ & & + \frac{\partial_t z_{nk}}{z_{nk}}\bigg\{\frac{1}{2}{\rm tr}_{(2st^2)}\left [ \frac{p_k+d^2}{p_k+\frac{2}{3}r-2\bar{\lambda}_k}\right ] + \frac{1}{2}{\rm tr}_{(1t)}'\left[\frac{p_k+d^2 } { p_k+\frac{2\alpha-1}{4}r-2\alpha\bar{\lambda}_k}\right ] \nonumber\\ & & + \frac{1}{2}\sum\limits_{l=0}^{1}\left[d_l(4,0)\frac{p_k(\lambda_l(4,0 ) ) -\lambda_l(4,0)}{p_k(\lambda_l(4,0 ) ) -\frac{4\alpha}{3\alpha-1}\bar{\lambda}_k}\right]\nonumber\\ & & -\frac{1}{4\alpha}{\rm tr}_{(0)}''\bigg [ \bigg\{\left[p_k-2\bar{\lambda}_k\right ] \left[p_k+\frac{\alpha-1}{2}r-2\alpha\bar{\lambda}_k \right]\bigg\}^{-1}\nonumber\\ & & \times\bigg\{\left((1 - 3\alpha)\left[(3-\alpha)p_k + \frac{\alpha-1}{2}r \right]+4\alpha(\alpha+1)\bar{\lambda}_k\right)(p_k+d^2)\nonumber\\ & & -3(1-\alpha)^2\sqrt{p_k}\sqrt{p_k-\frac{r}{3}}\,\left [ \sqrt{p_k}\sqrt{p_k-\frac{r}{3}}-\sqrt{-d^2}\sqrt{-d^2-\frac{r}{3}}\right ] \bigg\}\bigg]\bigg\}\end{aligned}\ ] ] with @xmath850
. our result ( [ 105 ] ) agrees with eq .
( 3.22 ) of ref .
@xcite , up to a few ( typographical ) errors occuring in eq .
( 3.22 ) . to be more precise , the prime at the @xmath851-term in lines 1 and 4 of eq . (
[ 105 ] ) , the factor @xmath852 appearing in the first term of lines 3 and 5 , and the factor @xmath853 contained in the term @xmath854 in line 7 are mistakenly left out in the corresponding equation of ref .
@xcite .
expanding the flow equations for @xmath855 and @xmath343 , eqs .
( [ 74 ] ) and ( [ 75 ] ) , with respect to @xmath34 , using the relation @xmath856 and setting , as in @xcite , @xmath857 leads to @xmath858\right\}+{\cal o}(\bar{\lambda}_k)\;,\end{aligned}\ ] ] @xmath859 + \bar{\lambda}_k k^2\left[-\frac{13}{24}q^1_1\right.\right.\nonumber\\ & & + \left.\left.\left(\frac{17}{24}+\alpha\right)q^2_2 + \eta_k\left(-\frac{1}{8}\widetilde{q}^1_1+\left(\frac{47}{24 } + \alpha\right)\widetilde{q}_2 ^ 2\right)\right]\right\ } + { \cal o}(\bar{\lambda}_k^2)\;.\end{aligned}\ ] ] in this form the flow equations for the couplings are suitable for a comparison with the corresponding results in eqs . (
4.6)-(4.9 ) of ref .
apart from the contributions from matter fields , which are also considered there , the results differ by a factor @xmath860 in front of the @xmath861-terms .
presumably , this deviation can be explained by a wrong sign introduced in @xcite for a certain term .
this term is a contribution from the ghosts produced by heat kernel expanding the last two terms in line 3 of eq .
( [ 105 ] ) , and it carries the prefactor @xmath862 .
we may conclude that , apart from these corrections , our results agree with those in @xcite , which is as it should be since the same cutoff is used .
[ harm ] in this section we introduce the spherical harmonics @xmath349 , @xmath350 and @xmath351 for symmetric transverse traceless @xmath863 tensors @xmath77 , transverse @xmath864 vectors @xmath153 , and scalars @xmath80 on a @xmath113-dimensional spherical background @xmath63 .
these harmonics form complete sets of orthogonal eigenfunctions with respect to the covariant laplacians acting on @xmath757 tensors , @xmath758 vectors and scalars , i.e. they satisfy @xmath865 and , after proper normalization , @xmath866 here @xmath867 and @xmath868 are the unit matrices in the spaces of @xmath757 tensors and transverse vectors , respectively . in eq .
( [ 094 ] ) the @xmath836 s denote the eigenvalues with respect to @xmath176 where @xmath319 refers to the spin of the field under consideration and @xmath869 takes the values @xmath870 .
furthermore , the second upper index at @xmath349 , @xmath350 and @xmath351 , @xmath871 , takes the degeneracy of the eigenvalues into account .
it assumes values from one to @xmath837 with @xmath837 the degree of degeneracy . in ref .
@xcite explicit expressions for @xmath836 and @xmath837 are derived which can be found in table 1 .
the eigenvalues are expressed in terms of the curvature scalar @xmath872 of the sphere with radius @xmath64 . in ref .
@xcite it is also shown that the spherical harmonics @xmath349 , @xmath350 and @xmath351 span the spaces of @xmath757 tensors , @xmath758 vectors and scalars so that we may expand arbitrary functions @xmath77 , @xmath153 and @xmath80 according to @xmath873 here the coefficients @xmath874 , @xmath875 and @xmath876 are countably infinite sets of constants that are uniquely determined by @xmath77 , @xmath153 and @xmath80 .
( [ 067 ] ) may now be used to expand also any symmetric non-@xmath877 tensor and nontransverse vector in terms of spherical harmonics since they may be expressed in terms of @xmath757 tensors , @xmath758 vectors and scalars by using the decompositions ( [ f ] ) , ( [ p ] ) , see e.g. @xcite . in this context
it is important to note that the @xmath878 modes @xmath879 and the @xmath880 modes @xmath881 satisfy the killing equation ( [ 09 ] ) and the scalar equation ( [ 010 ] ) , respectively , and that @xmath882 .
as discussed in subsection [ 2b ] , arbitrary symmetric rank 2 tensors receive no contribution from these modes . in the case of arbitrary vectors
it is the constant scalar mode that does not contribute .
such modes have no physical meaning and have to be omitted therefore .
eigenfunction & spin @xmath319 & eigenvalue @xmath836 & degeneracy @xmath837 & + @xmath883 & 2 & @xmath884 & @xmath885 & @xmath886 + @xmath887 & 1 & @xmath888 & @xmath889 & @xmath890 + @xmath891 & 0 & @xmath892 & @xmath893 & @xmath894 +
[ heat ] in this part of the appendix we supply the tools necessary for the evaluation of functional traces and we derive the heat kernel expansions for laplacians acting on differentially constrained fields . as a first step we consider a functional trace of the form @xmath895)}[f(-d^2)]&=&\int d^dx\,\sqrt{g}\,{\left < x\left|f(-d^2 ) \right|x\right>_{\mu_1\ldots\mu_s}}^{\mu_1\ldots\mu_s}\nonumber\\ & = & \int d^dx\,\left[\left(f(-d^2)\right)^{\mu_1\ldots\mu_s\nu_1\ldots\nu_s } \,\left({\bf 1}_{(s[c])}\right)_{\nu_1\ldots\nu_s\mu_1\ldots\mu_s}\delta^d(x - y ) \right]_{x = y}\end{aligned}\ ] ] here @xmath433 is an arbitrary smooth function whose argument is replaced with the covariant laplacian defined on the space of spin-@xmath319 fields with a possible symmetry and/or transversality constraint @xmath123 , as indicated by the subscript @xmath896)$ ] at the trace .
note that @xmath433 inherits the matrix structure from the corresponding laplacian .
furthermore , @xmath897)}$ ] denotes the unit matrix in the space of independent field components .
given a closed riemannian manifold @xmath898 we now assume that @xmath899 is a complete set of orthonormal functions on @xmath898 spanning the space of fields under consideration . then by making use of the completeness relation @xmath900)}\right)_{\nu_1\ldots\nu_s\mu_1\ldots\mu_s}\ , \frac{\delta^d(x - y)}{\sqrt{g(x)}}=\frac{1}{2 } \sum\limits_k\left(u^k_{\nu_1\ldots\nu_s}(x)\,u^k_{\mu_1\ldots\mu_s}(y ) + u^k_{\mu_1\ldots\mu_s}(x)\,u^k_{\nu_1\ldots\nu_s}(y)\right)\end{aligned}\ ] ] eq .
( [ hk14 ] ) can be written as @xmath901)}[f(-d^2)]=\int d^dx\,\sqrt{g}\,\sum\limits_k u^k_{\mu_1\ldots\mu_s}(x)\left(f(-d^2)\right)^{\mu_1\ldots\mu_s\nu_1\ldots \nu_s}u^k_{\nu_1\ldots\nu_s}(x)\;.\end{aligned}\ ] ] clearly if @xmath899 is taken to be a complete set of orthonormal eigenfunctions with respect to the covariant laplacian such that @xmath902 eq .
( [ hk20 ] ) boils down to @xmath903)}[f(-d^2)]=\sum\limits_{k}f\left(\widetilde{\lambda}_k \right)\;.\end{aligned}\ ] ] in general the evaluation of such traces is a formidable task and one has to resort to approximations .
the most familiar such approximation is the `` early time '' expansion of the diagonal heat kernel @xmath904 for @xmath905 .
( we write @xmath906 with @xmath907 . )
it has been discussed in many references @xcite : @xmath908 here @xmath909 is an arbitrary smooth matrix potential and the @xmath910 s are tensor polynomials proportional to the @xmath911-th power of the curvature and endowed with the same matrix structure as @xmath912 .
they depend on the space of fields under consideration . for operators @xmath912 acting on unconstrained fields , i.e. fields with independent components
, the first three coefficients take the form and @xmath913 .
however , for truncations containing invariants quadratic in the curvature we shall also need @xmath914 , see @xcite . ]
@xmath915 here @xmath916 is the unit matrix in the space of field components , and @xmath917 furthermore , @xmath918 is the curvature operator defined as the commutator @xmath919\equiv d_\mu d_\nu - d_\nu d_\mu\;.\end{aligned}\ ] ] inserting the asymptotic expansion ( [ hk22 ] ) into eq .
( [ hk14 ] ) we obtain for the functional trace of the heat kernel @xmath920=\left(\frac{{\rm i } } { 4\pi { \sf t } } \right)^{\frac{d}{2}}\!\int d^dx\,\sqrt{g}\left\{{\rm tr}\,a_0-{\rm i}{\sf t}\ , { \rm tr}\,a_2(x;q)-{\sf t}^2\,{\rm tr}\,a_4(x;q)+{\cal o}\left({\sf t}^3\right ) \right\}\end{aligned}\ ] ] where @xmath921 is the matrix trace with respect to the tensor or spinor indices .
let us have a closer look at the coefficients @xmath910 . for arbitrary scalars
@xmath80 , @xmath918 vanishes since @xmath922 . in the case of arbitrary spin-@xmath319 fields @xmath923 with integer spin @xmath595 we obtain @xmath924 hence , for scalars , @xmath925 receives no contribution from @xmath926 , while for fields with nonzero spin it amounts to a nonvanishing contribution with a @xmath927-matrix structure . for arbitrary vectors and rank-2 tensors
we find , respectively , @xmath928 from now on we restrict our considerations to matrix potentials @xmath909 of the form @xmath929 with @xmath205 a real constant , and we assume that the metric corresponds to a maximally symmetric background @xmath63 . setting @xmath930 we obtain @xmath931 and @xmath932 , independently of the spin @xmath319 of the field . here
the dependence on @xmath319 is totally encrypted in the unit matrix @xmath933 .
this is not the case for the coefficient @xmath925 .
it is given by @xmath934\ , { \bf 1}_{(0)}\,r^2\;,\end{aligned}\ ] ] @xmath935_{\mu\nu}=\frac{1}{360}\left[\frac{5d^3 - 7d^2 + 6d-60 } { d^2(d-1)}+60q+180q^2\right]\left({\bf 1}_{(1)}\right)_{\mu\nu}\,r^2\;,\end{aligned}\ ] ] @xmath936_{\mu\nu\rho\sigma } & = & \frac{1}{360}\left[\frac{5d^3 - 7d^2 + 6d-120}{d^2(d-1)}+60q+180q^2\right ] \left({\bf 1}_{(2)}\right)_{\mu\nu\rho\sigma}\,r^2\nonumber\\ & & + \frac{1}{3d^2(d-1)^2}\left(g_{\mu\nu}g_{\rho\sigma}-g_{\mu\sigma } g_{\nu\rho}\right)r^2\end{aligned}\ ] ] for scalars , vectors and rank-2 tensors , respectively , with @xmath937 , @xmath938 and @xmath939 .
up to this point we considered only unconstrained fields . for fields subject to constraints like transverse vectors @xmath153 and @xmath757 tensors @xmath77 we can not directly apply eqs .
( [ hk4 ] ) or ( [ hk10 ] ) , ( [ hk11 ] ) .
however , the heat kernel coefficients for @xmath153 and @xmath77 can be computed from those of the unconstrained fields using the decompositions ( [ p ] ) , ( [ f ] ) for arbitrary vectors @xmath726 and arbitrary symmetric tensors @xmath70 . from appendix [ harm ] we can infer that the sets of orthonormal @xmath815-eigenfunctions @xmath940 and @xmath941 span the spaces of _ all _ vectors @xmath726 and of _ all _ symmetric tensors @xmath70 , respectively . here
the @xmath351 s are the normalized spherical harmonics of eq .
( [ 093 ] ) .
now we insert these eigenfunctions into the trace formula ( [ hk20 ] ) with @xmath433 taken to be an exponential .
then we use the commutation relations of appendix [ commut ] in order to pull the @xmath74 s from the tt - decomposition through the exponentials and to combine them to laplacians .
this leads to the following decomposition for the traced heat kernels of the unconstrained vectors and symmetric tensors in terms of the heat kernels for the differentially constrained fields : @xmath942= { \rm tr}_{(1t)}\left[e^{-{\rm i}{\sf t}\left(d^2+q\,r\right)}\right ] + { \rm tr}_{(0)}\left[e^{-{\rm i}{\sf t}\left(d^2+\frac{dq+1}{d}r\right)}\right ] -e^{-{\rm i}{\sf t}\,\frac{dq+1}{d}r}\;,\end{aligned}\ ] ] @xmath943&= & { \rm tr}_{(2st^2)}\left[e^{-{\rm i}{\sf t}(d^2+qr)}\right ] + { \rm tr}_{(1t)}\left[e^{-{\rm i}{\sf t}\left(d^2+\left(\frac{d+1}{d(d-1 ) } + q\right)r\right)}\right]\nonumber\\ & & + { \rm tr}_{(0)}\left[e^{-{\rm i}{\sf t}\left(d^2+\left(\frac{2}{d-1 } + q\right)r\right)}\right]+{\rm tr}_{(0)}\left[e^{-{\rm i}{\sf t}(d^2+qr ) } \right ] -e^{-{\rm i}{\sf t}\left(\frac{2}{d-1}+q\right)r}\nonumber\\ & & -(d+1)\,e^{-{\rm i}{\sf t}\left(\frac{1}{d-1}+q\right)r } -\frac{d(d+1)}{2}\,e^{-{\rm i}{\sf t}\left(\frac{2}{d(d-1)}+q\right)r}\;.\end{aligned}\ ] ] the last term of eq .
( [ hk2 ] ) and the last three terms of eq .
( [ hk1 ] ) arise from those spherical harmonics @xmath351 and @xmath350 which are not contained in the sets of eigenfunctions ( [ hk23 ] ) and ( [ hk21 ] ) . to be more precise , the last term in eq .
( [ hk2 ] ) comes from the constant eigenmode @xmath944 of the operator @xmath945 .
furthermore , the last but second and the last but first term in eq .
( [ hk1 ] ) take account of the eigenmodes @xmath946 and @xmath881 of the operator @xmath947 , respectively .
as discussed in subsection [ 2b ] the @xmath881 s satisfy the scalar equation ( [ 010 ] ) and are therefore in a one - to - one correspondence with the pckv s of @xmath63 . the last term in eq .
( [ hk1 ] ) comes from the eigenmodes @xmath879 of the operator @xmath948 , which are the kv s of @xmath63 .
these subtraction terms compensate for the corresponding unphysical contributions contained in the _ complete _ traces for the constrained fields on the rhs of eqs .
( [ hk2 ] ) , ( [ hk1 ] ) .
this can be seen as follows .
consider the functional trace of eq .
( [ hk20 ] ) . omitting the contributions from the modes @xmath949 , we denote the functional trace involving only the remaining modes with @xmath950)}[f(-d^2)]$ ] .
( [ 096 ] ) implies the following relation between @xmath951)}[f(-d^2)]$ ] and the _ complete _ trace @xmath952 ) } [ f(-d^2)]$ ] : @xmath953)}[f(-d^2)]={\rm tr}_{(s[c])}[f(-d^2 ) ] -\sum\limits_{k\in\{k_1,\ldots , k_n\}}f\left(\widetilde{\lambda}_k\right)\;.\end{aligned}\ ] ] this rule indeed yields the last term in eq .
( [ hk2 ] ) and the last three terms in eq .
( [ hk1 ] ) . as the next step
we insert the asymptotic expansion ( [ hk3 ] ) into both sides of eqs .
( [ hk2 ] ) and ( [ hk1 ] ) and compare the coefficients of @xmath404 .
this leads to the following seeley coefficients for the constrained fields : @xmath954 @xmath955 the terms proportional to the @xmath956 s originate from the subtraction terms on the rhs of eqs .
( [ hk2 ] ) , ( [ hk1 ] ) which are due to the unphysical eigenmodes .
these terms have an expansion of the form @xmath957 , while the terms of the heat kernel expansion are of the form @xmath958 . comparing powers of @xmath959 , only under the condition @xmath960 a given term
@xmath961 contributes to @xmath962 .
hence for @xmath51 , @xmath871 fixed , the seeley coefficients @xmath963 for the differentially constrained fields receive a contribution from a term of the form @xmath964 at most for one specific value of the dimensionality @xmath113 . in particular , the subtraction terms in eqs .
( [ hk2 ] ) , ( [ hk1 ] ) do not contribute to @xmath965 , while @xmath966 and @xmath967 on the lhs of eqs .
( [ hk17 ] ) , ( [ hk18 ] ) receive contributions from terms of the form @xmath968 and @xmath969 , @xmath970 , respectively .
the matrix traces on the rhs of eqs .
( [ hk17 ] ) , ( [ hk18 ] ) can now be evaluated by using the heat kernel coefficients for the ( differentially ) unconstrained fields .
for scalars we have @xmath971 ( @xmath937 ) . for vectors
the traces are evaluated according to @xmath972_{\mu\nu}$ ] so that we obtain from eqs .
( [ hk4 ] ) , ( [ hk10 ] ) @xmath973\,r^2\;.\end{aligned}\ ] ] in order to determine @xmath963 for symmetric tensor fields we have to symmetrize the heat kernel for unconstrained rank-2 tensors according to @xmath974 before we can apply eq .
( [ hk3 ] ) with eqs .
( [ hk4 ] ) and ( [ hk11 ] ) .
this leads to @xmath975_{\mu\nu\rho\sigma}\right|_{(2s)}&= & \frac{1}{4}\big(\left.\left[a_n(q)\right]_{\mu\nu\rho\sigma}\right|_{(2 ) } + \left.\left[a_n(q)\right]_{\nu\mu\rho\sigma}\right|_{(2)}\nonumber\\ & & + \left.\left[a_n(q)\right]_{\mu\nu\sigma\rho}\right|_{(2 ) } + \left.\left[a_n(q)\right]_{\nu\mu\sigma\rho}\right|_{(2)}\big)\end{aligned}\ ] ] and in particular @xmath976 . for tensors
the matrix traces are computed according to @xmath977_{\mu\nu\rho\sigma}$ ] which yields @xmath978\,r^2\;.\end{aligned}\ ] ] finally we insert the matrix traces eqs .
( [ hk12 ] ) , ( [ hk16 ] ) and @xmath979 into eqs .
( [ hk17 ] ) , ( [ hk18 ] ) and determine the heat kernel coefficients for transverse vectors and @xmath757 tensors .
the results are summarized in table 2 .
let us add a final remark concerning the applicability of the asymptotic expansion ( [ hk22 ] ) . since it is valid only in the limit @xmath905
it is clear that it can not be integrated over @xmath980 or @xmath981 term by term , in general .
however , this is possible if the heat kernel is integrated against a `` test '' function which suppresses large values of @xmath319 or @xmath38 .
this is indeed the case for our application of the asymptotic expansion presented in subsection [ trace ] . ' '' '' field & @xmath982&@xmath983 & @xmath984 + [ 0.8ex]@xmath757 & & @xmath985 & ' '' '' @xmath986 + [ -0.8ex]tensor&[-2.2ex ] @xmath987 & + @xmath988 & ' '' '' @xmath989 + [ 0.8ex]@xmath758 & & [ 0.4ex]@xmath990 & ' '' '' @xmath991 + [ -0.8ex]vector & [ -2.2ex ] @xmath992 & [ -0.4ex]@xmath993 & @xmath994{0mm}{10mm}+q\,\frac{(d+2)(d-3)+6\,\delta_{d,2}}{6d}r^2+q^2\ , \frac{d-1}{2}r^2 $ ] + ' '' '' scalar & @xmath995 & @xmath996&@xmath997 +
[ coeff ] in this appendix we define the coefficient functions which appear in eqs .
( [ 44])-([55 ] ) of subsection [ 4c ] and in eq .
( [ 100 ] ) of appendix [ inv ] .
@xmath998 \left[2\alpha(d-1)-(d-2)\right]}{d^2\alpha}\;,\nonumber\\ & & f_{s2}(d,\alpha)\equiv\frac{4(d-2)(d-1)(\alpha-1)^2}{d^2\alpha}\end{aligned}\ ] ]
[ commut ] in the following we summarize the commutation relations which were used in order to derive eq .
( [ 44 ] ) of subsection [ 4c ] , and eqs .
( [ hk2 ] ) and ( [ hk1 ] ) of appendix [ heat ] .
they are valid for the class of maximally symmetric backgrounds .
@xmath999\exp\left(\bar{d}^2\right ) \left(\bar{d}^\mu\bar{d}^\nu -\frac{1}{d}\bar{g}^{\mu\nu}\bar{d}^2\right)\widehat{\sigma}\nonumber\\ & & \;\;\;\;\;=\frac{d-1}{d}\widehat{\sigma}\bar{d}^2\left(\bar{d}^2 + \frac{\bar{r}}{d-1}\right)\exp\left(\bar{d}^2+\frac{2\bar{r}}{d-1}\right)\ , \widehat{\sigma}+{\rm cov.\;divergence}\end{aligned}\ ] ]
[ approx ] in the following we determine the approximate formula for the position of the non - gaussian fixed point discussed in subsection [ 5c ] . in a first approximation we set @xmath586 and determine @xmath490 from the condition @xmath1000 alone . solving this equation for @xmath490 leads to @xmath1001 which , for @xmath1002 , boils down to @xmath1003\bigg\}^{-1}\;.\end{aligned}\ ] ] here @xmath1004 are @xmath113- and @xmath314-dependent coefficients defined as @xmath1005 employing the exponential shape function ( [ h6 ] ) with @xmath545 , and setting @xmath41 and @xmath315 , for instance , eq .
( [ h1 ] ) yields @xmath588 .
here we used that for this shape function @xmath1006 , @xmath547 , @xmath1007 , @xmath1008 .
as a different approximation scheme , we determine @xmath44 from a set of taylor - expanded @xmath329-functions . using @xmath1009 we expand the @xmath329-functions ( [ 001 ] ) and ( [ 004 ] ) about @xmath1010 and obtain @xmath1011\lambda_k\,g_k\nonumber\\ & & + \frac{1}{2}d(d+1)(d-2)(4\pi)^{1-\frac{d}{2}}\omega_d\ , \phi^{1}_{d/2}(0)\,g_k^2+{\cal o}\left({\rm g}^3\right)\;,\nonumber\\ \mbox{\boldmath $ \beta$}_g(\lambda_k , g_k;\alpha , d)&=&(d-2)\,g_k-(d-2 ) \omega_d\,g_k^2+{\cal o}\left({\rm g}^3\right)\;.\end{aligned}\ ] ] here @xmath505 and @xmath518 are defined as in eqs .
( [ gfp3 ] ) and ( [ gfp13 ] ) , and @xmath1012 stands for terms of third and higher orders in the couplings @xmath1013 and @xmath1014 . now @xmath490 is obtained as the nontrivial solution to @xmath1015 , which reads @xmath1016 inserting eq .
( [ h4 ] ) into @xmath1017 and neglecting also the terms quadratic in the couplings the condition @xmath1018 leads to @xmath1019 using the shape function ( [ h6 ] ) with @xmath545 we obtain from eqs .
( [ h4 ] ) and ( [ h5 ] ) in @xmath41 dimensions @xmath1020 which yields @xmath1021 for @xmath315 , for instance .
99999999 for a review see : c. bagnuls , c. bervillier , hep - th/0002034 ; + t. r. morris , prog .
suppl . 131 ( 1998 ) 395 for a review see : j. berges , n. tetradis , c. wetterich , hep - ph/0005122 ; + c. wetterich , hep - ph/0101178 k. g. wilson , j. kogut , phys .
rept . 12 ( 1974 ) 75 ; + k. g. wilson , rev .
47 ( 1975 ) 773 j. polchinski , nucl .
b 231 ( 1984 ) 269 m. reuter , c. wetterich , nucl .
b 417 ( 1994 ) 181 and nucl . phys .
b 427 ( 1994 ) 291 ; for an earlier formulation see : m. reuter , c. wetterich , nucl .
b 391 ( 1993 ) 147 and nucl . phys .
b 408 ( 1993 ) 91 m. reuter , phys .
d 57 ( 1998 ) 971 and hep - th/9605030 for a brief introduction see : m. reuter in _ annual report 2000 of the international school in physics and mathematics , tbilisi , georgia _ and hep - th/0012069 l. f. abbott , nucl .
b 185 ( 1981 ) 189 ; + f. freire , d. f. litim , j. m. pawlowski , phys . lett .
b 495 ( 2000 ) 256 ; hep - th/0101108 s. l. adler , rev .
mod . phys . 54
( 1982 ) 729 c. wetterich , gen .
30 ( 1998 ) 159 j. f. donoghue , phys .
72 ( 1994 ) 2996 ; phys .
d 50 ( 1994 ) 3874 a. bonanno , m. reuter , phys .
d 60 ( 1999 ) 084011 and gr - qc/9811026 a. bonanno , m. reuter , phys . rev .
d 62 ( 2000 ) 043008 and hep - th/0002196 a. bonanno , m. reuter , hep - th/0106133 ; for an application of similar methods to the very late universe see also astro - ph/0106468 n. c. tsamis , r. p. woodard , phys .
b 301 ( 1993 ) 351 ; ann . phys .
( ny ) 238 ( 1995 ) 1 ; + i. antoniadis , e. mottola , phys .
d 45 ( 1992 ) 2013 ; + i. antoniadis , p. mazur , e. mottola , phys . lett .
b 444 ( 1998 ) 284 s. weinberg in _ general relativity , an einstein centenary survey , _
s. w. hawking , w. israel ( eds . ) , cambridge university press , 1979 w. souma , prog .
102 ( 1999 ) 181 for a recent discussion see : j .-
sumi , w. souma , k .-
aoki , h. terao , k. morikawa , hep - th/0002231 w. souma , gr - qc/0006008 j. w. york jr . , j. math .
( 1973 ) 456 d. dou , r. percacci , class .
( 1998 ) 3449 s. falkenberg , s. d. odintsov , int .
phys . a 13 ( 1998 )
607 l. n. granda , s. d. odintsov , grav .
cosmol . 4 ( 1998 ) 85 and phys .
b 409 ( 1997 ) 206 ; l. n. granda , europhys .
42 ( 1998 ) 487 ; a. a. bytsenko , l. n. granda , s. d. odintsov , jetp lett . 65 ( 1997 ) 600 o. lauscher , m. reuter , in preparation m. reuter , c. wetterich , nucl .
b 506 ( 1997 ) 483 and hep - th/9605039 a. h. chamseddine , m. reuter , nucl . phys .
b 317 ( 1989 ) 757 e. mottola , j. math . phys .
36 ( 1995 ) 2470 c. wetterich , phys .
b 301 ( 1993 ) 90 u. ellwanger , m. hirsch , a. weber , z. phys .
c 69 ( 1996 ) 687 d. f. litim , j. m. pawlowski , phys .
b 435 ( 1998 ) 181 ; hep - th/9901063 m. reuter , f. saueressig , in preparation r. gastmans , r. kallosh , c. truffin , nucl .
b 133 ( 1978 ) 417 ; + s. m. christensen , m. j. duff , phys .
79 b ( 1978 ) 213 h. kawai , m. ninomiya , nucl .
b 336 ( 1990 ) 115 i. jack , d. r. t. jones , nucl . phys .
b 358 ( 1991 ) 695 ; + h. kawai , y. kitazawa , m. ninomiya , nucl . phys .
b 404 ( 1993 ) 684 ; + t. aida , y. kitazawa , j. nishimura , a. tsuchiya , nucl . phys .
b 444 ( 1995 ) 353 r. t. seeley , proc . symp .
pure math . 10
( 1967 ) 288 ; + p. b. gilkey , j. diff .
( 1975 ) 601 ; + a. o. barvinsky , g. a. vilkovisky , phys .
( 1985 ) 1 ; + g. a. vilkovisky , cern - th-6392 - 92 ( 1992 ) m. a. rubin , c. r. ord@xmath1022ez , j. math . phys . 25 ( 1984 ) 2888 ; j. math . phys . 26 ( 1985 ) 65 b. allen , phys . rev .
d 34 ( 1986 ) 3670 e. s. fradkin , a. a. tseytlin , nucl . phys .
b 234 ( 1984 ) 472 t. r. taylor , g. veneziano , nucl . phys .
b 345 ( 1990 ) 210 | a new exact renormalization group equation for the effective average action of euclidean quantum gravity is constructed .
it is formulated in terms of the component fields appearing in the transverse - traceless decomposition of the metric .
it facilitates both the construction of an appropriate infrared cutoff and the projection of the renormalization group flow onto a large class of truncated parameter spaces .
the einstein - hilbert truncation is investigated in detail and the fixed point structure of the resulting flow is analyzed .
both a gaussian and a non - gaussian fixed point are found .
if the non - gaussian fixed point is present in the exact theory , quantum einstein gravity is likely to be renormalizable at the nonperturbative level . in order to assess the reliability of the truncation a comprehensive analysis of the scheme dependence of universal quantities
is performed .
we find strong evidence supporting the hypothesis that 4-dimensional einstein gravity is asymptotically safe , i.e. nonperturbatively renormalizable .
the renormalization group improvement of the graviton propagator suggests a kind of dimensional reduction from 4 to 2 dimensions when spacetime is probed at sub - planckian length scales .
hep - th/0108040 + mz - th/01 - 21 + ultraviolet fixed point and + generalized flow equation of + quantum gravity + _ institute of physics , university of mainz + staudingerweg 7 , d-55099 mainz , germany _ + |
based on combined 3-year data , the icecube collaboration has reported recently@xmath0 the discovery of a diffuse background of high energy ( he ) astrophysical neutrinos with energy above 60 tev that extends at least up to 3 pev , in excess ( 5.7@xmath1 significance level ) of that produced by cosmic ray ( cr ) interactions in the atmosphere .
the neutrino events did not point back to any identified sources and are consistent with an isotropic distribution of arrival directions .
their best fit power - law to the energy flux per flavor above 60 tev is @xmath2 so far , the origin of this diffuse background of high energy neutrinos is not known .
a natural source of high energy astrophysical neutrinos is the decay of mesons produced in collisions of cr nuclei with matter and radiation .
several models of cosmic ray production of neutrinos in / near cosmic ray sources@xmath3 have been used to estimate the astrophysical diffuse background of high energy neutrinos .
but , uncertain assumptions and free adjustable parameters prevented meaningful conclusions .
moreover , no evidence of neutrino emission from point - like or extended cr sources was found in four years of icecube data@xmath4 .
these include supernova remnants , the nearest most luminous gamma ray bursts ( grbs ) such as@xmath5 130427a at redshift @xmath6 , the giant elliptical galaxy m87 at a distance of 16 mpc that contains the nearest active galactic nucleus ( agn ) in the northern sky , and markarian 421 , at a distance of 133 mpc , the brightest @xmath7-ray blazar in the northern sky .
in particular , the null detection of high energy neutrinos with icecube from 100 stacked grbs has been used to set a strong limit on the contribution of grbs to the diffuse background of high energy neutrinos@xmath8 .
but , this flux limit was based on the assumption that the neutrino emission in grbs roughly coincides in time and beaming with the prompt gamma - ray and early afterglow emissions , as expected@xmath9 in the fireball models of grbs .
grb neutrinos , however , may be produced mainly through _ hadronic production of mesons by grb cosmic rays in the giant molecular clouds ( gmcs ) where most core colapse sne take place _ rather than through photo - production of mesons , which are assumed to take place in grb `` fireballs '' during their prompt gamma ray and early afterglow emissions@xmath9 .
such hadronic production of mesons , which is expected in the cannonball model of grbs@xmath10 , is delayed and spread over much larger beaming angles than the prompt gamma ray and early afterglow emissions ( the early time gev emission from grbs observed with fermi - lat , most probably , is not from @xmath11 decay but from inverse compton scattering by the high energy electrons of their self emitted synchrotron radiation@xmath12 ) .
a large spread in the neutrino arrival times ( due to cr diffusion through the gmc ) reduces the detected flux from individual grbs and stacked grbs .
hence , limits on the flux of grb neutrinos@xmath8 based on no detection in selected time and sky windows centered on visible grbs may not be applicable to the diffuse flux of neutrinos from all past grbs , which may contribute significantly and even dominate the isotropic component of the diffuse background of high energy neutrinos .
moreover , in this letter , we show that the spectrum , sky distribution and intensity of the diffuse background of high energy neutrinos recently discovered with the icecube detector@xmath0 are compatible with those of a non - isotropic galactic background expected from hadronic collisions of very high energy cosmic rays in the galactic interstellar medium ( ism ) plus an isotropic extragalactic background expected from hadronic collisions near source and in the intergalactic medium ( igm ) of high energy cosmic rays accelerated by the highly relativistic jets , which produce grbs in star forming galaxies .
because of feynman scaling@xmath13 , very high energy ( vhe ) cosmic rays with a power - law flux @xmath14 per - nucleon that collide with the baryons of an `` optically thin '' gas produce mesons whose leptonic and semileptonic decays produce neutrinos at a rate @xmath15 per unit volume . the constant @xmath16 depends@xmath17 only on the power - law index @xmath18 , @xmath19 is the inelastic pp cross section , and @xmath20 is the baryon density of the gas . since the average energy of the produced neutrinos is @xmath21 that of the cr nucleons , production of neutrinos in the 60 tev - 3 pev energy - range requires cr nuclei with energy _ per nucleon _ well above the cr `` knee '' where @xmath22 mb . the cr flux _ per nucleon _ in the 100 pev - 10 eev energy range is roughly@xmath23 @xmath24 with @xmath25gev .
this flux represents quite well the cr flux _ per nucleon
_ between the cr knee and cr ankle .
presumably , these very high energy ( vhe ) crs are accelerated by the highly relativistic jets of supernovae of type ic that produce grbs , most of which point away from earth@xmath26 .
most of the gas and dust in the galaxy ( @xmath27 ) resides in a relatively thin galactic disc whose mean surface density within @xmath28 kpc from its center is@xmath29 @xmath30 , and beyond it the mid plane density drops exponentially in the radial direction like @xmath31 $ ] with @xmath32 with a scale height @xmath33 kpc and where @xmath34 kpc is the distance of the solar system from the center of the disc . this gas and dust
are embedded in a large cosmic ray halo with roughly a constant density in the galactic disc .
hence , the total neutrino luminosity of the galaxy is given roughly by @xmath35 where @xmath36 for @xmath37 .
a rough estimate of the neutrino event rate in icecube due to galactic @xmath38 with @xmath39 tev , is @xmath40 is the mean effective area of icecube for neutrinos of the @xmath41 flavors ( assuming equal mixing due to oscillations ) .
for an energy - weighted@xmath3 @xmath42 above 60 tev , and @xmath43 as parametrized in reference 14 , eq.(3 ) yields @xmath44 neutrino events pointing back to the galactic disc and peaked towards the galactic center .
the contribution of crs accelerated in episodes of mass accretion onto the galactic central massive black hole may contribute significantly to this concentration . all together
, the energy spectrum of these galactic neutrino background is expected to have an @xmath45 spectrum , and its sky distribution is expected to follow roughly that of the diffuse galactic gamma rays in the energy range 100 mev - 100 gev , which are produced mostly by hadronic @xmath11 production and leptonic bremsstrahlung in the galactic disc and are not absorbed significantly by the background light in the galaxy .
there is mounting observational evidence that long duration grbs and their afterglows are produced by the interaction of highly relativistic jets of plasmoids ( cannonballs ) of ordinary matter ejected in stripped envelope supernova explosions , mainly of type ic , with the radiation and matter along the jets trajectories@xmath46 .
in particular , the prompt @xmath7-ray pulses and early - time x - ray flares are produced by the jet electrons through inverse compton scattering ( ics ) of photons in a light halo ( glory ) surrounding the progenitor star . such a light halo is formed by scattered stellar light from ejecta of pre - supernova eruptions@xmath47 .
thus , the total kinetic energy of the baryons in the jet is @xmath48 times @xmath49 , the total emitted gamma ray energy .
the jet decelerates by the swept in particles in front of it .
its random magnetic fields transform its kinetic energy by fermi acceleration to crs energy with a spectrum @xmath50 .
the crs escaoe the jet by diffusion through its random magnetic fields . for a kolmogorov spectrum@xmath51 of the random magnetic fields ,
their diffusion coefficient is @xmath52 , which yields a residence time @xmath53 and a cr flux @xmath54 where @xmath55 .
the cr jets move through the gmcs with practically the speed of light and produce narrow beams of high energy neutrinos mainly through @xmath56 and @xmath16 production in hadronic collisions with the baryons along their path in the gmcs . under the assumption of feynman scaling@xmath13 , cr spectrum @xmath57 , and a baryonic column density @xmath58 of the gmc , roughly , a fraction@xmath59 @xmath60 of the cr flux is converted to @xmath61 flux with the same @xmath62 power - law energy spectrum .
star formation in gmcs in the local universe ( redshift @xmath63 ) seems to have a threshold around a surface density@xmath64 @xmath65 , i.e. , a baryon column density @xmath66 .
such a column density is consistent with the mean column density @xmath67 inferred@xmath68 from x - ray absorption after correcting it for the observed metallicity@xmath69 of grb host galaxies @xmath70 , i.e. , @xmath71 .
hence , for a ballistic motion of the jet with approximately the speed of light through the gmc , we adop @xmath72 . assuming isotropic emission , the gamma ray energy emitted by a grb is @xmath73 where @xmath74 is the grb luminosity distance in a flat @xmath75cdm universe with a hubble constant @xmath76 km / s mpc , matter density @xmath77 , dark energy density @xmath78 and baryon density @xmath79 , all in critical mass units@xmath80 , and @xmath81 is the observed @xmath7-ray fluence from the grb . if the @xmath7-ray energy is beamed into a solid angle @xmath82 , then the true grb gamma - ray energy is @xmath83 , and the true grb rate is @xmath84 . consequently , the total power in @xmath7-ray emission by grbs at redshift @xmath85 is @xmath86 _ independent of the beaming angle of grbs_. noting that the full sky rate of grbs is@xmath87 @xmath88 , the flux of grb neutrinos of all flavors per sr at earth can be written as @xmath89 ^ 2}\,\left[{e\over m_p}\right]^{-1/3}\ ] ] where the summation extends over all @xmath90 long duration grbs with known redshift and @xmath91 , which were reported before may 15 , 2014 in the greiner catalog of grbs@xmath92 and in the gcn circulars archive@xmath93 , respectively .
the distribution of such long grbs as function of @xmath85 is shown in fig . 1 . for that distribution , eq .
( 5 ) yields @xmath94 _ per neutrino flavor_. to a very good approximation , all the vhe cosmic ray nuclei from grbs reach shortly ( on a cosmic time - scale ) the igm without any energy loss . in the igm
, they continue to produce neutrinos in collisions with the igm baryons .
most of the baryons in the universe ( @xmath95 reside in the igm with a density @xmath96 .
hence , the effective column density encountered by a cr nucleon , which is ejected into the igm at redshift @xmath85 , is @xmath97 . using a conversion coefficient @xmath98 for such @xmath99 in eq .
( 5 ) , the igm contribution to the diffuse vhe energy isotropic neutrino background due to grbs is @xmath100 _ per neutrino flavor_.
the bolometric luminosity per unit volume in the local universe ( @xmath101 ) is@xmath102 @xmath103 .
the bolometric luminosity of our milky way ( mw ) galaxy is @xmath104 .
the ratio @xmath105 is approximately also the observed ratio of their core collapse supernova rates and consequently also of their grb rates . assuming that the vhe neutrino luminosity of external galaxies ( eg ) due to hadronic cr interactions in their ism is proportional to their grb rates , then a rough estimate of the cummulative neutrino flux from all external galaxies is @xmath106 the factor 2.9 is the effective value of @xmath107 inferred from grbs x - ray absorption in the grb host galaxies , weighted by the star formation history@xmath108 as in eq .
( 6 ) yields an estimated @xmath109 contribution to the diffuse extragalactic background of vhe neutrinos , where @xmath110 is the average galactic neutrino background per sr .
meson production in hadronic interactions of very high energy cosmic rays in the galactic ism is expected to produce a diffuse galactic background of very high energy neutrinos with an energy flux @xmath111 .
the predicted number of galactic neutrino events ( all flavors ) in icecube with @xmath112 tev is roughly 3.6 events per year .
their neutrino arrival directions are expected to trace the ism column density in their arrival directions , i.e. , point back mainly to the galactic disc with a concentration towards the galactic center .
their sky distribution is expected to follow closely that of the diffuse galactic gamma ray background in the energy range 100 mev - 100 gev where the ism is transparent to the @xmath7-rays that are produced mostly by @xmath11 decay and leptonic bremsstrahlung .
meson production in hadronic interactions of very high energy cosmic rays accelerated by the highly relativisic jets , which produce grbs within giant molecular clouds in star forming galaxies , where most grbs take place , is expected to produce an isotropic background of very high energy neutrinos with a per flavor flux @xmath113 .
such an energy flux produces @xmath114 neutrino events per year with @xmath112 tev in the icecube detector .
cosmic ray interactions in the ism of external galaxies are expected to contribute to the _ isotropic _ background of very high energy neutrinos a per flavor energy flux @xmath115 .
such an energy flux produces @xmath116 neutrino events per year with @xmath112 tev in the icecube detector .
fig . 2 , compares the per - flavor flux@xmath0 of vhe neutrinos with energy above 60 tev measured with icecube@xmath0 and the predicted flux due to the hadronic interaction of cosmic rays in the galactic ism ( represented there by an equivalent isotropic mw flux ) , in giant molecular clouds of star forming galaxies where most core collapse supernovae , and hence grbs , take place , and in the igm .
it shows that the predicted flux is compatible with that observed by icecube .
special effort was made to base our estimates only on general considerations and priors , and avoid completely free adjustable parameters .
however , in view of the large uncertainties in the values of the priors , the good agreement between the expected and the observed neutrino flux by icecube can not be used to draw firm conclusions beyond the statement that they are compatible . obviously , much larger statistics are needed to test conclusively whether the diffuse background of vhe neutrinos is a sum of a galactic non - isotropic background with a spectrum @xmath117 and intensity proportional to the galactic column density in their arrival direction , plus an isotropic extragalctic background with a harder spectrum @xmath118 .
r. abbasi , et al .
( icecube collaboration ) , _ astrophys .
j. _ * 732 * , 18 ( 2011 ) ; m. g. aartsen , et al .
( icecube collaboration ) , _ astrophys .
j. _ * 779 * , 132 ( 2013 ) ; m. g. aartsen , et al .
( the icecube collaboration ) , e - print arxiv:1406.6757 .
e. waxman and j. bahcall , _ phys .
lett . _ * 78 * , 2292 ( 1997 ) ; j. p. rachen and p. meszaros , _ phys .
d _ , * 63 * , 023003 ( 2000 ) ; c. d. dermer,_astrophys
. j. _ * 574 * , 65 ( 2002 ) ; d. guetta , et al .
_ astroparticle phys . _ * 20 * , 409 ( 2004 ) ; l. a. anchordoqui , et al .
_ astroparticle phys .
_ * 29 * , 1 ( 2008 ) , i. cholis , and d. hooper , j. cos
. astroparticle phys . * 06 * , 030 ( 2013 ) ; k. murase and k. ioka , _ phys .
lett . _ * 111 * 121102 ( 2013 ) ; and references therein .
we have used the parametrization in ref .
25 of the star formation rate compiled by a. m. hopkins and j. f. beacom , _ astrophys .
j. _ * 651 * , 142 ( 2006 ) ; n. a. reddy and c. c. steidel , _ astrophys .
j. _ * 692 * , 778 , ( 2009 ) . | the diffuse background of very high energy extra - terrestrial neutrinos recently discovered with icecube is compatible with that expected from cosmic ray interactions in the galactic interstellar medium plus that expected from hadronic interactions near source and in the intergalactic medium of the cosmic rays which have been accelerated by the jets that produce gamma ray bursts ( grbs ) . |
video microscopy provides access to the full phase space information .
we record the trajectories of about 2300 colloidal particles confined to two dimensions at a flat water - air interface .
the species a ( diameter @xmath12 ) and b ( @xmath13 ) have a relative concentration of @xmath14 where @xmath15 and @xmath16 are the number of particles of both species in the field of view .
this prevents the system from crystallization .
+ the superparamagnetic nature of the particles lets us control the particle interactions _ in situ _ by an external magnetic field @xmath17 .
this is expressed by the dimensionless system parameter @xmath18 which effectively acts as an inverse temperature . here
, @xmath19 denotes the area density and is computed via a voronoi tessellation . @xmath20
represent the susceptibilities of species a and b , respectively .
after equilibration at low @xmath21 , the system was cooled down stepwise . for different states up to @xmath22 snapshots were analyzed . with a frame rate of @xmath23 , sampling times up to some @xmath24 seconds were achieved .
this is sufficiently long to probe dynamics even in the relaxation time regime @xmath25 for highest system parameters @xmath21 .
additional details of the setup are described elsewhere @xcite .
+ we follow the well established data evaluation scheme described in @xcite .
the harmonic part of the systems potential energy is bilinear in the displacements @xmath26 times the `` spring constant '' given by @xcite @xmath27 for the monolayer the dynamical matrix @xmath28 is a @xmath29 matrix and describes the elastic coupling between particles .
it is given by a sum of the second derivatives of the interparticle potential and @xmath30 index the two spatial dimensions .
elasticity naturally originates from the spring constants of the system .
these are the eigenvalues @xmath31 of the dynamical matrix we are interested in .
+ the displacement @xmath32 of particle @xmath33 is taken relative to its equilibrium position and its fourier component is given by @xmath34 .
we calculate those positions to be the center of mass of the trajectory @xmath35 , where the average is taken over a time interval @xmath36 significantly longer than any short time relaxation process . in the equation
, @xmath37 corresponds to the number of snapshots during the time interval @xmath36 and in an amorphous solid those time averaged positions converge to the quasi - equilibrium positions .
the expression for @xmath38 holds as long as @xmath36 is smaller than the relaxation time @xmath25 and particles are confined in their `` cage '' .
@xmath25 was extracted from msd measurements for high @xmath21 .
+ in an amorphous system we can not calculate the dynamical matrix @xmath39 a priori , but using the equipartition theorem we can measure it s eigenvalues @xmath31 .
the equipartition theorem reads @xmath40 where @xmath41 is the number of particles , each with two translational degrees of freedom . with eqn .
[ [ eq : harmonic_energy ] ] it follows @xmath42 the subscript @xmath43 denotes polarizations @xmath44 and depends on the relative orientation of @xmath26 and @xmath45 . for @xmath46 and @xmath47
we study longitudinal and transverse behaviour , respectively .
@xmath48 is termed `` dispersion relation '' but we keep in mind that in our system , oscillatory motion is overdamped due to the surrounding solvent . by applying this method , we implicitly assume the investigated modes to be plane waves .
this holds in the long wavelength limit corresponding to classical continuum elasticity theory @xcite .
there , the 2d system is considered to be a homogeneous structureless solid .
theoretical background on this approach can be found in @xcite .
+ from the dispersion relations , the elastic constants can be derived in the long wavelength limit @xcite . in a two dimensional isotropic
solid , the elasticity tensor @xmath49 possesses only two independent elements .
these two elements can be expressed via the lam coefficients @xmath6 and @xmath50 of continuum elasticity theory .
we extract the lam coefficients from the `` dispersion relation '' as : @xmath51 \ , , \label{eq : elastic_constants1 } \\ \frac{k_bt}{a_0 ^ 2\,\mu } & = & \lim_{\vec q \rightarrow 0 } [ \,\frac{1}{q^2 } \langle | u_\perp(\vec q ) |^2 \rangle\ , ] \ , .\label{eq : elastic_constants2}\end{aligned}\ ] ] here , @xmath52 is the mean interparticle distance , defining the area density as @xmath53 . in two dimensions , @xmath6 gives the shear modulus and @xmath54 the bulk modulus .
as we extract the elastic constants for @xmath55 , we are only accounting for long wavelength modes in elastic continuum theory , which justifies our approach even in disordered materials .
+ mean square displacement ( msd ) for both species of particles at different temperatures .
the kink at high temperatures slowly develops into a plateau .
however , it is impossible to distinguish unambiguously between solid and fluid state from this dynamical data .
the black solid line corresponds to free diffusion and the vertical dashed line to the sampling time @xmath36 used in the following analysis to determine the elasticity . ]
recent work on vibrational properties of colloidal glasses relied on the diagonalisation of the covariance matrix of the displacement field @xcite . from the spectrum of eigenvalues ,
this two dimensional principal component approach derives the density of vibrational states . in all cases ( hard spheres and soft pnipam particles ) an anomalous low frequency behavior
was found , which was connected to the boson peak known from scattering experiments in molecular or polymeric glasses @xcite .
this excess of modes was attributed to the non - isotropy of disordered systems at short to intermediate lengthscales . in atomic and molecular glasses ,
recent progress showed that long wavelength modes undergo a crossover from propagation to localization due to diffraction upon decreasing the wavelength @xcite . by analyzing the eigenvector structure of the covariance matrix ,
the spatial features of the corresponding modes could be visualized in the colloidal case @xcite .
it was concluded that the low frequency spectrum consisted mainly of localized `` swirling '' type of modes .
since this does not go along with our approach of plane waves in a continuous medium , we study the structure of modes via principal component analysis as well at the end of the manuscript . from the particle displacements @xmath56 ,
we compute the covariance matrix @xmath57 where @xmath33 and @xmath58 correspond to all components of all particle displacements . by diagonalisation
we obtain the spectrum of eigenvalues @xmath59 .
following @xcite we use the inverse as frequencies of the modes in our systems , providing a mapping @xmath60 .
the corresponding eigenvectors reveal the structure of those modes we are interested in . in the calculations , we follow the procedures described in @xcite . because principal component analysis requires large statistics , we use @xmath61 frames for constructing the covariance matrix . at this number
the results have statistically converged .
to start off , we present the measured mean squared displacement ( msd ) @xmath62 of the particles in fig .
[ figure_msd ] for different system temperatures . here
, @xmath41 presents the total number of particles which are located at positions @xmath63 at time @xmath64 .
it clearly shows the expected slow down of dynamics . at high temperatures ,
the data show a mere kink which slowly develops into a plateau for intermediate lag times @xmath65 . for longer times
@xmath25 , the dynamics exhibits ( sub ) diffusive behavior corresponding to alpha relaxation processes .
this becomes clear by comparing to the solid black line with slope one , marking free diffusion .
it is noteworthy that from this dynamic data alone , the exact temperature of a transition from the fluid state to a solid ( amorphous ) state can hardly be identified .
+ therefore we focus on spectroscopy of acoustic excitations . fig .
[ fig : disprel ] shows the measured normalized `` dispersion relations '' for different interaction parameters analyzed from trajectories of big particles . filled and empty symbols represent spring constants for longitudinal and transverse waves , respectively .
the curves are averaged over different directions of @xmath45 , taking advantage of the the isotropy of the system .
the error bars contain contributions from the finite optical resolution of the microscopy technique ( assumed to be @xmath66 ) , which affects the amplitudes of modes in fourier space .
as expected , they show an initial quadratic growth ( in agreement with debye behavior ) and exhibit a maximum near @xmath67 where the system is stiffest and propagation of waves fades out .
+ dispersion relations of longitudinal and transverse excitations rescaled with the linear dependence on @xmath21 expected in harmonic systems ( filled and empty symbols in top and bottom plot , respectively ) . the curves for different temperatures do not collapse , which can only be explained if structural rearrangement occurs .
due to the better optical resolution only trajectories of big particles are analyzed here and in the following plots . ] usually , the growing amplitude of the curves with increasing @xmath21 reflects the stiffening of the system upon cooling . in a harmonic crystal ,
the `` dispersion relation '' scales directly with the coupling parameter since the spring constants grow quadratically with @xmath68 and thus go linear with @xmath21 @xcite .
accordingly , the curves for different temperatures should collapse onto one single master curve if divided by the interaction strength .
interestingly , we do not find this scaling in the amorphous solid . besides the magnetic field , @xmath21 only depends on intrinsically fixed parameters as laboratory temperature , magnetic susceptibility or area density .
since the particle - particle interactions ( and therefore the `` spring constants '' ) depend crucially on the respective distance , a subtle change in structure might explain this behavior . with increasing interaction parameter
, the particles try to maximize their distances at constant voronoi area in order to minimize energy .
+ because in the disordered state the equilibrium particle positions are not fixed for all times , the structure of the system might adapt to the increased pressure without affecting the macroscopic area density .
this effect is barely visible in the structure factor .
+ on a side note , the growth of the dispersion amplitude mirroring the increasing rigidity tackles another unique feature of our model system .
since the pair potential is purely repulsive and the system is investigated at constant volume ( and particle number ) , the ensemble is intrinsically under pressure within the boundaries given by the sample cell . since pressure equals an energy density ( that is , energy per area in 2d ) , the pressure is given by the control parameter @xmath21 , too , such that temperature and pressure can not be varied independently . a quantitative analysis ( see fig . [
fig : moduli ] ) shows that the value of the dimensionless pressure given by @xmath21 is of the same order of magnitude compared to dimensionless elastic moduli .
such pressure affects the sound velocities given by the slope of the dispersion relation .
+ @xmath6 and @xmath69 are given by the intercept of an extrapolation ( indicated by green arrows ) derived from the longitudinal and transverse dispersion bands ( filled and empty black circles ) for a solid ( blue ) and fluid ( red ) sample .
the gray shaded region is used for fitting the data and the inset shows the same plot with logarithmic scaling . ] now , we turn to the main results , namely the elastic constants derived from the `` dispersion relation '' .
[ fig : extrapolation ] shows exemplary the dispersion relation for @xmath70 and @xmath71 divided by @xmath72 according eqn .
[ eq : elastic_constants1 ] and [ eq : elastic_constants2 ] .
the solid curves are fits to the data to get the elasticity indicated by green arrows .
we chose an intermediate regime ( @xmath73 ) for the extrapolation @xmath55 in the shaded region where the data fits best .
while in @xcite a linear fit was used to get the intercept we here expand @xmath74 and fit quadratically .
30 measurement runs were performed at various temperatures . due to prior work @xcite ,
the transition was expected at @xmath75 .
therefore , we focused on good temperature resolution in that temperature range , resulting in high quality data each with measurement length of at least @xmath76 seconds . the result is shown in fig .
[ fig : moduli ] .
the error bars contain contributions from the finite optical resolution and the extrapolation of reduced dispersion relations towards small @xmath77 as described by eqn .
[ [ eq : elastic_constants1 ] ] and [ [ eq : elastic_constants2 ] ] .
the statistics for the latter was greatly improved by averaging over different directions of @xmath45 . in the limits of these errors ,
the shear modulus below @xmath78 equals zero , regardless of the applied external magnetic field . for increasing temperatures above ,
the moduli grow linearly , reflecting the linear `` stiffening '' of the interaction potential . with the amount of data
, a zoom in on the shear modulus ( shown in the lower plot ) clearly reveals an abrupt rise of @xmath6 , consolidating the results of simulation @xcite and delivering the first experimental evidence for a discontinuous behavior of the shear modulus at the glass transition .
this jump separates the fluid phase from the amorphous state , as indicated by the shading from red to blue .
averaging about the data in the transition region , we find @xmath79 .
the scattering of the data is attributed to dynamical heterogeneity present in our system @xcite which have been shown to be correlated with elastic heterogeneity @xcite .
the dashed lines are fits to the moduli in the region @xmath80 . for the amorphous system
we find a slope of @xmath81 and @xmath82 .
this compares to measurements for dipolar monodisperse _ crystals _ , where we found @xmath83 and @xmath84 with a melting temperature at @xmath85 , see ref .
the latter is consistent with a thermodynamic calculation for a dipolar crystal at @xmath86 which yields a ratio of bulk to shear modulus of @xmath87 @xcite .
an increased ratio @xmath88 compared to crystals is proposed by nonaffine displacements affecting mainly the shear modulus of systems with short range interaction @xcite . for the dipolar monolayer the ratio @xmath89
is decreased , dominantly due to a reduced bulk modulus .
+ temperature dependent bulk- and shear modulus .
the data suggest a jump of moduli at the transition temperature , which is located at @xmath79 .
the lower plot zooms in on the ordinate . in the solid ,
the moduli grow linearly due to increased pressure indicated by dashed lines and we extrapolate this linear behaviour down to the fluid .
the inset shows a zoom of ordinate and abscissa : the data are not compatible with a continuous increase of elasticity as e.g. given by the dotted lines . ]
concerning the height of the jumps we are not aware of predictions of universal behavior of elasticity in thermal amorphous systems as e.g. the famous @xmath90 for youngs modulus in 2d melting @xcite , but the measured values in glass and crystal are of the same order .
for the jump in the bulk modulus @xmath91 a factor @xmath92 is calculated by mode coupling theory for hard spheres in 3d @xcite and implicitly given in @xcite where it is shown that 3d and 2d systems behave comparably . for the dipolar system at hand
we observe a factor @xmath93 for the glass and @xmath94 for a crystal @xcite . in comparison to hard discs we attribute this increased factor to the softness of the dipolar potential .
this discontinuous behaviour distinguishes the glass transition from jamming phenomena , for which the modulus was found to grow continuously from zero .
this is encompassing granular systems @xcite , percolating gels with attractive interactions @xcite and furthermore systems exhibiting friction between particles @xcite .
this might be an ubiquitous difference between thermal and granular systems .
+ while a recent discussion about 2d and 3d glass claims to find fundamental differences @xcite we expect the qualitative behavior to be robust against changes in dimensionality . in 2d crystals
the existence of peierls instabilities @xcite or mermin - wagner fluctuations @xcite as long wavelength density fluctuations ( of the order of system size ) is well established .
peierls argument using relative distance fluctuations is not based on periodicity ( just a typical particle distance is needed unlike e.g. in a gas ) and thus we expect such long wavelength density fluctuations ( with diverging amplitude in the thermodynamic limit ) also in 2d glass .
using local coordinates namely cage - relative displacements ( called modified dynamic lindemann - parameter for 2d crystals @xcite ) as suggested in @xcite would solve most of the discrepancies found in @xcite .
in the present analysis global coordinates are used and mermin - wagner fluctuations are expected to suppress the elastic response in the limit of infinite time scales .
thus the time interval @xmath36 to determine the quasi - equilibrium position has to be chosen carefully .
finite time effects investigated by varying the time interval @xmath36 used to determine the particle `` equilibrium position '' . on the glassy side , the shear modulus does not decay to zero even for the longest integration times .
data of the system at @xmath71 in the solid ( blue ) and @xmath95 in the fluid ( red ) while the bulk and shear moduli are marked by filled and empty symbols .
the inset shows the high frequency region in the frozen state , which suggests a power law behavior .
upper and lower black lines are functions with exponents @xmath96 and @xmath97 . ] in fig .
[ fig : ft ] , the finite time effects of the evaluation procedure are shown . by changing the averaging time
@xmath36 determining the quasi - equilibrium positions , different frequency regimes are probed .
short averaging times yield a significant increase in elastic moduli , reflecting the expected stiffening at high frequencies even for fluids @xcite . for low frequencies ( large @xmath36 )
, in contrast , the shear modulus of the fluid state decays to zero , while in the glassy state , it decays into a plateau , even for times up to @xmath24 seconds .
this stability of the shear modulus is remarkable , because other quantities measuring the system dynamics already show signs of relaxation .
this averaging time used in fig .
[ fig : moduli ] is @xmath98 indicated by the green arrow in fig .
[ fig : ft ] ( limited by the longest accessible time from the fluid data ) . comparing with fig .
[ figure_msd ] this is well after the onset of the alpha relaxation time @xmath25 for the datasets in the vicinity of the transition .
in contrast to the shear modulus , the bulk modulus is much more sensitive to increased dynamics , as it does not show a well developed plateau for low frequencies .
this is attributed to the fact that transverse shear waves have much larger amplitudes than longitudinal compression waves , such that distortions of the wavefront affect properties derived from the latter much more strongly .
this striking behavior proves the shear modulus to be a good quantity to investigate the glass transition .
the inset finally shows a log - log plot of the high frequency regime for the frozen state .
for about two decades , both moduli show a power law behavior .
the blue lines are functions with exponents of @xmath99 , describing high frequency bulk and shear properties , respectively .
this power law behavior challenges theoretical descriptions .
+ in the last section we turn to the analysis of the shape of the excitations .
[ fig : modestructure]a ) shows the spectrum of eigenvalues @xmath100 of the displacement field covariance matrix @xmath101 for the system at @xmath102 , a temperature slightly below the glass transition temperature .
the most _ principal _ components ( carrying the most weight ) correspond to low `` frequencies '' due to the mapping @xmath103 .
because @xmath101 has the dimensions of square meters as it measures amplitudes , the `` frequencies '' have the unit @xmath104 .
only a valid mapping of amplitudes , wavelength , and frequencies ( implicitly using a dispersion relation ) connects a length scale given by the eigenvalues of the covariance matrix to frequencies used in the density of states @xmath100 .
an interesting feature of @xmath100 is the double peak structure , centering at @xmath105 and @xmath106 .
we interpret this as a reminiscence of van - hoove type singularities as they appear e.g. in crystals due to the vanishing sound velocity at the brillouin zone edge @xcite .
naively one could argue that first and second peak represent longitudinal and transverse modes respectively , but since pca is a priori insensitive to polarizations the connection is less trivial . nevertheless , fig .
[ fig : modestructure]b ) shows the density of states as function of temperature .
the peaks shift to lower frequencies with increasing temperature ( decreasing interaction strength ) reflecting a softening of the system as expected .
note that the first peak grows at the expense of the second upon approaching the fluid phase .
this resemblance with a density of states known from a crystalline solid also shows up in the reduced density of states @xmath107 which is shown in the inset of fig .
[ fig : modestructure]a ) . scaling with
@xmath108 yields a constant value for the low frequency region up to the first peak .
this is in accordance with the debye model @xmath109 , where @xmath110 is the dimensionality of the investigated system . in the debye model , it is plane wave excitations which make up the low frequency part of the spectrum .
putting issues regarding the amplitudes frequency mapping aside , the mode structure is clearly resolved by the eigenvectors of the most principal components of the displacement field . for decreasing wavelength ,
these are shown in graphs [ fig : modestructure]c - f with the corresponding frequency noted above .
those frequencies are also indicated by red arrows in fig .
[ fig : modestructure]a ) . while [ fig : modestructure]e , d ) show some kind of lissajous - pattern , for modes of a given frequency but different phases , they start to show incoherence for higher frequencies indicated by a random distribution of arrows . the coherence length of the imaged virtual motion
gets notably smaller , reflecting the growing influence of disorder .
this culminates at frequencies accounting to the first peak , where the mode structure appears almost disordered in fig .
[ fig : modestructure]f ) .
nonetheless , care has to be taken since pca was shown to generate mode pattern from mixed states at higher frequencies ; the low frequency modes on the other hand are expected to be pure states @xcite . in fig .
[ fig : modestructure]c ) one can clearly verify the plane wave character of the excited states in the low frequency limit . accordingly , the ansatz to equally distribute energies via the equipartition theorem on plane wave like modes for @xmath111 ( eqn . [ eq : equipartition ] ) is reasonable .
we have experimentally investigated the elastic properties of a colloidal glass former in great detail . by tuning the interactions _ in situ _ , we have accessed the well defined displacement field in fourier space at numerous effective temperatures and used the equipartition theorem to calculate the dispersion relations .
scaling with the interaction parameter reveals a hidden structural change upon cooling which is too subtle to be picked up by other functions . in the limit of long wavelength ,
the dispersion relations yield the elastic moduli of compression and shear , of which particularly the latter is of great interest for an deeper understanding of the glass transition .
it has been demonstrated that the shear response unambiguously separates the fluid state from the amorphous solid .
the transition itself exhibits a discontinuous behavior , which is in agreement with theoretical predictions and simulations and distinguishes it from granular matter or gelation type scenarios .
finite time analysis reveals a striking stability of the shear - related measurements , which are proven to be decoupled from relaxation phenomena like cage breaking .
this makes the shear modulus an exceptionally good quantity to investigate the glass transition . j. j. crassous , m. siebenbrger , m. ballauf , m. drechsler , d. hajnal , o. henrich , and m. fuchs , _
shear stresses of colloidal dispersions at the glass transition in equilibrium and in flow _ , j. chem .
phys . , * 128 * , 204902 ( 2008 ) j. j. crassous , l. casal - dujat , m. medebach , m. obiols - rabasa , r. vincent , f. reinhold , v. boyko , i. willerich , a. menzel , c. moitzi , b. reck , and p. schurtenberger , _ structure and dynamics of soft repulsive colloidal suspensions in the vicinity of the glass transition _ , langmuir , * 29 * , 10346 ( 2013 ) k. n. nordstrom , e. verneuil , p. e. arratia , a. basu , z. zhang , a. g. yodh , j. p. gollub , and d. j. durian , _ microfluidic rheology of soft colloids above and below jamming _
lett . , * 105 * , 175701 ( 2010 ) k. chen , w. g. ellenbroek , z. zhang , d. t. n. chen , p. j. yunker , s. henkes , c. brito , o. dauchot , w. van saarlos , a. l. liu , and a. g. yodh , _ low - frequency vibrations of soft colloidal glasses _ , phys
. rev .
lett . , * 105 * , 025501 ( 2010 ) e. courtens , m. foret , b. hehlen , b. ruffl , and r. vacher , _ the crossover from propagating to strongly scattered acoustic modes of glasses observed in densified silica _ , j. phys .
: cond .
, * 15 * , 1279 ( 2003 ) b. ruffl , m. foret , e. courtens , r. vacher , and g. monaco , _ observation of the onset of strong scattering on high frequency acoustic phonons in densified silica glass _ , phys . rev . lett .
, * 90 * , 095502 ( 2003 ) the numbers are estimated from fig .
[ fig : moduli ] in the manuscript and fig . 2 of @xcite .
for the crystal we compare values in the fluid and in the solid for temperatures where the renormalization procedure does not play a role .
v. prasad , v. trappe , a. d. dinsmore , p. n. segre , l. cipelletti , and d. a. weitz , _
universal features of the fluid to solid transition for attractive colloidal particles _
, faraday discuss .
, * 123 * , 1 ( 2003 ) t. still , c. p. goodrich , p. j. yunker , s. schoenholz , a. j. liu , and a. g. yodh , _ phonon dispersion and elastic moduli of two - dimensional disordered colloidal packings of soft particles with frictional interactions _ , phys
. rev .
e , * 89 * , 012301 ( 2014 ) | a solid - amorphous or crystalline - is defined by a finite shear modulus while a fluid lacks such .
we thus experimentally investigate the elastic properties of a colloidal glass former near the glass transition : spectroscopy of vibrational excitations yields the dispersion relations of longitudinal and transverse phonons in the glassy state . from the long wavelength limit of the dispersion relation
we extract the bulk and the shear modulus .
as expected , the latter disappear in a fluid and we measure a clearly resolved discontinuous behaviour of the elastic moduli at the glass transition .
this not only determines the transition temperature @xmath0 of the system but also directly addresses recent discussions about elasticity during vitrification .
we show that low frequency excitations in our system are plane waves such that continuum elasticity theory can be used to describe the macroscopic behaviour . in general
, there are several ways to define glassy systems @xcite .
for instance , a characteristic change in the thermodynamic properties volume and enthalpy upon cooling a fluid may be used to define the glass transition temperature @xmath0 .
another property often investigated for molecular or atomic glasses is the viscosity @xmath1 .
taking into account the rapid slowing down of dynamics , the glassy state is reached when the viscosity exceeds @xmath2
pa@xmath3s upon crossing the freezing temperature . however , all of these definitions are difficult to apply to soft matter systems . in studies on the mesoscopic or microscopic scale , access to
the macroscopic property @xmath1 often is difficult , if not impossible .
secondly , these systems are `` softer '' up to a factor of @xmath4 ( 2d ) or @xmath5 ( 3d ) which rules out the idea of an universal , viscosity - dependent categorization .
thus , other ways have been suggested to pin down @xmath0 @xcite .
a manifestation of solidity is the emergence of rigidity , expressed by the shear modulus @xmath6 becoming finite .
microscopically , prerequisites for solidity are long ranged dynamical correlations .
while it is easy to understand how such long range correlations emerge in a crystal ( based on the concept of long ranged periodic order ) , solidity is much less easy to grasp on an amorphous , disordered background . as the shear modulus is zero in fluids ,
we expect a significant change of @xmath6 at the onset of vitrification @xcite .
although @xmath6 is a macroscopic property , it is measurable locally via the equipartition theorem which makes this concept well suited for soft matter systems @xcite . from an experimental point of view , this approach is advantageous because it relies on measuring the change of a quantity from zero to a finite value . on the other hand , the usual approach of measuring the divergence of dynamical quantities like
the viscosity is much harder to realize experimentally because it requires access to a much larger parameter space .
+ elasticity of complex fluids and their viscoelastic behaviour has been tackled mostly by rheological measurements so far @xcite . in general
, these methods allow the study of frequency dependent storage @xmath7 and loss @xmath8 modulus .
the former describes the elastic contribution , while the latter covers dissipative processes upon shear .
thus , a non - vanishing @xmath9 in the limit @xmath10 indicates a solid .
these quantities have been investigated extensively by experiment as well as theory @xcite , and from the data , a clear distinction between solid and fluid state might be drawn .
this is also true for flow curves , which might be not only acquired by conventional rheometers but also by microfluidic rheology @xcite .
here we address solely the elasticity by internal fluctuations without external driving . +
while it has been shown recently that the zero - frequency shear modulus @xmath11 is a good quantity to observe the glass transition @xcite , there has been an ongoing debate on the behaviour of @xmath6 when an amorphous solid forms .
replica theory approaches @xcite predict a continuous growth from zero upon cooling .
this agrees with the view from granular systems , where theory predicts critical fluctuations close to the jamming transition of these athermal systems which also lead to an algebraic growth of @xmath6 from zero @xcite . on the other hand ,
another recent replica theory calculation reveals a discontinuous jump at the glass transition @xcite .
this complies with mode coupling theory predictions @xcite where @xmath6 is connected to a non - ergodicity parameter appearing in a stress auto - correlation function . a discussion about the discontinuous elasticity in the context of random first - order transition is given in @xcite .
in addition , simulations of hard discs also showed a discontinuous onset of the shear modulus @xcite by accessing the system with the equipartition theorem .
similar results from novel four - point correlation calculations back this up further @xcite .
+ here , we evaluate microscopy data from a two dimensional colloidal model system with soft interactions in equilibrium .
this gives us access to information on the single particle scale as opposed to atomic or molecular systems , where only ensemble data may be studied . by determining the displacement field and subsequent analysis in fourier space ,
the dispersion relations of acoustic - like excitations are obtained and provide insight into the macroscopic elastic properties of the system . as the system is cooled , we are able to confirm the discontinuous nature of the shear modulus at the glass transition . |
the kelly strategy is attractive to practitioners because of its robust and optimal properties , e.g. that it dominates any other strategy in the long run and minimizes the time to reach a target , as shown by @xcite , and among others by @xcite , @xcite , @xcite , @xcite , @xcite , @xcite .
it also carries intuitive appeal because of its connection to information theory , which was part of the original motivation in @xcite , and in particular by the fact that optimal investing is one and the same thing as optimal application of the available information ; see @xcite and @xcite . on the other hand , the kelly strategy in its pure unconstrained form , which we will refer to as the ` free kelly ' strategy , generally leads to a too aggressive strategy with too high leverage and risk to be of practical applicability .
real investors are subject to various constraints , and a practical version of the kelly principle can be formulated as : optimize the objectively expected compound growth rate of the portfolio subject to the relevant constraints .
it is in this form that the kelly principle has become increasingly popular among practitioners over the last 20 years or so . in this paper
we are concerned with the optimal strategy in this sense , for the particular case where the constraint is in the form of a stop - level .
unlike for most other forms of constraints , the effect of a stop - level on the optimal strategy varies with the portfolio value and with the time remaining until the stop - level is adjusted ( reset ) .
the related problem of finding the growth optimal strategy for a portfolio subject to a drawdown constraint ( a trailing stop - loss rule ) was solved using a combination of hjb and martingale methods by @xcite , see also @xcite and section [ grossmanzhou ] below .
the plan of the paper is as follows .
section [ sec : moti ] provides some background , and section [ sec : hjb ] reviews the hjb for the investment problem and introduces notation . in section [ sec : nonlinpde ] we derive a non - linear pde for the optimal strategy , and in section [ sec : known ] we consider a few examples with known solutions , and discuss to what extent the equation can be applied . in section [ koswslr ]
we provide a numerical solution for the kelly strategy subject to a periodically reset stop - loss rule . section [ sec : conclusion ] concludes .
consider a discretionary portfolio manager running a trading book in a global macro hedge fund .
such a trader is subject to various constraints when selecting the appropriate level of risk to run .
apart from leverage , stress - test and concentration limits and maybe constraints on the asset selection , the book will typically be subject to a value - at - risk ( var ) limit . in practice
most traders will run their book significantly below these limits because there is another kind of constraint : the stop level .
the stop may take several forms .
it is for example common for a book with monthly liquidity to be subject to explicit soft and hard stop levels between liquidity dates . a book with a @xmath0 var limit at
@xmath1 1 day , could for example have a soft stop at @xmath2 mtd ( month - to - date ) where the var limit drops to @xmath3 and a hard stop at @xmath4 mtd where the var limit drops to zero for the remainder of the month . in any given month
the stop - level is a fixed number of percentage points below the portfolio value at the beginning of the month .
any triggered stop will be released and reset after the first coming month - end , where investors will be given a chance to withdraw their funds . hired ( non - partner )
traders managing a relatively small sub - book as part of a larger hedge fund are typically also subject to an explicit terminal stop at around @xmath5 ytd ( year - to - date ) , at which point the book is closed and they lose their job . even founders and senior partners , who may not be subject to any explicit stop levels , will in effect operate under an implicit stop level .
this is due to the fact that if the drawdown from the high - water mark becomes too large , then there will be no performance fees for the foreseeable future to pay for the operation , and the fund is likely to find itself in a downward spiral of brain - drain , investor withdrawals , and lower returns .
the desire to avoid this situation will in effect mean that the fund can only operate at full risk when it is near its high - water mark .
indeed , it has happened more than once that a hedge fund has simply closed after a steep drawdown , only for the managers to open a new fund in a different setting .
history has of course also shown what can happen when a trading operation is run without stops or with unenforceable stops due to illiquidity or management failure .
leaving the moral aspects aside , it is clear that stop - levels , implicit or explicit , are a major part of the constraints limiting the risk in most funds , and that the constraining effect of the stop level scales with the portfolio value and the time remaining until the stop level is reset .
the penalty for hitting the stop is lost opportunity . for a terminal stop
the opportunity cost is potentially infinite since the loss is all future earnings .
a strategy can try to avoid hitting the stop by reducing the risk as the portfolio value approaches the stop level , but in practice it is impossible to guarantee that the stop will not be hit unless the risk is zero . in a model world where the risky assets are modelled using brownian motion , it is possible to avoid hitting the stop by reducing the risk , but this does not mean that there is no opportunity loss . scaling down the risk also scales down potential returns , and increases the amount of time the portfolio end up spending in the vicinity of the stop . thus there is a kind of ` dead zone ' just above the stop level , where the portfolio has very little risk , and it can become virtually stuck . the optimal strategy must find the right balance between scaling the risk down too quickly versus too slowly , foregoing potential returns in both cases . with a terminal stop , or an infinitely long time to reset
, the solution is to effectively ignore the capital below the stop level , and invest in the kelly optimal way as if the capital above the stop level was the total capital . doing this , the trader is only risking the fraction of the capital he can survive to lose .
this strategy which is independent of time , but a function of the portfolio value is discussed in section [ sec : kwts ] , and the related optimal strategy for a portfolio subject to a trailing maximum drawdown limit is discussed in [ grossmanzhou ] . a stop level which resets at the end of each period is more interesting from a practical point of view . with a periodically reset stop - level
the penalty for hitting the stop , or becoming stuck in the dead zone , is less severe , since it is only a temporary loss of opportunity .
if there is only a few days left before the reset , the portfolio value must be very near the stop - level for this to have any effect , but in the other limit where there is a very long time to the reset , the optimal strategy must approach the same strategy as the one for a terminal stop / infinite horizon .
it is therefore clear that an optimal strategy will scale the risk up and down , not only according to how far away the stop is from the current portfolio value , but also according to the time remaining until the stop - level reset .
it is the kelly growth optimal strategy for this problem , discussed in detail in section [ koswslr ] , which is our main objective .
a realistic modelling of the problem would be very complicated , depend on idiosyncratic details , and is beyond our ambitions in this paper .
instead our goal is to study the optimal strategy in the simplest non - trivial model with a stop - rule .
as it turns out , the resulting strategy , discussed in section [ koswslr ] , corresponds quite closely to common trader intuition , and can provide a rough guide to the appropriate risk level for practical portfolios .
we use the standard minimal portfolio model consisting of a risky asset displaying geometric brownian motion and an exponentially growing risk - free asset , respectively defined by the evolution @xmath6 with @xmath7 the growth rate and @xmath8 the volatility of the risky asset , @xmath9 the risk - free rate , and @xmath10 the forward looking increment of a wiener process .
a self - financing trading strategy , which invests the fraction @xmath11 in the risky asset and the fraction @xmath12 in the risk - free asset , will result in a portfolio value which evolves according to the equation @xmath13dt + \alpha\sigma dw_t.\ ] ] the analysis can be simplified by introducing the discounted relative portfolio value @xmath14 we can consider @xmath15 to be an index for the portfolio with @xmath16 and with @xmath17 the total return at time @xmath18 as measured in time zero dollars .
alternatively , if @xmath9 is the rate of inflation , it can be considered the real return .
this definition eliminates the constant risk - free drift term @xmath19 from the evolution equation for the discounted portfolio value @xmath20,\ ] ] where we have indicated that the strategy may depend on the value of the portfolio as well as time .
we are concerned here with problems for which the objective is the maximization or minimization of a terminal ` reward ' function @xmath21 . for a kelly growth optimal portfolio
this would be the period growth rate @xmath22 .
the value function is defined as the expectation of the terminal reward over paths starting from the intermediate point @xmath23 at time @xmath18 , and evolving according to ( [ eq : hjb : dpi ] ) with the optimal control , i.e.@xmath24.\ ] ] it is significant that we have excluded a consumption type term in the objective , ( a running cost / reward ) , as it would complicate the analysis considerably , but using the methods of appendix [ app : legendre ] such terms can be handled at least in simple cases ( by adding the equivalent terms to ( [ eq : lt06 ] ) ) .
the value function @xmath25 and the control @xmath11 satisfies the hamilton - jacobi - bellman equation @xmath26 which must be solved subject to the final time boundary condition @xmath27 the equations ( [ eq : hjb2])-([eq : hjb2cond ] ) can also be used to find the strategy minimizing the expectation of a terminal objective @xmath21 by replacing @xmath28 by @xmath29 . in either case
the objective function @xmath30 must have a form that makes the problem well - defined . as it stands
the hjb problem defined by ( [ eq : hjb2 ] ) and ( [ eq : hjb2cond ] ) is rather complicated as it involves two unknown functions @xmath25 and @xmath11 linked by a combined pde / maximization equation .
the trick which renders the problem solvable is that under certain conditions we can perform the optimization over the control before solving for @xmath25 , ( see e.g. @xcite ) .
the condition is essentially that @xmath25 must have a shape which makes the optimization problem in ( [ eq : hjb2 ] ) have a well - defined solution .
this must be confirmed a posteriori for the actual solution , ( but in many applications it is rather obvious ) .
in essence @xmath25 must be concave for a maximization problem and convex for a minimization problem . according to bellman s principle
, the optimal control must also be optimal over each sub - interval . in the present context
this means that @xmath11 must optimize ( [ eq : hjb2 ] ) at each instant of time .
this is effected by setting @xmath31 , which gives the formal solution @xmath32 substituting this back into ( [ eq : hjb2 ] ) , the control is eliminated resulting in the hjb equation for the value function alone @xmath33 the problem is then reduced to solving ( [ eq : hjb : value ] ) subject to the terminal time boundary condition ( [ eq : hjb2cond ] ) together with any additional boundary conditions ( for example conditions specified on the edges @xmath34 , which could be asymptotic ) .
the standard procedure is thus to start by solving for the value function , and then as the second step to calculate the optimal strategy from ( [ eq : hjb : value : control ] ) .
a hint that ( [ eq : hjb : value ] ) hides a simpler underlying structure is provided by the fact that a legendre transform turns it into a linear second order pde as discussed in appendix [ app : legendre ] .
it is natural to ask , if there is a pde for the optimal strategy without any reference to the value function .
such an equation would not only be interesting in its own , but would potentially allow us to circumvent the value function , and could have considerable practical benefits . in this section
we show how it is possible to eliminate the value function , and derive a non - linear partial differential equation which is obeyed by the optimal control itself .
the derivation requires that the functions involved are sufficiently smooth in the interior of the domain , but in practice this is not necessarily a great limitation .
the starting point is the hjb for the value function ( [ eq : hjb : value ] ) and the equation for the optimal control ( [ eq : hjb : value : control ] ) .
if we take the hjb equation ( [ eq : hjb : value ] ) together with the equations which follow by operating on it with @xmath35 and @xmath36 and the equation for the optimal control ( [ eq : hjb : value : control ] ) together with the equations which follow by operating on both sides by @xmath37 , @xmath35 and @xmath36 , then we have a system of seven equations for the unknowns @xmath11 , @xmath38 , @xmath39 , @xmath40 , @xmath41 , @xmath42 , @xmath43 , @xmath44 , @xmath45 , @xmath46 , @xmath47 .
our aim is to derive a single equation only involving the stochastic control and its derivatives @xmath11 , @xmath38 , @xmath39 , @xmath40 . on the surface
it would look like we would need eight equations in order to be able to eliminate the seven derivatives of the value function and still be left with a single equation for the control , but note that the stochastic control as given by ( [ eq : hjb : value : control ] ) only depend on the value function through the ratio between the two derivatives @xmath48 , which means that only six equations are needed to eliminate all the @xmath25 s .
if we start by using four equations to eliminate @xmath44 , @xmath45 , @xmath46 , @xmath47 , the three remaining equations can be put in the form @xmath49,\\
\partial_{\pi}^2 j & = \frac{(\mu - r)^2}{2\sigma^2 } \frac{(\partial_{\pi } j)^2}{\partial_t j}.\end{aligned}\ ] ] this gives us two equations for the ratio @xmath50 , respectively @xmath51.\end{aligned}\ ] ] it follows that the optimal strategy must obey the equation @xmath52 this is a non - linear pde which is second order in the portfolio value and first order in time .
the non - linearity is manifested in the @xmath53 factor on the right hand side . like the hjb for the value function
, it must be solved backwards in time from a final time boundary condition . in general the solution
must also obey boundary conditions in the @xmath23-direction which may be asymptotic .
it is reminiscent of a non - linear diffusion equation in reverse time , but it does not strictly have the required structure ( see appendix [ app : legendre ] ) . the equation is independent of the effective drift rate @xmath54 , but dependency on this quantity may enter via the boundary conditions . since any constant will solve ( [ eq : alpha : pde ] ) , it is of little use for problems with such a solution .
rather ( [ eq : alpha : pde ] ) is applicable to problems requiring a non - trivial solution in the sense of an optimal strategy which is a non - constant function of the portfolio value and time .
the second caveat is that we must be able to establish the boundary conditions for the control for example from limit or asymptotic properties which the solution must obey .
for some problems it may not be obvious how to find these boundary conditions in which case it will be necessary to revert to the hjb for the value function .
if , on the other hand , we have a strategy solving ( [ eq : alpha : pde ] ) , then the corresponding value function can be found from the first order equation ( [ eq : alpha2value ] ) from which it is determined up to a constant .
the equation ( [ eq : alpha : pde ] ) can be transformed into different forms which may be more or less convenient for a particular investment problem .
some examples are discussed in appendix [ app : transf ] .
an explanation for the form of the equation , and the nonlinearity in particular , can be found as follows .
if we describe the optimal strategy by the amount invested in the risky asset rather than by the fraction of the portfolio value , i.e. using @xmath55 , then the equation takes the form @xmath56 equation ( [ eq : gamma : pde ] ) can also be obtained directly from the legendre transformed hjb problem as discussed in appendix [ app : legendre ] .
let @xmath57 denote the discounted risky asset price and @xmath58 the discounted bond price as per ( [ eq : model01])-([eq : model02 ] ) and ( [ eq : disc ] ) .
then @xmath59 and @xmath60 , and the discounted portfolio value evolves according to the prescription @xmath61 the process ( [ eq : port : gamma ] ) is self - financing and is a special case of the general self - financing feedback strategy @xmath62 with the self - financing condition @xmath63 , with @xmath64 , where @xmath65 is the number of risky asset shares and @xmath66 the number of risk - free bonds in the portfolio at time @xmath18 . since @xmath60 , @xmath66
does not contribute to the dynamics , but is a slave keeping track of the accounting of the cash generated / needed for the re - balancing @xmath67 . in our case the portfolio ( [ eq : port : gamma ] ) can be decomposed as @xmath68 , with @xmath69 the value of the risky assets , and @xmath70 the value of the risk - free investment . whereas the portfolio @xmath15 is self - financing by construction , the risky asset investment , @xmath71 ,
does not alone represent a self - financing quantity because a general self - financing strategy can redistribute the portfolio freely between @xmath72 and @xmath73 as long as it keeps the sum unchanged . because we are focussing on smooth strategies here , we can link the increment @xmath74 to the increments in the portfolio value , @xmath75 , and time , @xmath76 , via the expansion @xmath77dt
+ ( \partial_{\pi}\gamma)d\pi_t.\end{aligned}\ ] ] it follows that when @xmath72 represents an optimal trading strategy obeying ( [ eq : gamma : pde ] ) , this reduces to the expression @xmath78 this means that although @xmath72 is not self - financing for a general sub - optimal trading strategy , it will , for the optimal trading strategy , behave as a self - financing quantity with the total increment in @xmath72 ( from market evolution plus re - balancing ) directly proportional to the increment in the portfolio ( which because the portfolio is self - financing is given by market evolution alone ) .
the same holds for the risk - free part of the portfolio which for the optimal strategy obeys @xmath79 .
we note in particular that the nonlinearity of ( [ eq : alpha : pde ] ) and ( [ eq : gamma : pde ] ) is a consequence of the feedback nature of the portfolio evolution as specified by equation ( [ eq : port : gamma ] ) , and that the optimal strategy displays a certain balance between the risky and risk - free parts of the portfolio by only redistributing between these in proportion to the total increment in the portfolio .
in this section we comment on the application of the strategy equation ( [ eq : alpha : pde ] ) to some standard investment problems , where there is a known analytic solution to the hjb equation .
the purpose is to investigate to what degree ( [ eq : alpha : pde ] ) can be applied with particular focus on the boundary conditions needed .
also [ sec : conststrat ] and [ sec : kwts ] establishes solutions which the optimal strategy in section [ koswslr ] must approach asymptotically .
the examples have been chosen to illustrate problems where the optimal strategy is respectively constant , [ sec : conststrat ] , value dependent , [ sec : kwts ] , and dependent on both value and time , [ sec : brownestrat ] .
the free kelly strategy is the unconstrained strategy maximizing the period growth rate and can also be interpreted as the strategy maximizing the terminal logarithmic utility , i.e.@xmath80.\ ] ] the well - known optimal strategy , @xmath81 , is to invest a constant fraction , @xmath82 of the portfolio in the risky asset .
the corresponding value function is given by @xmath83 being constant ( [ eq : kelly : alpha ] ) is a trivial solution to ( [ eq : alpha : pde ] ) .
a slight generalization of ( [ eq : kellyutil ] ) is the power function utility @xmath84 , with @xmath85 , which is also referred to by the acronym crra for constant relative risk aversion .
the value function for this utility is given by @xmath86 and the corresponding optimal strategy is the constant @xmath87 the free kelly strategy corresponds to the special case @xmath88 , and a fractional kelly strategy , @xmath89 , investing a constant fraction @xmath90 of the free kelly strategy , corresponds to optimizing a power function utility with @xmath91 .
a rational investor would never invest more than the free kelly strategy unless forced to do so by some constraint .
this means that @xmath90 would normally be less than one and @xmath92 therefore negative corresponding to a utility which is capped to the upside .
benchmarked fractional kelly strategies have recently been investigated in detail by @xcite .
if an investor has a critical level @xmath93 below which his wealth must never fall , and therefore can only put at risk the part of his capital exceeding this level , then the capital below the critical level is in effect irrelevant to his utility .
a kelly investor will therefore optimize the growth rate of the capital exceeding the critical level .
this goal is expressed in the value function @xmath94,\ ] ] with the hjb solution given by @xmath95 the corresponding optimal strategy is @xmath96 this is the unconditional free kelly strategy multiplied by the fraction that the excess capital constitutes out of the total capital . we observe that ( [ eq : kf03 ] ) is a solution to ( [ eq : alpha : pde ] ) because it is independent of time and obeys @xmath97 .
it is therefore completely defined as the time independent solution to ( [ eq : alpha : pde ] ) subject to the boundary conditions @xmath98 and @xmath99 for @xmath100 .
we note that ( [ eq : kf03 ] ) is a ` constant proportion portfolio insurance ' ( cppi ) strategy with the multiplier equal to the free kelly strategy , i.e. the amount invested in the risky asset is @xmath101 .
more generally we conclude that any time independent optimal strategy must take the form @xmath102 for some constants @xmath103 and @xmath104 .
@xcite , ( see also @xcite ) , have solved the problem of finding the optimal strategy for an investor subject to a maximum drawdown rule for a general class of utility functions .
the essence of their solution can in the present context be stated as follows .
investors who in accordance with the kelly principle are seeking to maximize the long term real / discounted capital growth rate @xmath105,\ ] ] subject to the maximum drawdown constraint ( with @xmath106 a given constant ) @xmath107 where @xmath108 is the high - water mark @xmath109 should follow the strategy of investing an amount equal to the free kelly strategy times the excess over the moving stop level in the risky asset , i.e. @xmath110 , or in other words , follow the strategy @xmath111 we note the obvious similarity between ( [ eq : kf03 ] ) and ( [ eq : gz04 ] ) , but refer to the references above for more information .
@xcite has shown that the problem of finding the strategy which maximizes the probability of reaching a specified real return target , @xmath112 , in a finite time period , @xmath113 , has the following solution .
as argued by browne , the probability that the target is hit at any time @xmath114 is equivalent to the probability that the terminal wealth hits the target , because once the target is reached the probability maximizing strategy would be to invest exclusively in the risk - free asset .
the value function for this problem is thus defined by @xmath115 with our notation browne s solution to the problem is given by the value function ( with @xmath116 ) @xmath117 and the corresponding optimal strategy is given by @xmath118 with @xmath119 , and with @xmath65 the standard normal density and @xmath120 the corresponding cumulated probability function .
straight forward differentiation will show that ( [ eq : browne:03 ] ) is indeed a solution to ( [ eq : alpha : pde ] ) . as shown in appendix
[ app : transf ] , ( [ eq : browne:03 ] ) is an example of a solution which can be found from ( [ eq : alpha : pde ] ) via separation of the variables . it would nevertheless have been difficult to arrive at ( [ eq : browne:03 ] ) directly from ( [ eq : alpha : pde ] ) , because although the boundary condition @xmath121 is obvious , this is not the case for other boundary conditions . the optimal strategy ( [ eq : browne:03 ] )
diverges in the limit @xmath122 to such a degree that the probability of hitting the target level approaches a finite value as the remaining time goes to zero , i.e. @xmath123 for @xmath124 .
a strategy with finitely capped leverage would have @xmath125 for @xmath126 .
we consider the investment problem of finding the strategy which optimizes the long term growth rate when investment is subject to a periodically reset stop - loss rule . to study this problem
we consider the minimal model introduced in ( [ eq : model01])-([eq : hjb : dpi ] ) with an investment universe consisting of one risky and one risk - free asset .
the stop - loss rule means that if in a given period the value of the portfolio @xmath15 drops to a certain level @xmath127 ( the stop - level ) , then the risk limit drops to zero , and the portfolio must be invested exclusively in the risk - free asset for the remainder of that period , and therefore the portfolio value will remain at @xmath127 . at the beginning of a new period of duration @xmath128 the stop - loss rule is reset , and the level @xmath127 is set a fixed number of percentage points below the opening portfolio value @xmath129 . in this model
consecutive investment periods are therefore both independent and identical in distribution .
we seek a strategy defined on the domain @xmath130\times [ 0,t]$ ] .
since we know that the optimal strategy must solve the hjb problem and in particular the strategy pde ( [ eq : alpha : pde ] ) , it follows that it must be smooth in the interior of the domain of definition , and that it is completely determined by the boundary conditions .
it is intuitively clear that in the limit where the portfolio value is far above the stop - level , the stop is irrelevant and the optimal strategy will equal the free kelly strategy , i.e. @xmath131 for @xmath132 . on the other hand ,
the stop - rule means that the optimal strategy must equal zero along the lower boundary on the portfolio value , i.e. @xmath133 for all @xmath134 $ ] .
finally it is clear that sufficiently near the final time boundary as @xmath122 with a portfolio value strictly above the stop - level , @xmath135 , the optimal strategy will approach the free kelly strategy with the limit @xmath136 for @xmath135 .
this follows from the fact that the underlying process is continuous in time , and therefore in the limit @xmath122 the stop - rule becomes irrelevant for a portfolio with finite leverage when the portfolio value is strictly above the stop - level , because there is not time enough left for the portfolio value to drop to the stop - level . to summarize we have the three dirichlet boundary conditions @xmath137 , \\
\alpha(\pi_c , t)=0 , & \;\;\text { for } t\in [ 0,t],\\\ \alpha(\pi , t)=\alpha_k , & \;\;\text { for } \pi>\pi_c.\end{aligned}\ ] ] to solve the problem we employ the strategy laid out in ( [ eq : transf01])-([eq : transf04a ] ) in appendix [ app : transf ] , i.e. change variables from @xmath138 to @xmath139 and use the scaled optimal strategy @xmath140 .
the new variables are respectively the scaled reciprocal portfolio value @xmath141 and the scaled time to period end @xmath142 with @xmath143 .
the domain of definition for the variables is @xmath144\times [ 0,\infty]$ ] , where @xmath145 is the stop - level , @xmath146 corresponds to the limit @xmath132 , and @xmath147 is the final time boundary .
the equation to solve is thus ( [ eq : transf04a ] ) subject to the boundary conditions @xmath148 these boundary conditions determine the solution uniquely , but as discussed in appendix [ app : transf ] , the asymptotic behavior along the fourth boundary , which is the limit @xmath149 , also follows . in this limit @xmath150 and
the opportunity cost of hitting the stop - level is prohibitively high .
the optimal strategy can not have an explicit dependence on time in this limit , and must therefore approach ( [ eq : kf03 ] ) which in the transformed variables reads @xmath151 for @xmath149 .
we are not aware of an analytic solution to ( [ eq : transf04a ] ) obeying the particular boundary conditions ( [ eq : ubound01])-([eq : ubound03 ] ) , but solving it numerically is straight forward . the solution is well behaved and although more advanced methods can be used , since @xmath152 the explicit euler method , which can be implemented in just a few lines of code , is convergent for @xmath153 .
0.35 for @xmath144\times [ 0,1]$ ] .
the ` dead zone ' is indicated in red and the free kelly limit in purple .
( b ) a specific example for an ex - ante sharpe ratio of @xmath154 and a period of one month .
the scaled strategy @xmath155 is plotted as function of time to period end for four different portfolio values specified by the distance to the stop level in percent . from the bottom
the four curves correspond to stop levels which are respectively @xmath156 , @xmath157 , @xmath158 , @xmath159 below the current portfolio value .
time to period end is indicated on the abscissa in weeks.,title="fig : " ] 0.35 for @xmath144\times [ 0,1]$ ] .
the ` dead zone ' is indicated in red and the free kelly limit in purple .
( b ) a specific example for an ex - ante sharpe ratio of @xmath154 and a period of one month .
the scaled strategy @xmath155 is plotted as function of time to period end for four different portfolio values specified by the distance to the stop level in percent . from the bottom
the four curves correspond to stop levels which are respectively @xmath156 , @xmath157 , @xmath158 , @xmath159 below the current portfolio value .
time to period end is indicated on the abscissa in weeks.,title="fig : " ] figure [ fig : ustrat](a , b ) and [ fig : asymp](a , b ) show results calculated with this method .
[ fig : ustrat](a ) shows the function @xmath160 , and [ fig : ustrat](b ) shows examples of @xmath155 for four fixed distances above the stop level .
the ex - ante sharpe ratio is @xmath161 , and the four curves correspond to stop levels which are respectively @xmath156 , @xmath157 , @xmath158 and @xmath159 below the current portfolio value , ( i.e. @xmath162 and @xmath163 ) .
the time period in [ fig : ustrat](b ) is one month corresponding to a fund with monthly stop - level reset , and is indicated on the abscissa in weeks remaining until the reset . 0.35 as a function of @xmath164 for four fixed values @xmath165 , @xmath166 , @xmath167 and @xmath168 .
( b ) the scaled strategy @xmath160 as a function of @xmath169 for @xmath170 corresponding to a stop - level @xmath171 below the portfolio value . in both graphs
the dashed line shows the asymptotic limit for @xmath149.,title="fig : " ] 0.35 as a function of @xmath164 for four fixed values @xmath165 , @xmath166 , @xmath167 and @xmath168 .
( b ) the scaled strategy @xmath160 as a function of @xmath169 for @xmath170 corresponding to a stop - level @xmath171 below the portfolio value . in both graphs
the dashed line shows the asymptotic limit for @xmath149.,title="fig : " ] figure [ fig : asymp](a , b ) illustrates the asymptotic convergence towards the terminal stop strategy as @xmath169 increases .
the main parameter setting the scale for the solution is the ex - ante sharpe ratio of the risky asset @xmath172 .
the optimal fraction @xmath173 only depends on the distance to the stop - level through @xmath141 and the time left through @xmath142 .
the characteristic time scale @xmath174 is determined by the ex - ante sharpe ratio .
the free kelly leverage is given by the ratio between the sharpe ratio and the underlying volatility , i.e. @xmath175 , but the volatility of the free kelly portfolio equals the sharpe ratio , i.e. @xmath176 . in practice
other constraints are likely to kick in before the leverage reaches the free kelly limit @xmath177 , and the leverage will then be capped below this limit .
if the portfolio is subject to a var limit corresponding to a maximum volatility @xmath178 , then @xmath179 , and the scaled strategy is subject to the cap @xmath180 . with an ex - ante sharpe ratio of @xmath154 as in figure [ fig : ustrat](b ) , a portfolio with a @xmath0 var limit at @xmath1 , 1 day ,
corresponding roughly to @xmath181 annual volatility , would for example be subject to the cap @xmath182 .
consider a portfolio with one risk - free and multiple risky assets each evolving according to the standard model of geometric brownian motion @xmath183 if @xmath184 denote the allocation to the @xmath185th risky asset , then the discounted portfolio value will evolve according to @xmath186 the free kelly portfolio is the portfolio optimizing the objective expectation of the growth rate , and it follows from @xmath187 = \left(\boldsymbol{\alpha}^{\dagger } ( \boldsymbol{\mu}-\boldsymbol{r } ) -\tfrac{1}{2}\boldsymbol{\alpha}^{\dagger}\boldsymbol{c } \boldsymbol{\alpha}\right)dt,\ ] ] ( with vectors and matrices indicated in bold and @xmath188 indicating transposition ) that this portfolio is given by the expression @xmath189 where @xmath190 is the covariance matrix @xmath191 . we can now consider the restricted optimization problem consisting in finding the optimal strategy for an investor who has decided only to invest in portfolios proportional to the free kelly portfolio .
this is a very restricted problem compared to the full multi - dimensional portfolio optimization , but it is not entirely irrational , since we know that this portfolio possesses many optimal properties , see e.g. @xcite , @xcite and also @xcite , @xcite , @xcite . the free kelly portfolio can itself be considered a ( risky ) asset , and we are therefore back to the situation discussed in the previous section .
the dynamics of the kelly portfolio is given by @xmath192 with @xmath193 , which means that the unconstrained kelly optimal allocation into the kelly portfolio is @xmath194 , as it must be for consistency .
it follows that we can find the optimal strategy for a trader subject to a stop - loss rule by the same procedure as for a single risky asset , and that the combined strategy can be expressed as @xmath195 where the optimal strategy @xmath196 , which obeys equation ( [ eq : alpha : pde ] ) with @xmath197 , is a factor between zero and one .
we have derived a non - linear partial differential equation for the optimal strategy which does not involve the value function .
the derivation requires that both the value function and the strategy should be sufficiently smooth in the interior of the domain .
the non - linearity of the equation for the optimal strategy is a result of the feedback nature of the portfolio evolution , and the equation expresses a certain balance between the increments in the risky and risk - free parts of the portfolio .
when boundary conditions can be established independently , the equation can be used to find the optimal strategy , and we have applied this method to find the kelly optimal strategy subject to a periodically reset stop - loss rule .
successful discretionary portfolio management is as much art as science , and intangible qualities like ` superior and unbiased judgement ' , ` alertness to position and funding stresses ' , ` mental resolution and confidence to act before the crowd ' , ` superior information gathering and networking ' , etc .
, are at least as important as any quantitative method
. nevertheless , this does not mean that the quantitative side of optimal portfolio management is unimportant .
the results derived in this paper can only provide a rough guide to real world portfolio optimization , mainly because they are based on a simplistic and benign model .
the main limitation is the model of market randomness based on brownian motion .
this has the consequence that all variables are continuous in time without gap risk , and allow the portfolio composition to be adjusted after each infinitesimal market move .
it is well known that real markets have more in common with lev - processes with index less than two , and display jumps ( gaps ) , serial correlation ( trends ) and volatility persistence ( clustering ) .
transaction costs , limited liquidity and slippage are other problems .
there is also the problem that even if the model were reasonable , we would still have to rely on statistical estimates of the true underlying parameters @xmath7 and @xmath8 .
thus the optimal strategy should incorporate the corresponding bayesian uncertainty .
therefore , the methods discussed here must be adapted accordingly before they can be applied . nevertheless , the results presented provide information about the optimal strategy in the limiting case of a very well behaved market .
algoet , p. h. and cover , t. m. , 1988 .
assymptotic optimality and assymptotic equipartition properties of log - optimum investment .
_ annals of probability _ , * 16 * , 876 - 898 . reprinted in @xcite .
kelly , j. l. , 1956 . a new interpretation of information rate .
_ bell sys . tech .
j. _ , * 35 * , 917 - 926 . reprinted in @xcite .
barron , a. r. and cover , t. m. , 1998 .
a bound on the financial value of information .
_ ieee transactions of information theory _ , * 34 * , 5 , 1097 - 1100 . reprinted in @xcite .
breiman , l. , 1961 .
optimal gambling systems for favourable games , _ proceedings of the 4th berkeley symposium on mathematical statistics and probability _ , * 1 * , 63 - 68 . reprinted in @xcite .
browne , s. , 1996 .
portfolio choice and the bayesian kelly criterion , _ adv .
_ , * 28 * , 1145 - 1176 .
browne , s. , 1999a . reaching goals by a deadline :
digital options and continuous - time active portfolio management , _ adv .
_ , * 31 * , 551 - 577 .
browne , s. , 1999b .
the risk and rewards of minimizing shortfall probability . _ the journal of portfolio management_. browne , s. , 2000a .
can you do better than kelly in the short run ? _ finding the edge : mathematical analysis of casino games _ , 215 - 231 , vancura , o. , cornelius , j. , eadington , w.r .
institute for the study of gambling and commercial gaming , university of nevada .
browne , s. , 2000b .
risk - constrained dynamic active portfolio management , _ management science _ , * 46 * , 1188 - 1199 . reprinted in @xcite .
cover , t. m. and thomas , j. a. , 2006 .
elements of information theory , second edition , _ wiley - interscience_. grossman , s. j. and zhou , z. , 1993 .
optimal investment strategies for controlling drawdowns . _ math . finance _ , * 3 * , 241 - 276 .
cvitanic , j. and karatzas , i. , 1995 . on portfolio optimization under `` drawdown '' constraints . _
i m a volumes in math . and its appl .
_ , * 65 * , 35 - 46 .
cvitanic , j. 1997 .
optimal trading under constraints .
_ financial mathematics : lectures given at the 3rd session of the centro internazionale matematico estivo , bressanone , italy 1996 .
_ , 123 - 190 , springer .
davis , m. and lleo , s. , 2008 . risk - sensitive benchmarked asset management .
_ quantitative finance _ , * 8 * , 415 - 426 .
davis , m. and lleo , s. , 2010 .
fractional kelly strategies for benchmarked asset management .
@xcite , 385 - 407 .
hakansson , n. h. , 1970 .
optimal investment and consumption strategies under risk for a class of utility functions .
_ econometrica _ , * 38 * , 587 - 607 .
reprinted in @xcite .
hakansson , n. h. , 1971 . on optimal myopic portfolio policies with and without serial correlation of yields , _
journal of business _ , * 44 * , 324 - 334 . reprinted in @xcite .
maclean , l. c. , thorp , e. o. , and ziemba , w. t. , editors , 2011 .
the kelly capital growth investment criterion : theory and practice , _ world scientific handbook in financial economic series _ , * 3 * , world scientific publishing co. pte . ltd .
merton , r. c. , 1990 . continuous - time finance .
_ blackwell_. thorp , e. o. , 1969 .
optimal gambling systems for favorable games . _ rev . int .
_ * 37 * , 273 - 293 . reprinted in @xcite
. thorp , e. o. , 1971 .
portfolio choice and the kelly criterion .
_ bus . and econ .
_ , 215 - 224 . reprinted in @xcite .
thorp , e. o. , 2006 .
the kelly criterion in blackjack , sports betting and the stock market .
_ handbook of asset and liability management , volume 1 , edited by s.a .
zenios and w. ziemba_. reprinted in @xcite .
yong , j. and zhou , x. u. , 1999 .
stochastic controls : hamiltonian systems and hjb equations , _ appl . of mathematics , stochastic modelling and
applied probability _ , * 43 * , springer .
ksendal , b. , 2003 .
stochastic differential equations : an introduction with applications , 6 ed .
, _ universitext _ , springer .
the hjb equation ( [ eq : hjb : value ] ) can be significantly simplified by a legendre transformation using as the new independent variable in place of @xmath23 , the slope of the tangent to the value function along the @xmath23-direction .
let the transformed value function @xmath198 be defined by @xmath199 with @xmath200 and inversely @xmath201 .
the legendre transformation requires that the equation @xmath202 has a unique solution for @xmath23 , i.e. that the inverse function @xmath203 exist at least in an interval around the point @xmath204 for each point in time @xmath18 . in the present context where @xmath25 is a value function ,
this is unlikely to be a problem , since @xmath204 , being the rate of increase with respect to @xmath23 , is a meaningful quantity .
the legendre transform is its own inverse and the properties of the transformation follow from the rules of differentiation . in general
the transformation is time - dependent , i.e. the value of @xmath204 corresponding to a fixed value of @xmath23 is a function of time : @xmath205 .
differentiating @xmath198 with respect to @xmath18 , while keeping @xmath23 fixed , therefore gives @xmath206 , but applying this procedure to ( [ eq : lt01 ] ) results in the simple relation @xmath207 because @xmath201 .
another useful relation which follow by differentiation is @xmath208 applying the rules of the transformation , it follows that the hjb equation ( [ eq : hjb : value ] ) is equivalent to the following linear second order pde for @xmath198 @xmath209 which must be solved subject to the transformed terminal time boundary condition , e.g. for the kelly utility @xmath210 , we would have @xmath211 , because at time @xmath128 we have @xmath212 .
let us denote by @xmath213 the transformed function corresponding to the optimal investment @xmath55 , via the definition @xmath214 it then follows from ( [ eq : hjb : value : control ] ) that @xmath215 and from ( [ eq : lt03 ] ) and ( [ eq : lt05 ] ) we have @xmath216 the transformed hjb problem is significantly simpler than the original problem , and if we operate on ( [ eq : lt05 ] ) with @xmath217 ( with @xmath204 fixed ) and on ( [ eq : lt045 ] ) with @xmath218 ( with @xmath18 fixed ) , then we get two equations from which we can eliminate @xmath219 .
the result is the following linear equation for the optimal investment @xmath220 ( [ eq : lt06 ] ) is a diffusion equation with time running backwards , ( i.e. it is a special case of the general non - linear one - dimensional diffusion equation with the structure : @xmath221 $ ] ) . to derive the corresponding equation for @xmath222 we have to remember that there is an implicit time - dependence in @xmath204 when @xmath23 is kept fixed and therefore @xmath223 .
otherwise the manipulations are straight forward and the result is ( [ eq : gamma : pde ] ) .
the non - linear pde for the optimal strategy ( [ eq : alpha : pde ] ) can be transformed in many ways . here
we focus on a few with potential applications .
let @xmath224 denote the free kelly strategy , and let @xmath196 denote the scaled control @xmath225 if we introduce the scaled dimensionless time variable @xmath226 then @xmath227 measures the time remaining in units of the characteristic time @xmath228 , and the equation for the optimal strategy @xmath229 takes the form @xmath230 let @xmath127 denote a portfolio value of particular interest .
depending on the setting this could for example be the starting value , a stop - loss , or a target level .
we can then define the scaled reciprocal portfolio value by @xmath231 the optimal strategy @xmath232 then obeys the equation @xmath233 this form is convenient for investment problems with a stop level @xmath127 , because @xmath234 translates to @xmath235 $ ] . since @xmath236 , the sign of @xmath237 is given by the curvature @xmath238 .
thus as @xmath149 the equation will seek to straighten out any curvature present in the initial state of the control at @xmath147 to approach a final state with @xmath239 consistent with the boundary conditions in the @xmath164-direction .
for an investment problem with a stop level and @xmath235 $ ] , this means that @xmath240 for @xmath149 , with @xmath241 , @xmath242 the respective boundary conditions . for an investment problem with an upper target level @xmath243 the reciprocal portfolio value is conveniently defined as @xmath244 , and the domain
@xmath124 then translates to @xmath245 . in this case
the asymptotic behavior as @xmath149 for a finite value of @xmath164 would be to approach a constant , @xmath246 , which equals the boundary value of the control at the target level .
the equation ( [ eq : transf04a ] ) for the optimal strategy has separable solutions of the form @xmath247 with constants @xmath248 and @xmath249 , and where the function @xmath250 must solve the ode @xmath251 browne s probability maximizing strategy ( [ eq : browne:03 ] ) belongs to this class .
the transformed solution resulting from the change of variables and scaling can be written as @xmath252 , with @xmath244 , where @xmath243 is the target level .
the function @xmath250 , which solves ( [ eq : transf08 ] ) with @xmath253 , is given by @xmath254 where @xmath65 and @xmath120 refer to respectively the pdf and cdf of the standard normal distribution . alternatively , ( [ eq : transf03 ] ) can be simplified in analogy with ( [ eq : gamma : pde ] ) by the transformation @xmath255 , which gives the pde @xmath256 for the scaled strategy variable @xmath257 .
the equation ( [ eq : transf06 ] ) has self - similar solutions of the form @xmath258 , where @xmath259 for some constants @xmath248 , @xmath260 and @xmath249 , and where the similarity function must solve the ode @xmath261 [ lastpage ] | from the hamilton - jacobi - bellman equation for the value function we derive a non - linear partial differential equation for the optimal portfolio strategy ( the dynamic control ) .
the equation is general in the sense that it does not depend on the terminal utility and provides additional analytical insight for some optimal investment problems with known solutions .
furthermore , when boundary conditions for the optimal strategy can be established independently , it is considerably simpler than the hjb to solve numerically . using this method we calculate the kelly growth optimal strategy subject to a periodically reset stop - loss rule . |
searching is undoubtedly one of the most basic problems in computer science and computational physics . in this context ,
searching is not just restricted to a physical database but could also be searching through a state space for an entry which fulfills a specific clause such as the constraint satisfiability problem ( @xmath0-sat ) .
the classical complexity of such a task scales linearly with the size of the dataset , @xmath1 , to be searched .
intuitively , it is easy to see this must be the case as every item must be checked in turn until the specific item is found . on average , half the items will have to be checked before the correct one is located .
this leads to the best classical scaling which can be achieved , @xmath2 .
one of the most important quantum algorithms discovered thus far is the searching algorithm of grover @xcite .
grover showed that an item could be found from a set of @xmath1 in a time quadratically faster than the classical case , @xmath3 .
grover s algorithm has been shown to be both optimal and also one of the few quantum algorithms which is provably faster than any possible classical algorithm @xcite .
several years after the introduction of this algorithm , shenvi , kempe and whaley @xcite gave a quantum search algorithm based instead on the discrete time quantum walk , which was first introduced with algorithmic applications in mind by aharonov et al .
@xcite and ambainis et al .
this quantum walk approach to the search problem is able to match the quadratic speed up of grover s algorithm .
the quantum walk search algorithm has been studied in detail and many improvements have been made since its introduction .
in fact , due to the many uses of searching in algorithms , the quantum walk search algorithm has become a standard tool in developing new quantum algorithms @xcite .
the quantum walk has also recently been shown to be universal for quantum computation and hence a computational primitive @xcite , again showing it is a powerful tool . in @xcite ,
the items of the dataset are laid out as the vertices of an undirected graph , specifically a hypercube of dimension @xmath4 , on which the quantum walk can be solved analytically @xcite
. other recent work by potoek et al .
@xcite has improved the original algorithm by adding an additional coin dimension , allowing the probability of the marked state to approach unity after just one run of the algorithm .
this brings the running time of the quantum walk search algorithm very close to the optimal for searching an unsorted dataset , @xmath5 .
zalka @xcite has previously shown that , for a probability of finding the marked state to be one , this is the best that can be achieved .
however , the hypercube studied in @xcite is a highly connected but non - physical structure . in order to make the algorithm more physical ,
the study of the search algorithm on lower dimensional structures was started by benioff @xcite .
he considered the additional cost of the time it would take a robot searcher to move between different spatially separated data points on @xmath6-dimensional lattices , stating that in two spatial dimensions , @xmath7 , no speedup was apparent .
subsequently , aaronson and ambainis ( aa03 ) @xcite introduced an algorithm based on a divide and conquer approach , contradicting this claim with a run time of @xmath3 in dimensions @xmath8 and @xmath9 when @xmath10 .
.summary of runtimes of quantum search algorithms in various dimensions . [ cols="<,^,^,^,^",options="header " , ] [ percprobs ] it is fairly obvious that at this critical percolation threshold , the properties of the lattice change significantly . for lattices with a percolation probability below the percolation threshold , it is clear that many of the sites in the lattice will be unreachable , whereas above the threshold the opposite is true ( though perhaps through a less direct route than in a fully connected lattice ) . due to their transport properties , percolation lattices
are widely used to model various phenomena including forest fires , disease spread and the size and movement of oil deposits . for a good introduction to both the theory and use of percolation lattices ,
see stauffer and aharony @xcite . we are using the percolation lattices as a description for the database arrangement that we wish to run the quantum walk search algorithm upon . as
the disorder introduced by using percolation lattices is random , we ran the search algorithm on many different percolation lattices ( 5000 ) , and averaged over the results .
it is obvious that at low probabilities of vertices ( or edges ) existing , that there may be sections of the graph that the quantum walk is unable to reach .
in fact , at very low probabilities , it is likely that the marked state will be in a small , unconnected region of the lattice where it will never be ` found ' . in these cases , this means the marked state will only ever be able to attain a small portion of the total probability .
we set the condition on the algorithm that the probability of the marked state must reach at least twice the value of the initial superposition in order for it to succeed .
similarly , the time to find this maximum probability is artificially smaller than it should be if the entire lattice was connected .
this is due to the walker only having to coalesce on the marked state over a small piece of the lattice . in order to combat this
, we set the time to find the marked state as zero if the algorithm failed .
if it succeeded , we took the reciprocal of the time to find the marked state .
after averaging over many different percolation lattices , we again took the reciprocal of this averaged time in order to give a clearer view on how the algorithm scaled with time .
we also set the probability of the marked state to be zero if the algorithm failed . in order to run the quantum walk search algorithm on percolation lattices
, we have to deal with the fact that the lattice is not @xmath6-regular . in this
setting , we can not just add self loops to make the lattice regular as in @xcite as we want to know exactly how the disorder affects the algorithm .
instead , we take the grover coin for the degree of the vertex in question and ` pad ' it out with the identity operator for the edges that are missing .
for example if we have a vertex with just edge 3 missing , the operator would be @xmath11 where @xmath12 represents the grover coin with edges 1 , 2 and 4 present . in the case of a two dimensional percolation lattice , there are 16 combinations of edges that can be present / missing . for a three dimensional percolation lattice ,
this increases to 64 combinations . in order to deal with this
, we maintain the labelling of the edges as previously and assign a binary number to each edge , depending on whether an edge is present or not .
the example above , eq . ( [ percgrover ] ) , would therefore be @xmath13 .
this creates the @xmath14 combinations we require .
there is then a fixed mapping between each binary number and the correct coin for each vertex .
in addition to the coin operator changing , we must also modify the initial state to account for the missing vertices or edges .
this could be done in several ways .
we try to stick as closely to the initial state of the basic quantum walk search algorithm by just splitting the state into an equal superposition over all the possible edges present .
we now show our initial results for the quantum walk search algorithm on two dimensional site percolation lattices .
we firstly show , fig .
[ twodprobperc ] , how the maximum probability of the marked state varies with both the size of the dataset and the percolation probability .
we see , as we would think intuitively , that as the percolation probability drops and the structure becomes less connected , the maximum probability of the marked state decreases . , varies with the size of the dataset and the percolation probability for site percolation in two dimensions.,scaledwidth=75.0% ] we note that the scaling of the maximum probability initially maintains the logarithmic scaling of the basic 2d lattice before eventually reverting to the scaling of the line , @xmath15 , at lower percolation probabilities . in the case of site percolation
, this change in scaling seems to occur at roughly probabilities below @xmath16 , not significantly higher than the critical percolation threshold .
this is expected as at the critical threshold , the structure has in general a single path from one side to the other , effectively a 1d lattice .
our numerical results match this behaviour , with the scaling of the probability of the marked state matching that of the line at this point . at
percolation probabilities higher than the critical threshold , we see a change in the prefactor to the scaling of the maximum probability of the marked state .
we show this prefactor to the logarithmic scaling in fig .
[ twodpercprobscaling ] .
it is easy to see that as soon as the percolation probability passes the critical threshold , @xmath17 , the scaling increases in a linear fashion .
we also note here , after investigation on a finer scale , that there is a gradual change in this prefactor scaling around the critical percolation threshold .
the time to find the marked state follows a similar behaviour , gradually changing from the quadratic scaling of the 2d lattice to a classical linear scaling as @xmath18 reduces .
we show the time to find the marked state for site percolation in fig .
[ twodtimesite ] . , to the scaling of the time to find the maximum probability of the marked state , from the data in fig .
[ twodtimesite ] , varies with the size of the dataset and the percolation probability for site percolation in two dimensions .
also shown is @xmath19 to indicate the lower bound of the algorithm ( dashed line).,scaledwidth=75.0% ] we see that when @xmath20 , the scaling of the time to find the marked state is very similar to the classical run time , @xmath2 .
the kinks in this scaling ( and the other percolation probabilities ) are just from averaging over many percolation lattices . given more time , a higher number could be run and thus a smoother scaling obtained .
it can be seen that the time to find the marked state seems to retain the quadratic quantum speed up , even in the presence of a non - trivial level of disorder . as in the work of leung
@xcite , it seems as though the scaling of the time to find the marked state may follow a fractional scaling from quadratic back to linear as , @xmath21 where @xmath22 is the time to find the marked state and @xmath1 is the size of the dataset .
we follow the analysis in @xcite to establish how the scaling of the time to find the marked state varies with the percolation probability .
we show , in fig .
[ twodperctimescaling ] , how the value of the coefficient @xmath23 varies as the level of disorder is increased .
we can see the quadratic speedup is maintained , @xmath24 , for percolation probabilities of roughly @xmath25 .
below this probability , the quantum speed up disappears gradually to end at the classical run time when @xmath26 .
this is for the same reason as in the scaling of the maximum probability of the marked state , at the critical threshold the structure is effectively a line . below the critical threshold ,
the algorithm fails ( the marked state is probably in a disconnected region ) .
we note here that the coefficient , @xmath23 , is not exactly 0.5 as we expect for the quadratic speed up .
this is most probably due to the fact that percolation lattices are random in nature , and we only average over a specific number . if we averaged over more , then we would see a more constant scaling of the coefficient at @xmath19 , i.e. a full quadratic speed up .
we now turn our attention to three dimensional site percolation lattices .
we follow the same analysis as in the two dimensional case .
we firstly show , fig .
[ threedprobperc ] , how the maximum probability of the marked state varies as the percolation probability is decreased .
we see , as in the two dimensional case , that the basic scaling of the maximum probability matches that of the three dimensional lattice until the percolation probability drops to roughly the critical percolation threshold , @xmath27 .
we show in fig .
[ threedpercprobscaling ] , how the prefactor to this scaling of the maximum probability varies with the percolation probability . in the same way as the two dimensional case , we see an almost linear scaling of the prefactor once the percolation probability has passed the critical threshold
. the scaling here does nt seem to be as close as in the two dimensional case .
this is probably because in the case of three dimensional percolation lattices , there are many more combinations of lattice which can be created . averaging over more of these lattices would most probably give a smoother fit .
, varies with the size of the dataset and the percolation probability for site percolation in three dimensions.,scaledwidth=75.0% ] the time to find the marked state , in the three dimensional case , follows the same behaviour as in the two dimensional percolation lattices .
we show in fig .
[ threedtimesite ] , how the time to find the marked state varies with the percolation probability .
we see , fig . [ threedperctimescaling ] , as in the two dimensional case , that the scaling coefficient , @xmath23 , gradually changes from the quadratic speed up to the classical run time .
again , we note that the quadratic speed up is maintained for a non - trivial amount of disorder before gradually changing to the classical run time at the point @xmath26 . , to the scaling of the time to find the maximum probability of the marked state , from the data in fig .
[ threedtimesite ] , varies with the size of the dataset and the percolation probability for site percolation in three dimensions .
also shown is @xmath19 to indicate the lower bound of the algorithm ( dashed line).,scaledwidth=75.0% ] we do note , as in the two dimensional case , that the coefficient is not exactly 0.5 .
this can be explained in the same way as the two dimensional percolation lattices , and averaging over more lattices should give a constant value of the coefficient @xmath23 .
in this paper , we have discussed various factors which affect the efficiency of the quantum walk search algorithm .
we introduce a simple form of tunnelling which allows us to modify the substrate we use as the database arrangement , and use this to interpolate between structures with varying dimensionality and degree .
we find that although the dependence on the spatial dimension of the underlying substrate is strong , it is not the only factor which affects the efficiency of the algorithm .
we also find secondary dependencies on the connectivity and symmetry of the structure .
in addition , we use percolation lattices to model disorder in the lattice in a simple way . in this case
we find , counter - intuitively , that the algorithm is able to maintain the quantum speed up even in the presence of non trivial levels of disorder .
we now discuss our findings for each factor in turn .
we have shown two different ways in which we can interpolate between structures of differing spatial dimension .
firstly , we use a tunnelling operator to vary specific edges of a lattice enabling us to gradually change the spatial dimension of the lattice . in this case
, we find a sudden change in the scaling of the maximum probability of the marked state as soon as there is even a very small probability of the edges existing .
this seems to indicate that the ` strength ' of the edges in the lattice is of little importance , with the dependence on the specific spatial dimension taking precedence .
however , we find that the prefactor to the scaling of this probability varies with the strength of the tunnelling edges , increasing as the tunnelling strength increases .
the basic scaling of the time to find the marked state is not affected by the change in dimensionality , we note though that the prefactor to the scaling decreases as the tunnelling strength increases , hence the algorithm becomes more efficient .
the other case we consider is the case of lattices with varying height or depth , for example , a 3d lattice with fixed width and height but of varying depth .
although this structure is still strictly three dimensional , when the depth is very low and the width ( height ) is large , the quantum walker will see the structure as almost a basic 2d cartesian lattice .
suprisingly , in this case we see a gradual change in scaling in the maximum probability of the marked state . at low depths of the lattice ,
the scaling is almost the same as the lower spatial dimensional structure gradually changing to the higher dimensional structure scaling as the depth increases to become equal to that of the other dimensions .
this highlights the importance of full symmetry in the quantum walk search algorithm .
we show how the search algorithm is affected by varying connectivity in regular lattices .
we use our simple model of tunnelling to allow us to interpolate between structures such as the square lattice ( @xmath28 ) and the triangular lattice ( @xmath29 ) . with this model , we are able to identify how the prefactors to the scaling of both the maximum probability of the marked state and the time to find the marked state vary with the connectivity of the structure .
the basic scaling of the time to find the marked state , @xmath3 , is not affected by the increase in connectivity but we find the prefactor to this scaling reduces as the connectivity of the structure being searched increases .
this is due to the additional paths the walker can take to coalesce on the marked state , thus increasing the efficiency of the algorithm in both two and three dimensions .
the maximum probability of the marked state is also affected by the connectivity of the underlying structure .
we find that the additional connectivity does not affect the basic scaling of @xmath30 in the two dimensional case . only moving to three spatial dimensions
allows the walker to find the marked state with a constant probability , @xmath31 .
however , we do note that in both two and three dimensions the prefactors to this scaling , in general , increase as the connectivity of the structure increases .
again , this increases the efficiency of the algorithm as it may not have to be repeated so many times .
we also find that the probability of the marked state does not increase uniformly with the additional connectivity .
we see the prefactor in the scaling drop and then recover itself before increasing as the tunnelling strength increases .
this is due to the dynamics of the quantum walk on a structure with some broken symmetry , i.e. low tunnelling strength between vertices .
we briefly investigated the dynamics of the walk by starting the walker in a single location and monitoring how quickly it spread outwards with varying tunnelling strengths .
this confirmed our results for the search algorithm as we found that the spread of the quantum walk also dropped for lower tunnelling strengths before recovering and eventually increasing at higher tunnelling probabilities .
however , this work on the spreading of the walk compared to tunnelling strength is by no means exhaustive and it would be interesting to look more deeply into this in the future .
we studied both two and three dimensional percolation lattices as a way to model disorder in the quantum walk search algorithm .
we are interested in how the algorithm performs with increasing disorder .
we use percolation lattices as a random substrate for the database arrangement we wish to search .
we find , in both the two and three dimensional cases , that as the level of disorder increases , the maximum probability of the marked state decreases . whilst the percolation probability is higher than the critical percolation threshold , the basic scaling of the maximum probability of the marked state matches that of the basic lattice ( in that spatial dimension ) .
once the percolation probability drop to the critical threshold , this scaling changes to that of the line , @xmath15 . this
is expected as at this point the structure is effectively a line .
we also note the prefactor to the scaling of the maximum probability of the marked state increases linearly once the percolation probability is greater than the critical threshold .
the time to find the marked state follows a similar behaviour .
we find that as the disorder increases , the time to find the marked state also increases .
surprisingly though , we note that the quadratic speed up is maintained for a non - trivial level of disorder , before gradually reverting to the classical run time , @xmath2 , as the disorder reaches the critical percolation threshold .
this seems to match the results of @xcite , which show a fractional scaling for the spreading of the quantum walk from a maximal quantum spreading to a classical spreading at and below the critical threshold .
however , this is in contrast to the work of krovi and brun @xcite who highlight the effect of localisation on the quantum walk when defects are introduced into the substrate .
both these factors indicate that the quantum walk search algorithm seems to be more robust to the effects of disorder and symmetry than the basic spreading of the quantum walk .
this could be due to the fact that the initial state of the walker is spread across the whole lattice .
we have seen that the algorithm becomes less efficient as the disorder increases , but at percolation probabilities greater than the critical threshold , the algorithm still seems to be viable , although more amplification of the result may be required .
+ + _ acknowledgments _ the authors would like to thank jiannis pachos and noah linden for useful and interesting discussions .
nl was funded by the uk engineering and physical sciences research council .
vk is funded by a uk royal society university research fellowship .
f. magniez , a. nayak , p. c. richter , and m. santha . .
in _ proceedings of the 20th annual acm - siam symposium on discrete algorithms ( soda ) _ , pages 8695 .
society for industrial and applied mathematics , 2009 . | we numerically study the quantum walk search algorithm of shenvi , kempe and whaley [ pra * 67 * 052307 ] and the factors which affect its efficiency in finding an individual state from an unsorted set .
previous work has focused purely on the effects of the dimensionality of the dataset to be searched . here
, we consider the effects of interpolating between dimensions , connectivity of the dataset , and the possibility of disorder in the underlying substrate : all these factors affect the efficiency of the search algorithm .
we show that , as well as the strong dependence on the spatial dimension of the structure to be searched , there are also secondary dependencies on the connectivity and symmetry of the lattice , with greater connectivity providing a more efficient algorithm .
in addition , we also show that the algorithm can tolerate a non - trivial level of disorder in the underlying substrate . |
giant h ii regions are ideal laboratories to understand the feedback of star - formation on the dynamics and energetics of the interstellar medium ( ism ) .
supernov and stellar winds arising in such regions are reponsible for producing shocks , destroying dust grains and molecules , while compressing molecular clouds and triggering subsequent star - formation .
they also allow the release of newly synthetized elements into the ism , altering its metallicity . in order to study the star - formation properties as a function of the environment , we observed three giant h ii regions spanning a wide range of physical conditions ( gas density , mass , age ) and chemical properties ( metallicity ) with the spitzer space telescope ( werner et al .
observations are part of the gto program pid#63 .
the regions are ngc3603 in the milky way , 30doradus ( hereafter 30dor ) in the large magellanic cloud ( lmc ) , and n66 in the small magellanic cloud ( smc ) .
the scope of this program is to address crucial issues such as the destruction of complex molecules by energetic photons arising from massive stars , the polycyclic aromatic hydrocarbon ( pah ) abundance dependence on metallicity , or conditions that lead to the formation / disruption of massive stellar clusters .
photometry with spitzer / irac ( fazio et al .
2004 ) has been performed and will be discussed in brandl et al .
( in preparation ) . the brightest mid - infrared ( mir ) regions ( knots , stellar clusters , shockfronts , ... ) were followed spectroscopically with the infrared spectrograph ( irs ; houck et al .
2004 ) . in lebouteiller et al .
( 2007 ) , we analyzed the spatial variations of the pah and fine - structure line emission across individual photodissociation regions ( pdrs ) in ngc3603 .
the two other regions will be investigated the same way in follow - up papers ( bernard - salas et al . in preparation ; whelan et al . in preparation ) . in this paper
, we introduce the full irs dataset ( low- and high - resolution ) of the giant h ii regions and we derive their chemical abundances .
a subsequent paper will be focused on the study of molecules and dust properties ( lebouteiller et al . in preparation ) .
elemental abundances in h ii regions are historically derived from optical emission - lines .
large optical telescopes , together with sensitive detectors makes it possible to determine the chemical composition of very faint h ii regions . because of dust extinction , optical spectra only observe ionized gas toward sighlines with low dust content . in this view
, the mir range allows analyzing denser lines of sight , with possibly different chemical properties because of small - scale mixing and/or differential depletion on dust grains .
mir emission - lines constitute the only way to measure abundances in more obscure regions , and these abundances ought to be compared to abundances from the optical range .
although the optical domain gives access to some of the most important elements to constrain nucleosynthethic and stellar yields ( c , n , o , ne , s , ar , fe ) , it does not include some essential ionization stages necessary for abundance determinations of certain elements , such as s iv or ne ii .
the mir range enables the abundance determination of ne , s , and ar , with the most important ionization stages observed .
iron abundance can be also determined from mir forbidden emission - lines , but with considerably larger uncertainty due to ionization corrections . finally , it must be stressed that abundance determinations in the optical are more sensitive to the electronic temperature ( @xmath1 ) determination as compared to the mir range .
the effect of @xmath1 on abundances determinations is a significant source of error in optical abundance results .
wu et al . ( 2007 ) recently studied a sample of blue compact dwarf galaxies ( bcds ) with the irs and found a global agreement between abundances derived from the optical and those derived from the mir .
this suggests that the dense lines of sight probed in the mir have a similar chemical composition as unextincted lines of sight and/or dense regions with possibly peculiar abundances do not contribute significantly to the integrated mir emission - line spectrum .
mir abundances of the bcds were calculated using mostly h@xmath2 or h@xmath0 lines from the optical as tracer of the hydrogen content , with significant uncertainties from aperture corrections , or different observed regions because of extinction .
the present sample of giant h ii regions provides the unique opportunity to measure accurate abundances , with a signal - to - noise ratio sufficiently high to observe directly the h i recombination line at 12.37@xmath3 m .
we provide abundances of ne , s , ar , and fe toward lines of sight with different physical properties ( pdrs , ionized gas , embedded source , stellar cluster , ... ) within each giant h ii region .
we first present the sample of the three giant h ii regions in [ sec : pres ] .
the data reduction and analysis are discussed in [ sec : observations ] .
we infer the ion abundances in [ sec : ionicab ] .
elemental abundances are determined in [ sec : eleab ] and are discussed in [ sec : discussion ] .
ngc3603 is located in the carina arm of the milky way , at around @xmath47kpc from the sun ( see e.g. , moffat et al .
1983 ; de pree et al . 1999 ) .
based on its galactic longitude ( 291.62@xmath5 ) , ngc3603 is at around 8.5kpc from the galactic center , i.e. , only somewhat further away than the sun ( @xmath48kpc ) .
this is the most massive optically visible h ii region in the milky way ( goss & radhakrishnan 1969 ) , being 100 times more luminous than the orion nebula .
ngc3603 is remarkably similar to r136 , the core of 30dor , in terms of its star density profile and its wolf - rayet ( wr ) content ( moffat et al .
ngc3603 has essentially a solar metallicity with @xmath6 ( o / h)@xmath48.39 - 8.52 ( e.g. , melnick et al . 1989 ;
tapia et al . 2001 ; garca - rojas et al .
2006 ; see also [ sec : eleab ] ) .
massive stars in the central stellar cluster influence heavily the surrounding ism morphology through stellar winds , notably by compressing molecular clouds ( nrnberger & stanke 2002 ) .
these massive stars are also responsible for most of the excitation in the h ii region through the large number of ionizing photons , with a lyman continuum flux of @xmath7s@xmath8 ( kennicut 1984 ; drissen et al . 1995 ) .
30dor is located in the large magellanic cloud , at a distance modulus of @xmath9 ( selman et al .
1999 ) i.e. , @xmath10kpc .
it is the largest and most massive h ii region complex in the local group , with the nebula being @xmath11 ( 200pc ) in diameter .
the metallicity of 30dor is a factor of @xmath120.6 below solar ( [ sec : eleab ] ) .
the dense core of very luminous and massive stars is known as r136 ( @xmath122.5pc in diameter ) . in r136 , there are 39 confirmed o3 stars ( massey & hunter 1998 ) as well as several wr stars ( e.g. , melnick 1985 ) .
stars from spectral types o3 to b3 are also detected as far away as 150@xmath13 from r136 ( bosch et al .
2001 ) .
n66 is the largest nebula in the smc ( henize 1956 ) .
the distance to n66 is about that of the smc ( 60.6kpc ; hilditch et al .
many massive stellar clusters are located across the region ( sabbi et al .
2007 ) , but most of the nebular ionization is thought to be due to ngc346 , the largest stellar concentration in the smc ( dreyer 1888 ) .
the metallicity of the ngc346 cluster is @xmath14z@xmath15 ( haser et al .
1998 ; bouret et al . 2003 ) and its age is @xmath123myr ( bouret et al .
the other @xmath16 fainter @xmath16 clusters have similar ages ( sabbi et al .
many h ii regions are located across the nebula , including the compact h ii region n66a , powered by its own stellar cluster .
several dozen o stars are confirmed in ngc346 , at least one of them as early as o3 type ( walborn & blades 1986 ; massey et al .
1989 ) .
llllll pointing & @xmath0 ( j2000 ) & @xmath17 ( j2000 ) & campaign / aorkey & sh / lh & spectral characteristics + ngc3603#1 & 11@xmath1814@[email protected] & -61@xmath519@[email protected] & 22/12080384 & 0.25 & ( sky ) + ngc3603#3 & 11@xmath1814@[email protected] & -61@xmath517@[email protected] & 22/12080384 & 0.27 & pdr + sio - abs + ngc3603#4 & 11@xmath1814@[email protected]&-61@xmath512@[email protected] & 22/12080384 & 0.37 & pdr + ngc3603#5 & 11@xmath1814@[email protected]&-61@xmath515@[email protected] & 22/12080384 & 0.32 & pdr + sio - abs + ngc3603#6 & 11@xmath1815@[email protected]&-61@xmath515@[email protected] & 22/12080384 & 0.26 & stellar , sio - em ( hd97950 ) + ngc3603#7 & 11@xmath1815@[email protected]&-61@xmath516@[email protected] & 22/12080384 & 0.27 & ig + ngc3603#8 & 11@xmath1815@[email protected]&-61@xmath515@[email protected] & 22/12080384 & 1 & pdr + ig + ngc3603#9 & 11@xmath1815@[email protected]&-61@xmath516@[email protected] & 22/12080384 & 0.44 & ig ( irs9a ) + 30dor#1 & 5@xmath1838@[email protected]&-69@xmath510@[email protected] & 21.1/12081152 & 0.02 & ( sky ) + 30dor#2 & 5@xmath1839@[email protected]&-69@xmath508@[email protected] & 21.1/12081408 & 0.29 & pdr + 30dor#3 & 5@xmath1838@[email protected]&-69@xmath506@[email protected] & 21.1/12081408 & 0.28 & pdr + ig ( r136 ) + 30dor#4 & 5@xmath1838@[email protected]&-69@xmath506@[email protected] & 21.1/12081408 & 0.375 & ig + 30dor#5 & 5@xmath1839@[email protected]&-69@xmath504@[email protected] & 21.1/12081408 & 0.28 & pdr + 30dor#6 & 5@xmath1838@[email protected] & -69@xmath506@[email protected] & 21.1/12081408 & 0.24 & pdr + 30dor#7 & 5@xmath1838@[email protected] & -69@xmath506@[email protected] & 21.1/12081408 & 0.31 & pdr + ig ( p644 ) + 30dor#8 & 5@xmath1838@[email protected]&-69@xmath504@[email protected] & 21.1/12081408 & 0.36 & pdr ( p4 ) + 30dor#10 & 5@xmath1838@[email protected]&-69@xmath502@xmath2114.@xmath134 & 21.1/12081152 & 0.60 & pdr + 30dor#11 & 5@xmath1838@[email protected]&-69@xmath504@[email protected] & 21.1/12081152 & 0.39 & pdr + ig + 30dor#12 & 5@xmath1838@[email protected]&-69@xmath504@[email protected] & 21.1/12081152 & 0.37 & pdr + sio - abs + 30dor#13 & 5@xmath1838@[email protected]&-69@xmath505@[email protected] & 21.1/12081152 & 0.29 & sio - em + 30dor#14 & 5@xmath1838@[email protected]&-69@xmath504@[email protected] & 21.1/12081152 & 0.35 & pdr ( p1429 ) + 30dor#15 & 5@xmath1838@[email protected]&-69@xmath505@[email protected] & 21.1/12081152 & 0.31 & pdr ( p1 ) + 30dor#16 & 5@xmath1838@[email protected]&-69@xmath505@[email protected] & 21.1/12081152 & 0.32 & ig ( p3 ) + 30dor#17 & 5@xmath1837@[email protected]&-69@xmath511@[email protected] & 21.1/12081152 & 0.65 & sio - abs ( 2mass j05375004 - 6911075 ) + n66#1 & 0@xmath1859@[email protected]&-72@xmath510@[email protected] & 8/4384768 & 0.18 & pdr + ig + n66#2 & 0@xmath1859@[email protected]&-72@xmath510@[email protected] & 8/438476 & 0.12 & pdr + n66#3 & 0@xmath1859@[email protected]&-72@xmath511@[email protected] & 8/438476 & 0.25 & pdr + n66#5 & 0@xmath1859@[email protected]&-72@xmath510@[email protected] & 26/16207872 & 1.06 & sio - abs + n66#6 & 0@xmath1859@[email protected]&-72@xmath510@[email protected] & 26/16207872 & 1.23 & sio - em ( ngc346 ) + n66#7 & 0@xmath1859@[email protected]&-72@xmath511@[email protected] & 26/16207872 & 1 & pdr + ig + n66#8 & 0@xmath1859@[email protected]&-72@xmath511@[email protected] & 26/16207872 & 0.75 & pdr + n66#9 & 0@xmath1859@[email protected]&-72@xmath511@[email protected] & 26/16207872 & 0.89 & pdr + n66#10 & 0@xmath1858@[email protected]&-72@xmath509@[email protected] & 26/16207872 & 1 & pdr + n66#11 & 0@xmath1859@[email protected]&-72@xmath509@[email protected] & 26/16207872 & 1 & pdr + ig + sio - em + n66#12 & 0@xmath1858@[email protected]&-72@xmath510@[email protected] & 26/16207872 & 0.54 & pdr + ig + n66#13 & 0@xmath1859@[email protected]&-72@xmath511@[email protected] & 26/16207872 & 1 & pdr + an observation log is presented in table [ tab : observations ] where we report the coordinates of each position , the module scaling factor ( [ sec : stitching ] ) , and the spectral characteristics .
a total of seven positions were observed in ngc3603 ( fig .
[ fig : irac3603 ] ) , fifteen positions in 30dor ( fig .
[ fig : irac30dor ] ) , and twelve positions in n66 ( fig .
[ fig : irac346 ] ) .
we focus our discussion on the observations from the sh and lh modules , which cover 9.9 - 19.6@xmath3 m and 18.7 - 37.2@xmath3 m respectively , with a spectral resolution of @xmath22 .
observations from the sl module ( 5.2 - 14.5@xmath3 m ; @xmath23 ) were used to extend the spectral coverage shortward of 9.9@xmath3 m .
we also included the first irs observation of n66 , originally designed to probe mir bright knots as part of pid#63 , but which resulted in a mispointing . these observations ( n66#1 , n66#2 , and n66#3 ; table [ tab : observations ] ) probe relatively low - excitation regions , with a few arcseconds offset from the originally intended mir sources .
the two - dimensional detector images were processed by the spitzer science center s pipeline reduction software ( version s13.2 ) .
we used the the basic calibrated data product .
rogue pixels and on - the - fly flagged data were removed using irsclean .
no sky - subtraction was performed for 30dor and n66 since the observations initially designed to be background images include prominent mir emission features ( lines and pahs ) . it must be noted that the regions are bright enough that the lack of background subtraction does not affect the measurements .
we took instead the opportunity to use these background spectra and investigate low - excitation regions .
source # 1 was used for sky - subtraction in ngc3603 ; it shows extremely weak emission - lines , with fluxes less than 1% that of lines in other positions .
high - resolution spectra were extracted from the full sh and lh apertures using the smart software developped at cornell ( higdon et al .
the low - resolution sl spectra were taken from lebouteiller et al .
( 2007 ) for ngc3603 , bernard - salas et al .
( in preparation ) for 30dor , and whelan et al .
( in preparation ) for n66 . the spectra from the sh and lh modules had to be scaled for most observations because the aperture sizes are different and because the sources sampled are not unresolved .
the sh aperture covers @xmath24 on the sky while the lh one covers @xmath25 and the two apertures are perpendicular to each other . if these apertures were uniformly illuminated , this would result in a correction factor of @xmath26 in the ratio of the sh to lh flux .
we applied a scaling factor to the flux in the lh observations to scale to the sh ones .
note that we align the dust continuum , and that the lines might behave differently ( see the discussion in
[ sec : density ] ) .
correction factors are reported in table[tab : observations ] and range from @xmath120.25 to 1 .
the only source for which the factor is significantly greater than 1 is n66#6 .
the sh observation contains an additional source as compared to the lh one , which explains the additional flux observed in the sh spectrum . for several other sources in n66 ,
no correction was needed , implying that the extent of those mir sources is smaller than the sh aperture .
the mir sources in ngc3603 and 30dor show sh / lh scaling factors usually between @xmath120.3 and @xmath120.4 , i.e. , consistent with extended emission .
the distance to 30dor and n66 is similar , but these objects show strong differences in the extent of their mir sources .
this suggests that the mir emission in n66 is mostly concentrated in small knots and that there is no intense extended dust continuum emitting across the whole nebula ( see also whelan et al . in preparation ) .
we also aligned spectra from the sl module to stitch those from the sh / lh modules .
final spectra are presented in the appendix .
we illustrate in fig .
[ fig : lines ] the detection of the most important lines in the mir spectrum of one observation .
line fluxes are presented in table [ tab : line_fluxes ] . in order to infer the flux from a given emission line
, we adjusted a first - order continuum to fit the data on both sides of the line .
the line was then fitted with a gaussian profile .
measurements were done in the combined data where the spectral orders and the two nod positions have been co - added .
llllllllllllll line & [ s iv ] & [ ne ii ] & [ ne iii ] & [ s iii ] & [ ar iii ] & [ s iii ] & [ ne iii ] & [ fe ii ] & [ fe ii ] & [ fe iii ] & [ si ii ] & h i + @xmath27 ( @xmath3 m ) & 10.51 & 12.81 & 15.56 & 18.71 & 21.83 & 33.48 & 36.01 & 17.89 & 26.00 & 22.93 & 34.82 & 12.37 + ngc3603#3 & 53.83 & 134.45 & 181.10 & 142.69 & 5.60 & 291.98 & 24.35 & 1.94 & 3.05 & 3.23 & 55.33 & 3.50 + ngc3603#4 & 49.30 & 280.13 & 162.93 & 399.72 & 7.82 & 542.34 & 29.58 & 2.48 & 8.33 & 9.55 & 137.78 & 5.90 + ngc3603#5 & 83.81 & 89.73 & 193.42 & 209.58 & 6.27 & 304.99 & 28.50 & 1.23 & 3.66 & 3.30 & 58.82 & 2.98 + ngc3603#6 & 20.98 & 11.77 & 36.80 & 35.40 & 1.45 & 52.25 & 6.85 & @xmath280.42 & @xmath281.34 & 0.86 & 8.94 & 5.86 : + ngc3603#7 & 1075.34 & 140.92 & & 833.37 & & & 134.64 & @xmath287.75 & & & @xmath2824.38 & 14.07 + ngc3603#8 & 1725.83 & 429.63 & & & & & 541.07 & @xmath2816.7 & & & @xmath28131.23 & 21.32 + ngc3603#9 & 521.82 & 246.0 & & 635.09 & & & 153.30 & @xmath2811.73 & & & @xmath2896.26 & 6.63 : + 30dor#2 & 4.35 & 3.18 & 12.50 & 9.83 & 0.26 & 16.80 & 1.22 & 0.08 & 0.07 & 0.19 & 2.95 & 0.24 + 30dor#3 & 36.12 & 8.00 & 60.19 & 37.77 & 0.94 & 59.25 & 5.73 & 0.31 & 0.18 & 0.60 & 7.16 & 1.00 + 30dor#4 & 102.71 & 7.84 & 107.02 & 53.26 & 1.52 & 68.58 & 8.68 & 0.58 & @xmath280.50 & 1.29 & 3.43 & 1.56 + 30dor#5 & 8.63 & 5.94 & 20.99 & 17.34 & 0.51 & 31.71 & 2.44 & 0.28 & 0.37 & 0.39 & 9.46 & 0.37 + 30dor#6 & 15.67 & 7.08 & 33.33 & 25.05 & 0.51 & 41.45 & 3.69 & 0.19 & 0.86 & 0.84 & 8.92 & 0.50 + 30dor#7 & 105.95 & 15.05 & 143.96 & 79.73 & 2.59 & 109.26 & 12.39 & 0.58 & 0.92 & 1.91 & 8.85 & 2.37 + 30dor#8 & 18.31 & 34.39 & 62.07 & 64.08 & 2.05 & 111.14 & 7.10 & 0.62 & 1.93 & 2.02 & 24.33 & 1.74 + 30dor#10 & 13.6 & 39.57 & 31.68 & 54.29 & 1.50 & 66.96 & 3.74 & 0.37 & 1.53 & 1.47 & 19.63 & 1.40 + 30dor#11 & 32.39 & 14.35 & 74.99 & 54.82 & 1.63 & 98.25 & 8.45 & 0.54 & 0.84 & 1.33 & 11.72 & 1.15 + 30dor#12 & 31.75 & 6.26 & 46.80 & 28.02 & 0.98 & 64.03 & 6.26 & 0.34 & 0.60 & 0.68 & 11.93 & 0.66 + 30dor#13 & 7.63 & 2.61 & 14.90 & 10.73 & 0.35 & 22.87 & 2.66 & 0.10 & @xmath280.15 & 0.29 & 3.49 & 0.27 + 30dor#14 & 131.11 & 27.69 & 207.41 & 117.35 & 3.14 & 168.87 & 15.93 & 0.86 & 2.75 & 2.54 & 34.57 & 3.09 + 30dor#15 & 151.33 & 28.29 & 245.48 & 142.31 & 4.55 & 222.20 & 22.61 & 0.83 & 2.40 & 3.10 & 24.24 & 3.13 + 30dor#16 & 150.39 & 14.23 & 183.24 & 97.64 & 2.92 & 126.17 & 14.16 & 0.81 & @xmath281.41 & 1.95 & 10.66 & 2.78 + 30dor#17 & 1.74 & 7.75 & 7.08 & 10.18 & 0.53 & 34.65 & 1.58 & 0.20 & 1.65 & 1.07 & 13.20 & 0.32 + n66#1 & 3.06 & 0.31 & 3.24 & 2.24 & 0.05 & 2.03 & 0.25 & 0.26 & @xmath280.05 & 0.01 & 0.54 & 0.09 + n66#2 & 2.31 & 0.28 & 2.41 & 1.64 & 0.03 & 1.03 & 0.13 & 0.11 & @xmath280.04 & @xmath280.03 & 0.37 & 0.083 + n66#3 & 1.46 & 0.12 & 1.47 & 0.95 & 0.04 & 1.18 & 0.16 & 0.04 & @xmath280.04 & @xmath280.03 & 0.31 & @xmath280.01 + n66#5 & 2.78 & 0.42 & 3.59 & 2.41 & 0.13 & 6.25 & 0.64 & 0.16 & @xmath280.08 & @xmath280.10 & 1.40 & 0.15 + n66#6 & 2.53 & 0.69 & 2.65 & 2.11 & 0.15 & 5.64 & 0.64 & @xmath280.11 & 0.24 & @xmath280.14 & 2.54 & 0.10 + n66#7 & 1.73 & 0.18 & 2.12 & 1.43 & 0.12 & 3.90 & 0.52 & @xmath280.11 & 0.22 & @xmath280.14 & 2.52 & 0.13 + n66#8 & 3.09 & 0.88 & 4.49 & 3.32 & 0.10 & 4.05 & 0.39 & @xmath280.06 & 0.07 & 0.04 & 0.84 & 0.16 + n66#9 & 1.49 & 0.55 & 2.53 & 1.89 & 0.07 & 3.25 & 0.44 & @xmath280.07 & @xmath280.07 & @xmath280.08 & 1.28 & 0.09 + n66#10 & 1.50 & 0.55 & 2.52 & 1.87 & 0.07 & 3.25 & 0.52 & @xmath280.07 & @xmath280.07 & @xmath280.08 & 1.29 & 0.10 + n66#11 & 1.49 & 0.55 & 2.51 & 1.94 & 0.07 & 3.27 & 0.29 & @xmath280.07 & @xmath280.07 & @xmath280.08 & 1.25 & 0.08 + n66#12 & 3.54 & 0.26 & 3.11 & 1.65 & 0.09 & 2.30 & 0.32 & @xmath280.10 & @xmath280.05 & @xmath280.04 & 0.48 & 0.13 + n66#13 & 2.20 & 0.38 & 2.88 & 2.04 & 0.08 & 4.36 & 0.42 & @xmath280.10 & @xmath280.07 & @xmath280.08 & 0.97 & 0.12 + the [ ne ii ] line at 12.81@xmath3 m is blended with a pah feature at @xmath1212.7@xmath3 m , and for this particular case we forced the continuum to fit the data around 12.75@xmath3 m . the observations ngc3603#6 and ngc3603#9 show a broad emission bump at @xmath1212.4@xmath3 m @xmath16 seen in both low - resolution and high - resolution spectra @xmath16 which is possibly stellar in origin .
the 12.28@xmath3 m h@xmath29 line and the 12.37@xmath3 m h i line are blended with this broad feature , and their contribution to the integrated emission is not clearly visible . for this reason , it was not possible to reliably measure their flux .
the statistical errors from the fit are smaller than other uncertainties such as errors on the flux calibration and stitching of the sh and lh module spectra ( [ sec : stitching ] ) .
we estimate the total measurement errors to range from @xmath1210% ( for h i , [ ne ii ] , [ ne iii ] , [ si ii ] , [ s iii ] , [ s iv ] , and [ fe ii ] ) to @xmath1220% ( for [ ar iii ] and [ fe iii ] ) .
fine - structure lines can be used to measure the ionic abundances relative to hydrogen . to derive the ionized hydrogen content , we made use of the h i recombination line at 12.372@xmath3 m ( humphreys @xmath0 , 7@xmath306 ) .
the hu@xmath0 line is detected in all the positions of each object , except in n66#3 , where we calculated an upper limit ( table [ tab : line_fluxes ] ) .
the positions ngc3603#6 and ngc3603#9 show a broad stellar emission bump preventing a reliable h i line flux estimate ( [ sec : measurements ] ) .
for all the other positions , the hu@xmath0 line is blended with another , relatively weaker , h i line at 12.387@xmath3 m ( 11@xmath308 transition ) .
we estimated and corrected for the contribution of this line using the tables of hummer & storey ( 1987 ) for case b recombination to calculate the flux ratio h i @xmath31/h i @xmath32 under the physical conditions described in [ sec : density ] .
the contribution of the h i @xmath31 line that we corrected for is @xmath33 . to determine the ionic abundance
, we first estimate the h@xmath2 flux from the hu@xmath0 line using the tables of hummer & storey ( 1987 ) for a given electron density and temperature .
we then infer the ionic abundances using the method described in bernard - salas et al .
the ionic abundance is defined as : @xmath34 where @xmath35 is the density of protons , @xmath36/ @xmath37 is the ratio of observed intensities , @xmath38 is the wavelength of the line and @xmath39 is the wavelength of h@xmath2 , @xmath40 is the effective recombination coefficient for h@xmath2 , a@xmath41 is the einstein spontaneous transition rate for the line and finally @xmath42/ @xmath43 is the ratio of the population of the level from which the line originates to the total population of the ion .
this ratio is obtained by solving the statistical equilibrium equation for a five level system and normalizing the total number of ions ( osterbrock 1989 ) .
collisional strengths were taken from the iron projectpradhan / ip.html_. ] ( hummer et al .
1993 ) .
+ in order to calculate the ionic abundances , it is necessary to estimate the electron density ( @xmath44 ) and the temperature ( @xmath1 ) in the nebula .
it is possible to estimate the electron density notably through diagnostics provided by the measurements of different lines from the same ionization stage of a given ion . by combining sh and lh observations , we have access to two independent line ratios : [ s iii]@xmath45/[s iii]@xmath46 , and [ ne iii]@xmath47/[ne iii]@xmath48 .
the [ ne iii ] line ratio is irrelevant for our data because it is sensitive to much higher densities ( @xmath49@xmath50 ) than what is usually found in giant h ii regions .
the [ s iii ] line ratio is sensitive to lower densities , but most of our data points lie in the flat regime where no reliable density can be determined ( fig .
[ fig : density]a ) .
it must be stressed that the [ s iii ] lines from which we derive the density do not belong in the same module .
hence the [ s iii ] line ratio determination is strongly affected by uncertainties in the module scaling factor . although the spectra were stitched to align ( [ sec : stitching ] ) , the regions observed in the two modules are not necessarily the same and might not share similar physical conditions .
we aligned the sh and lh spectra based on the dust continuum and it is likely that the line fluxes do not scale accordingly .
in fact , it can be seen in fig .
[ fig : density]b that the agreement between the electron density determinations from the mir and from the optical lines improves as the sh / lh module scaling factor reaches large values ( @xmath51 ) .
note that the outlier n66#11 is characterized by a highly uncertain density determination because of the extremely small [ s iii ] line ratio , in the low - density regime . as a conclusion ,
the stitching of the sh and lh module spectra implies significant systematic uncertainties on the electron density determinations .
we decided to use instead the values from the optical studies for the abundance determinations in [ sec : ionicab_results ] .
the electron temperature values were also taken from the optical analysis .
we assumed @xmath52k and @xmath53@xmath50 for ngc3603 ( melnick et al .
1989 ; garca - rojas et al .
2006 ) , @xmath52k and @xmath54@xmath50 for 30dor ( kurt & dufour 1998 ; peimbert 2003 ; tsamis et al . 2003 ; vermeij & van der hulst 2002 ) , and @xmath55k and @xmath54@xmath50 for n66 ( kurt & dufour 1998 ; peimbert et al .
the electron density and temperature were assumed to be uniform across each object .
the influence of possible variations of @xmath44 and @xmath1 within the gaseous nebula is discussed in [ sec : ionicab_results ] .
llllllll & ne ii / h & ne iii / h & s iii / h & s iv / h & ar iii / h & fe ii / h & fe iii / h + ngc3603#3 & 5.44e-5 & 3.53e-5 & 4.86e-6 & 3.57e-7 & 2.60e-6 & 2.50e-7 & 3.37e-7 + ngc3603#4 & 6.74e-5 & 1.89e-5 & 8.09e-6 & 1.94e-7 & 2.16e-6 & 4.06e-7 & 5.92e-7 + ngc3603#5 & 4.27e-5 & 4.43e-5 & 8.39e-6 & 6.53e-7 & 3.43e-6 & 3.53e-7 & 4.05e-7 + ngc3603#6 & @xmath284.8e-6 & 7.16e-6 & @xmath281.2e-6 & @xmath281.4e-7 & @xmath286.7e-7 & @xmath281.1e-7 & @xmath289.0e-8 + ngc3603#7 & @xmath281.4e-5 & @xmath287.54e-5 & @xmath287.1e-6 & @xmath281.8e-6 & @xmath282.3e-5 & @xmath288.9e-7 & @xmath282.0e-6 + ngc3603#8 & @xmath282.9e-5 & @xmath282.00e-4 & & @xmath281.9e-6 & @xmath285.6e-5 & @xmath286.1e-6 & @xmath281.6e-5 + ngc3603#9 & @xmath285.3e-5 & @xmath281.82e-4 & @xmath281.1e-5 & @xmath281.8e-6 & @xmath284.9e-5 & @xmath284.e-10 & @xmath281.1e-5 + ngc3603 ( average ) & 5.48e-5 & 2.64e-5 & 7.11e-6 & 4.01e-7 & 2.73e-6 & 3.36e-7 & 4.45e-7 + ( dispersion , % ) & 10.6 & 27.2 & 13.0 & 27.3 & 11.1 & 11.1 & 14.0 + 30dor#2 & 1.91e-5 & 3.58e-5 & 5.32e-6 & 4.06e-7 & 1.75e-6 & 7.86e-8 & 2.77e-7 + 30dor#3 & 1.15e-5 & 4.12e-5 & 4.88e-6 & 8.06e-7 & 1.51e-6 & 4.83e-8 & 2.09e-7 + 30dor#4 & 7.24e-6 & 4.72e-5 & 4.43e-6 & 1.48e-6 & 1.57e-6 & @xmath288.6e-8 & 2.90e-7 + 30dor#5 & 2.31e-5 & 3.89e-5 & 6.07e-6 & 5.21e-7 & 2.22e-6 & 2.69e-7 & 3.68e-7 + 30dor#6 & 2.03e-5 & 4.56e-5 & 6.48e-6 & 7.00e-7 & 1.64e-6 & 4.62e-7 & 5.86e-7 + 30dor#7 & 9.11e-6&4.16e-5 & 4.35e-6 & 9.98e-7 & 1.76e-6 & 1.04e-7 & 2.81e-7 + 30dor#8 & 2.84e-5&2.44e-5 & 4.77e-6 & 2.35e-7 & 1.90e-6 & 2.98e-7 & 4.05e-7 + 30dor#10 & 4.05e-5 & 1.55e-5 & 5.01e-6 & 2.17e-7 & 1.72e-6 & 2.93e-7 & 3.66e-7 + 30dor#11 & 1.79e-5 & 4.46e-5 & 6.15e-6 & 6.27e-7 & 2.27e-6 & 1.96e-7 & 4.03e-7 + 30dor#12 & 1.35e-5 & 4.83e-5 & 5.47e-6 & 1.07e-6 & 2.38e-6 & 2.43e-7 & 3.58e-7 + 30dor#13 & 1.38e-5 & 3.75e-5 & 5.11e-6 & 6.27e-7 & 2.07e-6 & @xmath281.5e-7 & 3.73e-7 + 30dor#14 & 1.29e-5 & 4.60e-5 & 4.92e-6 & 9.48e-7 & 1.64e-6 & 2.39e-7 & 2.87e-7 + 30dor#15 & 1.30e-5 & 5.37e-5 & 5.88e-6 & 1.08e-6 & 2.34e-6 & 2.06e-7 & 3.46e-7 + 30dor#16 & 7.34e-6 & 4.51e-5 & 4.54e-6 & 1.21e-6 & 1.69e-6 & @xmath281.4e-7 & 2.45e-7 + 30dor#17 & 3.53e-5 & 1.54e-5 & 4.18e-6 & 1.23e-7 & 2.70e-6 & 1.40e-6 & 1.18e-6 + 30dor ( average ) & 1.82e-5 & 3.87e-5 & 5.11e-6 & 7.37e-7 & 1.94e-6 & 3.20e-7 & 3.99e-7 + ( dispersion % ) & 13.6 & 7.5 & 3.3 & 13.4 & 4.6 & 30.9 & 14.6 + n66#1 & 4.09e-6 & 2.05e-5 & 2.51e-6 & 6.31e-7 & 7.12e-7 & @xmath281.3e-7 & 3.13e-8 + n66#2 & 4.64e-6 & 1.91e-5 & 2.30e-6 & 5.99e-7 & 5.37e-7 & @xmath281.6e-7 & @xmath281.2e-7 + n66#3 & @xmath282.0e-5 & @xmath286.72e-5 & @xmath287.71e-6 & @xmath282.18e-6 & @xmath284.12e-6 & @xmath287.3e-7 & @xmath286.8e-7 + n66#5 & 3.52e-6 & 1.44e-5 & 1.72e-6 & 3.65e-7 & 1.18e-6 & @xmath281.3e-7 & @xmath282.0e-7 + n66#6 & 8.69e-6 & 1.60e-5 & 2.26e-6 & 4.99e-7 & 2.06e-6 & 5.80e-7 & @xmath284.2e-7 + n66#7 & 2.52e-6 & 1.59e-5 & 2.30e-6 & 5.01e-7 & 1.25e-6 & 4.99e-7 & @xmath283.9e-7 + n66#8 & 6.92e-6 & 1.69e-5 & 2.22e-6 & 3.81e-7 & 8.50e-7 & 1.06e-7 & 7.48e-8 + n66#9 & 7.69e-6 & 1.70e-5 & 2.25e-6 & 3.26e-7 & 1.06e-6 & @xmath281.9e-7 & @xmath282.7e-7 + n66#10 & 6.92e-6 & 1.52e-5 & 2.00e-6 & 2.96e-7 & 9.52e-7 & @xmath281.7e-7 & @xmath282.4e-7 + n66#11 & 8.05e-6 & 1.76e-5 & 2.42e-6 & 3.41e-7 & 1.11e-6 & @xmath282.0e-7 & @xmath282.8e-7 + n66#12 & 2.59e-6 & 1.38e-5 & 1.30e-6 & 5.13e-7 & 9.00e-7 & @xmath288.9e-8 & @xmath288.8e-8 + n66#13 & 3.87e-6 & 1.40e-5 & 2.57e-6 & 3.50e-7 & 8.80e-7 & @xmath281.4e-7 & @xmath281.9e-7 + n66 ( average ) & 5.41e-6 & 1.64e-5 & 1.68e-6 & 4.36e-7 & 1.04e-6 & 3.95e-7 & 5.30e-7 + ( dispersion % ) & 12.1 & 3.7 & 5.4 & 7.6 & 10.9 & 30.2 & 29.0 + lllllll ne ii / h & ne iii / h & s iii / h & s iv / h & ar iii / h & reference & comment + + 5.48e-5 & 2.64e-5 & 7.11e-6 & 4.01e-7 & 2.73e-6 & ( this study ) & average ( 3 - 4 positions ) + & 1.00e-4 & 1.29e-5 & & 3.63e-6 & 1 & optical + + 1.82e-5 & 3.87e-5 & 5.11e-6 & 7.37e-7 & 1.94e-6 & ( this study ) & average ( 15 positions ) + & 5.89e-5 & 5.50e-6 & & 1.51e-6 & 4 & optical + 3.71e-5 & & 4.28e-6 & & 1.11e-6 & 5 & optical + 2.33e-5 & 5.34e-5 & 7.56e-6 & 1.32e-6 & 1.34e-6 & 2 & position # 1 , iso + & & 3.52e-6 & & 9.37e-7 & 2 & position # 1 , optical + 2.02e-5 & 6.04e-5 & 6.15e-6 & 1.46e-6 & 1.14e-6&2 & position # 2 , iso + & 3.63e-5 & 3.19e-6 & & 8.85e-7 & 2 & position # 2 , optical + 2.03e-5 & 5.84e-5 & 5.54e-6 & 1.48e-6 & 1.45e-6&2 & position # 3 , iso + & 3.79e-5 & 4.36e-6 & & 1.16e-6 & 2 & position # 3 , optical + 3.02e-5 & 7.37e-5 & 8.73e-6 & 1.27e-6 & 1.50e-6 & 2 & position # 4 , iso + & 3.88e-5 & 3.36e-6 & & 8.46e-7 & 2 & position # 4 , optical + + 5.41e-6 & 1.64e-5 & 1.68e-6 & 4.36e-7 & 1.04e-6 & ( this study ) & average ( 11 positions ) + 1.27e-5 & 1.74e-5 & 1.30e-6 & 4.19e-7 & 6.57e-7 & 2 & n66 , iso + & 1.57e-5 & 1.53e-6 & & 3.81e-7 & 2 & n66 , optical + & 1.35e-5 & 1.41e-6 & & 3.89e-7 & 3 & region a , optical + ionic abundances assuming uniform electron density and temperature are presented in table [ tab : ionicab ] . for comparison
, we report in table [ tab : ionicab_global ] the values derived from the optical . given the line flux uncertainties discussed in [ sec : measurements ] , we consider the measurement error on the ionic abundance determination to range from @xmath1215% ( for ne ii , ne iii , s iii , s iv , and fe ii ) to @xmath1225% ( for ar iii and fe iii ) .
additional systematic errors on the method are due to the assumed physical conditions .
it is likely that @xmath44 and @xmath1 vary across a single giant h ii region , and along a given line of sight . it has been argued that @xmath44 in galactic and extragalactic h ii regions is higher toward the brightest regions , reaching a few 100@xmath50 , while the faintest galactic h ii regions are characterized by uniform electron density on the order of 20 - 140@xmath50 ( e.g. , castaneda et al .
1992 ; copetti et al .
it can be seen in fig.[fig : ionicab_var ] that the ionic abundance determinations of ne ii and ne iii in the giant h ii regions of our sample are fairly insensitive to electron density in the range @xmath56-@xmath57@xmath50 .
the other ions do not show strong variations in their abundance for densities smaller than @xmath58@xmath50 .
furthermore , low - excitation faint regions in n66 ( n66#1 and n66#2 ) do not show ionic abundance determinations particularly lower than the other @xmath16 brighter @xmath16 positions in this object .
we conclude that internal variations ( or uncertainties ) in @xmath44 do not affect significantly the ionic abundance determinations . on the other hand , variations of the electron temperature may have a significant impact on the abundance determinations .
the uncertainty using different electron temperatures is much reduced when analyzing mir fine - structure lines as compared to the use of lines in the optical spectrum ( bernard - salas et al .
2001 ) . in first approximation ,
the mir ionic abundance determinations vary linearly with @xmath1 .
all the ionic abundances show similar trends with @xmath1 in fig.[fig : ionicab_var ] because of the dependence of the recombination coefficient of h@xmath2 with @xmath1 ( [ sec : ionicab_method ] ) . in gaseous nebulae of giant
h ii regions , the temperature is usually between 7500k and 15000k , while temperature fluctuations are on the order of 15 - 20% ( esteban et al .
2002 ; odell et al . 2003 ) .
we conclude that there might be a systematic error on the ionic abundance determinations up to 20% if @xmath1 is not uniform and vary by as much as 20% .
the final uncertainty on the ionic abundances is given by the sum of measurement errors and errors due to electron temperature variation .
total errors range from @xmath5915%@xmath5920% ( @[email protected] ) for ne iii , ne iii , s iii , s iv , and fe ii , to @xmath5920%@xmath5920% ( @[email protected] ) for ar iii and fe iii .
the dispersion of the ionic abundances we derive across the giant h ii regions ranges from @xmath123% to @xmath1230% ( table [ tab : ionicab ] ) , which is smaller than the total uncertainty ( @xmath1235 - 40% ) .
this implies that ( 1 ) the errors on the ionic abundance could be somewhat overestimated , and ( 2 ) electron temperature are unlikely to vary by more than 20% across a given giant h ii region .
the mir range gives the unique opportunity to observe the most important ionization stages of elements such as ne , s , and ar . as a result
, the elemental abundance determination requires no @xmath16 or little @xmath16 ionization correction factors .
iron abundance determination is much more uncertain because we do not observe the dominant ion in the ionized gas ( fe iv ) .
finally , we could not measure the total abundance of silicon , since we have access only to the si ii ion .
lrrrrr & i & ii & iii & iv & v h & 13.60 & & & & ne & 21.56 & 40.96 & 63.45 & 97.11 & 126.21 si & 8.15 & 16.34 & 33.49 & 45.15 & 166.77 s & 10.36 & 23.33 & 34.83 & 47.30 & 72.68 ar & 15.76 & 27.63 & 40.74 & 59.81 & 75.02 fe & 7.87 & 16.18 & 30.64 & 54.8 & 75.0 the total abundance of sulfur was calculated using the sum of the s iii and s iv ionization stages .
we used the [ s iii ] line at 18.71@xmath3 m instead of the one at 33.48@xmath3 m to estimate the ionic abundance of s iii , because the 18.71@xmath3 m line is measured in the same module as h i @xmath32 , hence allowing us to avoid aperture effects .
in addition , the 18.71@xmath3 m line is much less sensitive to electron density variations ( fig .
[ fig : ionicab_var ] ) .
no ionization corrections were made due to the presence of other ionization stages .
we can not exclude that some s ii is present in the ionized gas given the ionization potential ( ip ) of s ii ( table [ tab : ip ] ) .
the photoionization cross - section ( pics ) of s i is actually higher than that of s ii for energies @xmath6050ev ( verner et al .
it is however unknown which fraction of s ii resides in the ionized gas , in the neutral gas , or in the associated pdrs .
photoionization models as well as optical observations support the predominancy of s iii and s iv stages
. models of tsamis et al .
( 2005 ) for 30dor predict that @xmath1292% of sulfur is into s iii and s iv , while 8% is due to other ionization stages , mostly s ii .
peimbert ( 2003 ) and vermeij & van der hulst ( 2002 ) find consistent results observationaly , with s ii / s iii ranging from 3% to 8.5% . in n66 , the contribution of s ii is approximately 15% that of s iii ( peimbert et al .
2000 ; vermeij & van der hulst 2002 ) .
finally , in ngc3603 , s ii / s iii is only 1.2% ( garca - rojas et al .
hence , we might underestimate the sulfur abundance by @xmath1210% in 30dor and n66 while the ionization correction factor ( icf ) in ngc3603 is negligible .
the final uncertainty on s / h is due to measurement errors on the ionic abundances of s iii and s iv ( 15% each ; [ sec : measurements ] ) , and to the assumed physical conditions ( @xmath6020% , which affect s iii and s iv the same way , [ sec : ionicab_results ] ) .
the abundance of neon was calculated using both ne ii and ne iii ions .
we use the [ ne iii ] line at 15.56@xmath3 m instead of the one at 36.01@xmath3 m because it resides in the same module as the h i @xmath32 line , and because the 15.56@xmath3 m line is comparatively less sensitive to electron density variations ( fig .
[ fig : ionicab_var ] ) . as far as higher ionization stages are concerned
, the contribution of ne iv is expected to be negligible since the ip of ne iii is 63.45ev , i.e. , above the he ii absorption edge at 54.4ev in ionizing stars .
the pics of ne iii is actually smaller than that of ne ii for energies @xmath6070ev ( verner et al .
1996 ) , implying that ne iv should not exist in significant amount .
moreover , the [ o iv ] line at 25.89@xmath3 m is not detected in any of our spectra but one which will be discussed in lebouteiller et al .
( in preparation ) .
the ip of o iii is 54.9ev , as compared to 63.4ev for ne iii , hence the absence of o iv implies the absence of ne iv .
models of 30dor confirm these findings and predict that ne iv represents @xmath61% of the total neon , as compared to 86% for ne iii and 14% for ne ii ( tsamis et al .
. the proportion of ne iv should be negligible also in n66 and ngc3603 , resulting in similar negligible ionization corrections .
more specifically , given the fact that the global interstellar radiation field ( isrf ) hardness in n66 and ngc3603 is similar or lower than that in 30dor ( lebouteiller et al .
in preparation ) , we do not expect any significant contribution of ne iv in any of our objects . similarly to sulfur abundance ( [ sec : sulfurdet ] ) , the error on ne / h is @xmath5930% due to measurement errors , and @xmath6020% due to the assumed physical conditions .
llll & [ ar ii ] & ar ii / h & ar ii / ar iii + ngc3603#3 & 6.90 & 1.79e-7 & 6.9% + ngc3603#4 & 26.20 & 4.05e-7 & 18.8% + ngc3603#5 & 4.37 & 1.34e-7 & 3.9% + ngc3603#6 & 2.96 & 7.69e-8 & + ngc3603#7 & & & + ngc3603#8 & 4.47 & 1.91e-8 & + ngc3603#9 & & & + ngc3603 ( average ) & & 1.63e-7 & 9.9% + 30dor#2 & 0.11 & 4.24e-8 & 2.4% + 30dor#3 & 1.56 & 1.44e-7 & 9.5% + 30dor#4 & & & + 30dor#5 & 0.12 & 2.99e-8 & 1.3% + 30dor#6 & 0.33 & 6.08e-8 & 3.7% + 30dor#7 & 0.91 & 3.54e-8 & 2.0% + 30dor#8 & 2.45 & 1.30e-7 & 6.8% + 30dor#10 & 2.34 & 1.54e-7 & 9.0% + 30dor#11 & 0.10 & 7.99e-9 & 0.3% + 30dor#12 & & & + 30dor#13 & & & + 30dor#14 & 1.54 & 4.59e-8 & 2.8% + 30dor#15 & 0.66 & 1.94e-8 & 0.8% + 30dor#16 & 0.30 & 9.93e-9 & 0.6% + 30dor#17 & 0.24 & 7.10e-8 & 2.7% + 30doradus ( average ) & & 6.26e-8 & 3.5% + n66#1 & & & + n66#2 & & & + n66#3 & & & + n66#5 & 0.06 & 3.23e-8 & 2.7% + n66#6 & 0.10 & 7.66e-8 & 3.7% + n66#7 & & & + n66#8 & 0.03 & 1.71e-8 & 2.0% + n66#9 & 0.12 & 1.05e-7 & 9.9% + n66#10 & 0.09 & 6.95e-8 & 7.3% + n66#11 & & & + n66#12 & 0.03 & 1.61e-8 & 1.8% + n66#13 & 0.13 & 8.21e-8 & 9.3% + n66 ( average ) & & 6.55e-8 & 5.2% + given their ip ( table [ tab : ip ] ) , we expect _ a priori _ ar
ii , ar iii , and ar iv to be present in the ionized gas . through the sl module of the irs , we have access to the [ ar ii ] line at 6.99@xmath3 m
. we were able to estimate the ar ii ionic abundance using the method described in
[ sec : ionicab_method ] , with an estimated uncertainty as large as 25% , principally due to the low spectral resolution of the sl module together with the faintness of the line .
the results show that the ar ii / ar iii abundance ratio is smaller than 10% except toward the position # 4 in ngc3603 ( table [ tab : ionicab_arii ] ) .
the line of sight toward ngc3603#4 is characterized by a relatively low ionization degree , based on the ( ne iii / h)/(s
iii / h ) ionic abundance ratio ( [ sec : iondegree ] ) , which explains the large amount of ar ii .
it is not entirely clear , however , whether ar ii belongs to the ionized gaseous phase .
in fact , even though ar i has an ip of 15.76ev , i.e. , above that of h i , sofia & jenkins ( 1998 ) proposed that the pics of ar i is such that it can be ionized into ar ii when hydrogen is still in h i. the calculations of chemical abundances in the neutral gas of star - forming regions seem to agree with this proposal ( see e.g. , lecavelier et al . 2004 ; lebouteiller et al .
2005 , 2006 ) .
in addition , the study of the spatial variations of mir features across ngc3603 shows that the [ ar ii ] line intensity correlates with that of the pdr tracers such as the pahs , and not with usual ionized gas tracers such as [ ne ii ] , [ ne iii ] , [ s iii ] , or [ s iv ] ( lebouteiller et al . 2007 ) . hence , the contribution of ar ii we derived should be regarded as an upper limit .
+ + the ip of ar iii ( 40.74ev ) is similar to that of ne ii ( 40.96ev ) , thus it could be expected that , together with ar ii and ar iii , ar iv is present in the ionized gas , given the ubiquitous detection of [ ne iii ] in our spectra .
we consider however that ar iv contribution is not significant given the relatively low pics of ar iii as compared to that of ar ii for energies smaller than @xmath1270ev ( verner et al .
optical observations of ngc3603 show that ar iv contribution is indeed negligible , with ar iv@xmath62ar iii ( garca - rojas et al .
a similar value is found for 30dor , where the contribution of ar iv is @xmath122 - 3% that of ar iii ( peimbert 2003 ; tsamis et al .
2003 ) . in n66 on the other hand ,
ar iv could represent as much as 17% as compared to ar iii ( peimbert et al .
such a large contribution is not surprising given the relatively hard isrf in this object . to investigate in more detail the presence of ar iv
, we used the photoionization grids of stasiska ( 1990 ) . figure [ fig : ioncorr]a shows the correlation between the ionization fraction ratios @xmath63(ar iv)/@xmath63(ar iii ) and @xmath63(ne iii)/@xmath63(ne ii ) . in the spectra of the three giant h ii regions of our sample ,
@xmath63(ne iii)/@xmath63(ne ii ) ranges from @xmath40.28 to @xmath46.31 ( dashed lines in fig .
[ fig : ioncorr]a ) .
this corresponds to @xmath63(ar iv)/@xmath63(ar iii ) ranging from @xmath40.15 to @xmath40.64 . hence according to the models , ar
iii is the dominant stage , and corrections factors on the final argon abundance are on the order of 15 - 65% ( 0.06 - 0.21dex ) .
these ionization correction factors seem to be significantly larger than found in the optical ( ar iv / ar iii@xmath2817% , i.e. , @xmath280.07 dex ) . the optical spectra probe a gas with an ionization degree equal or higher
as compared to the gas probed in the mir ( [ sec : iondegree ] ) , hence the icf due to the presence of ar iv is likely smaller than @xmath1220% in the mir spectra .
the final results suggest that corrections are in fact negligible ( [ sec : argondisc ] ) .
the final error on ar / h is due to measurement errors on the ionic abundances of ar ii ( @xmath1225% ) and ar iii ( @xmath1220% ) , and to the assumed physical conditions ( @xmath6020% ) . given the ip of the iron ionization stages ( see table [ tab : ip ] ) , it is expected that most of the iron in the ionized gas is into fe ii , fe iii , and fe iv .
fe iv is not expected to be further ionized because of the helium absorption edges in stars at 54.4ev .
we have access to both fe ii ( 25.99@xmath3 m ) and fe iii ( 22.92@xmath3 m ) in the mir .
the fe ii ion does not necessarily arise in the ionized gas .
in fact , we find that the [ fe ii ] line flux correlates best with pdr tracers such as the pah emission or [ ar ii ] ( lebouteiller et al . in preparation ) .
on the other hand , the presence of fe iv could seriously hamper the iron abundance determination .
one way to probe the presence of fe iv is to study the variation of the ionic abundance ratio fe iii / fe ii as a function of the ionization degree ( see the application in m42 and m17 by rodrguez 2002 ) .
fe iii / fe ii is expected to decrease when the ionization becomes harder , due to the presence of fe iv . in our sample
, there is no clear correlation with the ionization degree as probed by [ ne iii]/[ne ii ] .
this means either that fe iv is not present in significant amount or that on the contrary it is the dominant stage over the whole range of physical conditions ( density , isrf hardness ) in the h ii regions .
we estimated the ionic abundance of fe iv using the photoionization grids of stasiska ( 1990 ) .
it can be seen in fig .
[ fig : ioncorr]b that @xmath63(fe iv)/@xmath63(fe iii ) lies between @xmath40.56 and @xmath46.21 . hence fe iv represents a significant fraction of the total iron in the ionized gas , even becoming the dominant ionization stages in regions with [ ne iii]/[ne ii ] greater than @xmath40.6 , i.e. , for most of our observations . in their optical study of ngc3603 , garca - rojas et al .
( 2006 ) assumed that fe iv@xmath64(fe ii+fe iii ) , confirming the predominancy of the fe iv ion .
the correction we applied due to the presence of fe iv ranges from a factor 0.6 to 6 ( table [ tab : eleab ] ) . the error on fe / h is due to measurement error on fe ii / h ( 15% ) and fe iii / h ( 20% ) , and to the assumed physical conditions ( @xmath6020% ) .
we corrected fe / h for the presence of fe iv .
the error on the icf is estimated to be as much as a factor 2 .
lllll & ne / h & s / h & ar / h & fe / h ( icf ) + ngc3603#3 & 7.95 & 7.08 & 6.41 & 5.98 ( 0.20 ) + ngc3603#4 & 7.94 & 6.92 & 6.33 & 6.13 ( 0.13 ) + ngc3603#5 & 7.94 & 6.96 & 6.53 & 6.14 ( 0.26 ) + ngc3603#6 & & & & + ngc3603#7 & & & & + ngc3603#8 & & & & + ngc3603#9 & & & & + error & @xmath65 & @xmath65 & @xmath66 & @xmath67 + ngc3603 ( average ) & 7.94 & 6.99 & 6.43 & 6.09 ( 0.19 ) + ( standard deviation ) & 0.00 & 0.09 & 0.11 & 0.09 + 30dor#2 & 7.74 & 6.76 & 6.25 & 6.01 ( 0.46 ) + 30dor#3 & 7.72 & 6.76 & 6.22 & 6.04 ( 0.63 ) + 30dor#4 & 7.73 & 6.77 & 6.20 & 6.32 ( 0.86 ) + 30dor#5 & 7.79 & 6.82 & 6.36 & 6.16 ( 0.36 ) + 30dor#6 & 7.82 & 6.86 & 6.23 & 6.42 ( 0.40 ) + 30dor#7 & 7.73 & 6.73 & 6.26 & 5.80 ( 0.21 ) + 30dor#8 & 7.72 & 6.70 & 6.31 & 6.10 ( 0.25 ) + 30dor#10 & 7.75 & 6.72 & 6.28 & 5.98 ( 0.16 ) + 30dor#11 & 7.80 & 6.83 & 6.36 & 6.26 ( 0.48 ) + 30dor#12 & 7.79 & 6.82 & 6.38 & 6.32 ( 0.54 ) + 30dor#13 & 7.71 & 6.76 & 6.32 & 6.20 ( 0.63 ) + 30dor#14 & 7.77 & 6.77 & 6.22 & 6.24 ( 0.52 ) + 30dor#15 & 7.82 & 6.84 & 6.37 & 6.33 ( 0.59 ) + 30dor#16 & 7.72 & 6.76 & 6.23 & 6.22 ( 0.83 ) + 30dor#17 & 7.71 & 6.63 & 6.44 & 6.57 ( 0.16 ) + error & @xmath65 & @xmath65 & @xmath66 & @xmath67 + 30dor ( average ) & 7.76 & 6.77 & 6.32 & 6.24 ( 0.35 ) + ( standard deviation ) & 0.02 & 0.03 & 0.09 & 0.11 + n66#1 & 7.39 & 6.50 & 5.85 & 5.30 ( 0.80 ) + n66#2 & 7.37 & 6.46 & 5.73 & + n66#3 & & & & + n66#5 & 7.25 & 6.32 & 6.07 & + n66#6 & 7.39 & 6.44 & 6.31 & + n66#7 & 7.27 & 6.45 & 6.10 & + n66#8 & 7.38 & 6.42 & 5.93 & 5.61 ( 0.35 ) + n66#9 & 7.39 & 6.41 & 6.03 & + n66#10 & 7.34 & 6.36 & 5.98 & + n66#11 & 7.41 & 6.44 & 6.05 & + n66#12 & 7.21 & 6.26 & 5.95 & + n66#13 & 7.25 & 6.33 & 5.94 & + error & @xmath65 & @xmath65 & @xmath66 & @xmath67 + n66 ( average ) & 7.34 & 6.36 & 5.97 & 5.48 ( 0.45 ) + ( standard deviation ) & 0.05 & 0.10 & 0.15 & 0.24 + lllllll ne / h & s / h & ar / h & fe / h & reference + + @xmath68 & @xmath69 & @xmath70 & @xmath71 & ( 1 ) + @xmath72 & @xmath73 & @xmath74 & & ( 2 ) + @xmath75 & @xmath76 & & & ( 3 ) + & & + + & @xmath77 & & & ( 3 ) + @xmath78 & @xmath79 & @xmath80 & @xmath81 & ( 4 ) + @xmath82 & @xmath83 & @xmath84 & @xmath85 & ( 5 ) + @xmath86 & @xmath83 & @xmath87 & & ( 6 , pos#1 ) + @xmath88 & @xmath89 & @xmath90 & & ( 6 , pos#2 ) + @xmath91 & @xmath92 & @xmath93 & & ( 6 , pos#3 ) + @xmath94 & @xmath95 & @xmath96 & & ( 6 , pos#4 ) + @xmath97 & @xmath98 & @xmath99 & & ( 7 ) + @xmath100 & @xmath101 & @xmath102 & & ( 8) + & & + + @xmath103 & @xmath104 & 5.81 & & ( 6 ) + @xmath105 & & & & ( 9 , n66a ) + @xmath106 & & & & ( 9 , n66nw ) + @xmath107 & & & & ( 10 ) + @xmath107 & @xmath108 & @xmath109 & & ( 12 , t@xmath110=0.0013 ) + & & + + + elemental abundances are presented in table [ tab : eleab ] and illustrated in fig .
[ fig : abundances ] .
the uncertainties are comparable to the abundance dispersion across each giant h ii region ( table [ tab : eleab ] ) , suggesting that the abundance variations we observe are mostly driven by uncertainties and are probably not intrinsic ( see also discussion in [ sec : dispersion ] ) .
elemental abundances from optical studies are summarized in table [ tab : eleab_opt ] .
the mir abundances of neon , sulfur , argon , and iron scale according to their abundance in the sun .
for the three objects in our sample , the elemental abundances derived from mir collisionally - excited lines ( cels ; red diamonds in fig .
[ fig : abundances ] ) agree globally with those derived from optical cels ( black dots ) .
because our abundance measurements are less affected by electronic temperature determinations , this suggests that the discrepancy often observed between optical recombination lines and cels is not principally due to temperature fluctuations ( see also liu et al .
2001 ; bernard - salas et al .
. the abundance of neon does not suffer from any significant uncertainty due to ionization corrections ( [ sec : neondet ] ) .
neon stands out as being the element showing the best agreement with the corresponding optical references ( [ tab : eleab_opt ] ) .
the average abundance we derive in ngc3603 ( @xmath111 ) is relatively close to the value ne / h=8.08 found by simpson et al .
( 1995 ) using the same mir cels as we observe , but combining different observations with possible aperture effects .
neon is also the element which shows the smallest scatter in its abundance across the various positions in each giant h ii region . as an illustration ,
the 15 positions in 30doradus show only 0.11dex dispersion , the 3 positions in ngc3603 give remarkably equal values , while the 11 positions in n66 show 0.20dex dispersion .
these findings , together with the fact that neon is not expected to be depleted on dust grains or incorporated in molecules makes it a reliable metallicity tracer , contrary to sulfur , argon , and iron ( see also [ sec : cosmic ] ) . for this reason ,
we consider it as a reference for our discussion .
the average abundances of sulfur we determined in the giant h ii regions corroborate the optical values ( fig . [
fig : abundances ] ) .
there might be an indication however that our determinations agree best with the lowest optically - derived abundances ( see discussion in [ sec : diffdep ] ) .
the sulfur abundance shows a larger dispersion across each giant h ii region than neon .
part of the larger dispersion could be attributed _ a priori _ to ionization corrections due to the presence of s ii .
however , an icf would change s / h by less than 0.05dex ( assuming s ii / s iii=0.1 ; [ sec : sulfurdet ] ) , which is smaller than the total uncertainty on the sulfur abundance determination ( @xmath65dex ) . in order to study in more detail the sulfur abundance in the h ii regions , we compare it to the neon abundance ( see [ sec : neondisc ] ) . as discussed in thuan et al .
( 1995 ) , ne , s , and ar are products of @xmath0-processes during both hydrostatic and explosive nucleosynthesis in massive stars .
these elements are thus thought to follow a parallel chemical evolution in stars except in the case of an extreme initial mass function ( imf ) because the stars synthetizing ne have somewhat larger masses than the stars producing s and ar ( woosley & weaver 1995 ) .
we observe that ne / h and s / h trace each other remarkably well in the three objects of our sample ( fig .
[ fig : abs ] ) .
this contrasts with the lack of correlation found by verma et al .
( 2003 ) in starburst regions using iso data , but this is consistent with ne and s production in massive stars .
+ + the ne / s ratio appears to be significantly larger than the solar value by @xmath120.2 - 0.3dex ( fig .
[ fig : abs ] ) . correcting
for the presence of s ii can not explain this discrepancy ( [ sec : sulfurdet ] ) .
note that an extreme imf is also unlikely to explain the sulfur depletion because s / h and ar / h show somewhat different behavior with respect to ne / h although s and ar are produced in similar mass stars ( [ sec : argondisc ] ) .
interestingly , the ne / s ratio was found to be on average larger in the pne of the magellanic clouds than in the milky way , which was attributed to sulfur depletion on molecules and dust grains ( marigo et al .
2003 ; henry et al .
2004 ; pottasch & bernard - salas 2006 ; bernard - salas et al .
this underabundance of sulfur with respect to other @xmath0-elements has also been observed in h ii regions in the milky way ( simpson et al .
2004 ) and in m83 and m33 ( rubin et al . 2007 ) with the same interpretation . unlike si or fe for instance ,
sulfur is not depleted toward cool diffuse clouds ( savage & sembach 1996 ) . however , in dense regions and in pdrs , it becomes significantly depleted ( simpson & rubin 1990 ; verma et al .
2003 ; pottasch & bernard - salas 2006 ) .
goicoechea et al .
( 2006 ) found that sulfur depletion could be up to a factor 4 , i.e. , 0.6dex in the horsehead pdr .
for even denser regions , it has been established that sulfur is about two orders of magnitude more depleted than c , n , and o in molecular clumps with densities @xmath112(h)@xmath113@xmath50 ( ruffle et al .
1999 ) .
the position disagreeing the most with the ne / s solar value is 30dor#8 , which shows prominent silicate absorption , where the ism is denser and where refractory elements are more likely to be depleted onto dust grains .
however , the ionized gas we measure the abundances from is not associated with the dense region , as indicated by the negligible amount of mir extinction toward 30dor#8 ( lebouteiller et al . in preparation ) .
furthermore , regions showing ratios closer to the solar ne / s ( e.g. , n66#7 , ngc3603#3 ) show pah emission , which is a sign of relatively dense regions .
hence , though sulfur depletion onto dust grains is likely to occur , there is no trend with the spectral charateristics ( pah - dominated , ionized gas , etc ... ) .
this is due to the fact that the ionized gas we probe is not entirely associated with the pdrs or the embedded regions , because of the complex geometry and structure of the lines of sight .
results in blue compact dwarfs ( izotov & thuan 1999 ; wu et al .
2007 ) , and galactic h ii regions ( simpson et al .
2004 ) are added for comparison in figure [ fig : abs ] .
bcds have optical spectra very similar to galactic h ii regions , and they sample the low - metallicity zone in the plot .
the giant h ii region n66 in the smc overlaps the most metal - rich bcds .
note that abundance measurements from the optical and mir studies of bcds disagree with each other ( crosses and filled circles in figure [ fig : abs ] ) .
however in the case of the bcds , the difference comes from a higher neon abundance measured in the mir , which could be due to the fact that the gas probed in the mir is enriched in heavy elements and that sulfur is depleted on dust grains .
it seems that the trend formed by the bcds and the h ii regions of our sample is significantly off the solar proportion for metallicities close to solar , while it barely follows the solar proportion at metalllicities lower or equal to that of n66 .
globally , there might be a hint that the ne / s ratio increases together with the metallicity , which would be an important proof of sulfur depletion on dust grains . in ngc3603 ,
ar / h it is somewhat smaller than the optical determination ( fig .
[ fig : abundances ] ) .
the argon abundances in n66 and 30dor are surprisingly large even when no icf accounting for ar iv is applied .
the argon abundances determined using only the ar ii and ar iii contributions match better the solar and bcd values than with icf correction due to the possible presence of ar iv ( fig .
[ fig : abar ] ) . as an illustration ,
when using an an icf to account for ar iv , we find ar / h in 30dor equal to the highest solar determinations ( while 30dor has clearly a subsolar metallicity with 0.60z@xmath114 ; [ sec : cosmic ] ) . these results are consistent with the fact that the main ionization stage in the dense gas probed in the mir is ar iii , with a small contribution due to ar ii ( [ sec : argondet ] ) .
+ + argon is produced by oxygen - burning in massive stars , and it is expected to follow both sulfur and neon nucleosynthesis . indeed the argon abundance ar / h correlates with ne / h in the giant h ii regions ( fig .
[ fig : abar ] ) , but with a somewhat larger dispersion than the ne / h vs. s / h correlation .
this is mostly due to large measurement uncertainties on the [ ar ii ] and [ ar iii ] line fluxes ( [ sec : measurements ] ) , and not to the contribution from ar iv ( [ sec : argondet ] ) .
there is no position in any of the h ii regions showing an argon deficiency with respect to neon , even at high metallicities , which is consistent with argon being undepleted on dust grains or in molecules , even in the densest regions probed .
the mir neon and argon abundance determinations in ngc3603 and 30dor agree well with the solar value ( fig .
[ fig : abar ] ) .
in contrast , about half of the positions in n66 shows a larger argon abundance than expected from the solar ne / ar .
verma et al .
( 2003 ) find that argon is overabundant relative to neon in a wide variety of star - forming galaxies , with metallicities from sub - solar ( bcds ) to super - solar ( wr galaxies ) .
there is no trend among the galaxies in their sample .
we could see the effect of enhanced argon abundance in n66 but it is more likely due to uncertaities in the abundance determination .
the average fe / h determinations in ngc3603 and 30dor agree with each other ( table [ tab : eleab ] ) , and barely match fe / h in the most metal - rich bcds ( fig .
[ fig : abfe ] ) .
the average iron abundance in ngc3603 ( @xmath115 ) compares well with the value derived from the optical ( @xmath71 ; garca - rojas et al .
2006 ) , and implies no differential depletion between the gases probed in the mir and in the optical ( [ sec : diffdep ] ) .
the iron abundance in 30dor shows a large dispersion including the optical value within the range . the fe / h value toward 30dor#17 is more than 3@xmath116 larger than the average iron abundance in 30dor .
the spectrum of 30dor#17 shows deep silicate absorption probably originating from dust associated with an ultracompact h ii region . in the same spectrum
, we also observe the high - excitation [ o iv ] line at 25.89@xmath3 m , which is known to originate around hot wolf - rayet stars ( schaerer & stasinska 1999 ) and in shock - heated gas ( lutz et al .
1998 ) . because the region is dominated by schocks from the nearby supernova remnant ( snr ) 30dorb ( chu et al . 1992 ) , it is likely that the [ o iv ] line traces gas shocked by the sn .
hence the enhanced iron abundance toward 30dor#17 is consistent with removal of iron atoms from dust grains due to shocks .
we plan to investigate in more detail the ism properties toward the snr using the new data provided by an accepted irs cycle 5 gto program .
+ there are only two fe / h determinations in n66 ( positions # 1 and # 8) , both of which are significantly smaller than the fe / h values in the two other giant h ii regions , and also smaller than the bcds having similar metallicities ( as traced by ne / h )
. this might be due to underestimated ionization corrections due to the presence of fe iv in this object ( [ sec : irondet ] ) . however , although n66 is characterized by a globally harder isrf as compared to the two other h ii regions in our sample , the positions n66#1 and n66#8 do not show significantly harder isrfs than all the positions in 30dor . hence the iron underabundance in n66 could be genuine .
the fe / ne ratio in the three giant h ii regions is significantly lower than the solar value ( fig .
[ fig : abfe ] ) . this can be explained by iron depletion on the surface of very small grains , which are the dominant dust component in the ionized gas ( see e.g. , cesarsky et al .
1996 ; verstaete et al . 1996
; lebouteiller et al . 2007 ) .
rodrguez ( 2002 ) observed that the [ o / fe ] ratio increases with metallicity in bcds , and explained it by iron depletion on dust at high metallicity .
we find that the trend formed by the bcds and the giant h ii regions flattens for large neon abundances , suggesting that the iron becomes more and more underabundant as the metallicity increases , being consistent with iron depletion on dust grains .
mir spectra allow us to probe regions that are obscured at optical wavelengths , possibly with different physical and chemical properties .
the average sulfur abundances we calculated from irs spectral lines in the giant h ii regions corroborate the optical determinations ( fig . [ fig : abundances ] ) , even though they likely fall towards their lower side .
interestingly , simpson et al . ( 1995 ) measured the mir [ s iii ] line flux , and used the [ s iv ] line flux in the literature to infer s / h=7.12 in ngc3603 , i.e. , close to our average determination for this object ( @xmath117 ; table [ tab : eleab ] ) .
these two mir determinations are significantly lower than [email protected] , which was obtained from optical lines by garca - rojas et al .
if the relative sulfur underabundance in the mir is real , this could hint at a differential sulfur depletion between the gases probed in the mir and in the optical .
however , it is likely that the underabundance is driven by uncertainties in the abundance determination methods .
ionization correction in the mir could represent 0.04 - 0.05dex ( [ sec : sulfurdet ] ) .
on the other hand , optical studies do not directly observe s iv , which in our case contribute around 5% to the total sulfur in ngc3603 , and as much as 24% in n66 and 17% in 30dor .
the icf applied in the optical could overcorrect the total sulfur abundance .
morevover , electronic temperature variations / uncertainties affect the optical determinations and to a less extent the mir determinations ( [ sec : ionicab_results ] ) .
the study of the refractory element iron supports the lack of differential depletion between the gases probed in the mir and in the optical .
we found that the iron abundance in ngc3603 and 30dor is similar when derived in the optical and in the mir ( [ sec : irondisc ] ) .
this confirms that the possible discrepancy seen in s / h between mir and optical observations is not due to additional depletion in the gas observed in the mir , but rather to uncertainties on the abundance determination .
this stresses the importance of deriving the abundance of a refractory element such as iron in dense regions .
we found that the chemical composition derived in the optical and in the mir is overall fairly similar ( fig .
[ fig : abundances ] ) . for this reason ,
it is possible to compare the ionic abundances from optical and mir wavelengths and interpret possible differences in physical conditions . in order to trace the gas excitation
, we use the ionic abundance ratio ( ne iii / h)/(s iii / h ) .
this ratio is a good approximation for ( ne iii / h)/(ne ii / h ) in h ii regions ( lebouteiller et al . in preparation ) and it can be derived in both optical and mir ranges .
based on the values in table [ tab : ionicab_global ] , the ( ne iii / h)/(s
iii / h ) ratio is @xmath118 in ngc3603 , i.e. , about two times smaller than the value in the optical ( @xmath119 ; garcia - rojas et al .
we find ( ne iii / h)/(s
iii / h ) = @xmath120 in 30dor , somewhat smaller than @xmath121 in the optical ( peimbert 2003 ) .
finally , we find ( ne iii / h)/(s
iii / h ) = @xmath122 in n66 , as compared to 10.3 in the optical ( vermeij & van der hulst 2002 ) .
hence the mir spectra seem to probe ionized gas with a degree of ionization equal to or lower than that of the gas probed in the optical
. this could be due to the fact that the sh and lh apertures from the irs do not sample the same spatial regions as those observed within narrow slits used in the optical .
it is also possible that the gases probed in the mir and in the optical have different physical conditions ( such as density ) .
stellar winds and supernov explosions from the current star - formation episode release newly produced elements into the surrounding ism on a short timescale ( few @xmath123yr ) .
it is thus natural to question the origin of the metals we observe in the ionized gas of the giant h ii regions .
we investigate in this section the total metal content of each giant h ii region by comparing their metallicities with the sun and other objects from the same host galaxies ( h ii regions , planetary nebul ) .
first it must be stressed that the solar abundance determinations have shown strong variations over the past few years ( see e.g. , pottasch & bernard - salas 2006 ) .
the gray stripes in fig .
[ fig : abundances ] illustrate the range of solar abundance determinations for each element over the 1998 - 2007 period .
neon and argon abundances are particularly poorly determined because of the absence of suitable lines in the solar photosphere .
abundance determinations of those elements are indirect and make use of coronal lines together with a correction using other @xmath0-elements such as magnesium or oxygen . considering the most extreme determinations among the recent studies ( asplund et al .
2006 ; feldman & widing 2003 ) , solar ne / h determinations range from @xmath124 to @xmath125 while ar / h ranges from @xmath126 to @xmath127 .
the sulfur solar abundance is better determined , with values from 7.14 ( asplund et al .
2006 ) to 7.33 ( grevesse et al . 1998 ) .
we consider neon as being the most reliable metallicity tracer available in our study , as compared to sulfur , argon , and iron ( [ sec : neondisc ] ) .
the average neon abundance is @xmath128 in ngc3603 , @xmath129 in 30dor , and @xmath130 in n66 ( table [ tab : eleab ] ) . considering an average solar value of 7.98
, the average neon abundance in ngc3603 implies essentially a solar metallicity ( @xmath40.91 z@xmath114 ) , while 30dor is @xmath40.60 z@xmath114 , and n66 is @xmath40.23 z@xmath114 .
the metallicity of ngc3603 agrees with the solar metallicity within uncertainties .
this is consistent with the abundance gradient in the milky way ( see , e.g. , martin - hernandez et al .
2002 ) given the fact that the galactocentric distance of ngc3603 is almost that of the sun ( 8.5kpc as compared to 8kpc ; [ sec : pres_3603 ] ) .
it is instructive to compare our abundances in 30dor and n66 with those in planetary nebul ( pne ) and in other h ii regions in the lmc and smc from bernard - salas et al .
the authors measured abundances using mir lines in a similar way to the present study so that the comparison does not suffer from significant systematic uncertainties .
bernard - salas et al . ( 2007 ) found that neon and sulfur abundances in pne agree relatively well with those in the h ii regions of the lmc and smc , implying that h ii regions have not been significantly enriched over the past few gyrs ( age of the pne progenitor ) .
the global neon abundance we inferred from the mir lines in 30dor is @xmath129 , remarkably close to the average value in the pne of the lmc @xmath28ne / h@xmath131=7.78 ( no quoted errors ) .
the global neon abundance we derive in n66 ( @xmath132 ) is consistent within error bars with the average @xmath28ne / h@xmath131 in the pne of the smc ( 7.43 ) .
in addition , it must be noticed that the metallicity in o dwarf stars in ngc346 , the stellar cluster associated with n66 , is 0.2 z@xmath114 ( bouret et al .
2003 ) , which agrees well with our metallicity determination in the ism of the h ii region ( @xmath40.23 z@xmath114 ) .
it seems that the chemical composition of giant h ii regions compares well with that of pne and young stars , suggesting that the ionized gas of the h ii regions has not been enriched significantly by the current star - formation episode , or by any previous episode more recent than @xmath601gyr .
if enrichment has really occured , the metallicity of the h ii regions has been enhanced by much less than a factor 2 .
full mixing requires that elements acquire the same physical conditions as the ionized gas , namely temperature and density , but also viscosity and molecular diffusion ( scalo & elmegreen 2004 ) . the effective mixing provided by turbulent diffusion in h ii regions allows metals to mix on spatial scales of a few hundreds parsecs over 100myr ( roy & kunth 1995 ; avillez & mac low 2002 ) .
hence abundance discontinuities between h ii regions in a single galaxy should be ubiquitous , especially in low - mass galaxies where rotational shear is weak ( roy & kunth 1995 ) .
however observations challenge this hypothesis and suggest that mixing can occur on spatial - scales as large as the galaxy size and over time - scales larger than the age of the h ii regions ( see e.g. , kobulnicky 1998 ; skillman & kennicut 1993 ; noeske et al .
2000 ; russel & dopita 1990 ) .
our results support such spatial- and time - scales and question the fate of metals released by massive stars in giant h ii regions . following the scenario of tenorio - tagle ( 1996 )
, the metals could undergo a long cycle in a hot phase before mixing with the surrounding ism .
the composite _ irac _ and _ chandra _ image of 30dor shows large cavities filled with hot gas , surrounded by colder and denser ionized gas shells and filaments ( brandl et al . in preparation ) .
these are supernova - driven outflows generated by the superstellar cluster ( see also e.g. , redman et al .
the temperature of metals in the hot cavities is too high for recombination to occur , and the density too small for collisionally - excited levels to be populated so that the abundances we derive from the mir spectra do not account for these newly released elements .
the superbubbles of hot gas in 30dor should remain inside the galaxy for hundreds of myr according to the scenario of recchi et al .
it could eventually lead to a superwind that will enable the escape of metals in the halo ( tenorio - tagle 1996 ) .
our observations probe the chemical abundances toward several positions within each giant h ii region .
neon and sulfur abundances show remarkably little dispersion in the three giant h ii regions ( fig .
[ fig : abundances ] ) .
vermeij & van der hulst ( 2002 ) analyzed 4 positions in 30dor and found neon abundances ranging between 7.84 and 7.94 . analyzing 15 positions
, we find a comparable dispersion , 0.11dex from the mininum to the maximum ne / h values . in the same object , vermeij & van der hulst ( 2002 )
find sulfur abundances ranging from 6.63 to 6.86 , and argon abundances ranging from 6.18 to 6.63 , which is a smaller dispersion than we measured ( but with fewer positions ) .
our results in ngc3603 and n66 also imply that the neon abundance and , to a lesser extent , the sulfur abundance show little dispersion in these regions .
part , if not all , of the dispersion of the elemental abundances is due to uncertainties in the abundance determinations , such as ionization corrections ( [ sec : eleab ] ) and the assumed nebula physical conditions ( [ sec : ionicab_results ] ) . in this section
, we investigate whether there is evidence for small - scale mixing .
the apparent homogeneity of the abundances contrasts with the wide variety of physical regions each line of sight samples toward each object ( pdr , stellar cluster , ionized gas , embedded regions ) .
this suggests that the ionized gas from which we measure the chemical abundances is not necessarily associated with these physical regions .
in fact , the lack of mir line extinction toward pdrs and toward embedded objects implies that the whole nebula is filled with ionized foreground gas ( lebouteiller et al . in preparation ) . can abundance inhomogeneities still be inferred ?
the material ejected from supernov could reach a hot coronal phase before falling back onto the galactic disk in the form of molecular droplets ( tenorio - tagle 1996 ) .
this material will then be photodissociated by massive stars and mix with the surrounding ionized gas . according to this model ,
mixing could occur at very small spatial scales ( @xmath601pc ) .
tsamis & pequignot ( 2005 ) introduced in their model of 30dor small - scale chemical inhomogeneities at the subparsec - sized scale .
the typical distance between the irs observations is @xmath124pc in ngc3603 , @xmath1220pc in 30dor , and @xmath1215pc in n66 .
hence our mir observations do not have the necessary spatial resolution to probe such chemical inhomogeneities . on the other hand , recchi et al .
( 2001 ) proposed that , mostly because of thermal conduction , the sne type ii ejecta start to cool down and mix with the cold gas within a few @xmath123yr before the formation of hot gas outflows . in this particular case ,
small - scale mixing could be observed locally . in order to assert whether small - scale mixing is responsible for the abundance variations we observe across each giant h ii region , we use ne / h in 30dor as a reference . the standard deviation of ne / h is 0.02dex , implying that the abundance of neon could vary as much as @xmath45% across the region .
interestingly , 5% of the neon abundance currently present in 30dor corresponds to the total neon enrichment that the star - forming dwarf galaxy izw18 experienced so far ( ne / h = 6.40 ; izotov et al .
kunth & sargent ( 1986 ) proposed that the metallicity of izw18 represents the minimum chemical enrichment of an h ii region from a starburst episode .
it is unlikely that the small - scale variations in a single giant h ii region correspond to the yields from a starburst episode , especially considering the fact that the metals in the ionized gas are not likely to be cotemporal with the star - formation episode ( [ sec : cosmic ] ) .
uncertainties on the measurements and on the abundance determination method ( @xmath65dex for ne / h , i.e. , as much as 55% ) are too large to constrain possible abundance fluctuations of less than 5% at the metallicity of 30doradus .
we analyzed the chemical abundances in the ism of three giant h ii regions , ngc3603 in the milky way , 30dor in the lmc , and n66 in the smc using the mir lines observed with the irs onboard spitzer .
* our observations probe the ism toward various physical regions , such as stellar clusters , ionized gas , photodissociation regions , and deeply embedded mir bright sources .
the spectra show the main ionization stages of neon , sulfur , and argon in the ionized gas .
we also detect [ fe ii ] and [ fe iii ] lines . * ionic abundances of ne ii , ne iii , s iii , s iv , ar ii
, ar iii , fe ii , and fe iii were derived .
the internal variation of electron density across a region has no impact on the ionic abundance determination . on the other hand
, we find that electron temperature uncertainties and/or intrinsic variations could be responsible for an error of 20% at most on the abundance determinations . based on the ( ne iii / h)/(s iii / h ) ionic abundance ratio
, we find that the optical spectra probe a gas with a degree of ionization equal to or higher than the gas probed in the mir .
* elemental abundances were determined from the ionic abundances .
no ionization corrections were needed , except for iron .
we find that neon , sulfur , and argon scale with each other , which is expected from stellar yields .
abundances do not show any dependence on the physical region ( pdr , stellar cluster , embedded region , ... ) . *
the ne / s ratio is larger than the solar value , and suggests that sulfur could be depleted onto dust grains .
the sulfur abundance in the mir agrees best with the lowest optical determinations , which is likely due to uncertainties in the abundance determinations . *
iron abundance shows a larger uncertainty than ne / h , s / h , and ar / h . the comparison of iron and neon abundances hints at significant depletion of iron onto dust grains at large metallicities .
the agreement with the optical determination of fe / h indicates however that there is no differential depletion on dust grains between the gas probed in the mir and in the optical .
* fe / h is found to be spectacularly large in one position , corresponding to a supernova remnant .
this strongly suggest that iron atoms have been released from dust grains due to schocks from the sn . *
the metallicity of ngc3603 agrees with the galactic abundance gradient .
the metallicities of 30doradus and n66 agree well with those of the pne in their respective host galaxies .
these findings suggest that the giant h ii regions did not experience a significant metal enrichment for at least 1gyr .
if enrichement occured , the metallicity was altered by less than a factor of two .
* neon and sulfur abundances show remarkably little dispersion in the three h ii regions ( e.g. , 0.11dex dispersion in 15 positions in 30dor ) .
small - scale mixing is apparently effective , abundance fluctuations are smaller than @xmath1255% . however , internal variations of the abundances are likely to be on the order of @xmath605% , and determining their existence would require a significant improvement of the data quality and of the method to be evidenced .
we would like to thank daniel kunth , suzanne madden , and marc sauvage for their useful comments .
we thank the referee for providing constructive comments .
this work is based on observations made with the _
spitzer space telescope _ , which is operated by the jet propulsion laboratory , california institute of technology , under nasa contract 1047 .
support for this work was provided by nasa through contract 1257184 issued by jpl / caltech .
vc would like to acknowledge the partial support from the eu tok grant 39965 .
_ bcd _ : blue compact dwarf .
_ cel _ : collisionally - excited line .
_ gto _ : guaranted time observation .
_ icf _ : ionization correction factor .
_ imf _ : initial mass function .
_ ip _ : ionization potential .
_ ism _ : interstellar medium .
_ isrf _ : interstellar radiation field .
_ lmc _ : large magellanic cloud .
_ pah _ : polycyclic aromatic hydrocarbon .
_ pdr _ : photodissociation region .
_ pics _ : photoionization cross - section .
_ smc _ : small magellanic cloud .
_ snr _ : supernova remnant .
asplund , m. , grevesse , n. , & sauval , a. j. 2006 , communications in asteroseismology , 147 , 76 | we investigate the chemical abundances of ngc3603 in the milky way , of 30doradus in the large magellanic cloud , and of n66 in the small magellanic cloud .
mid - infrared observations with the infrared spectrograph onboard the spitzer space telescope allow us to probe the properties of distinct physical regions within each object : the central ionizing cluster , the surrounding ionized gas , photodissociation regions , and buried stellar clusters .
we detect [ s iii ] , [ s iv ] , [ ar iii ] , [ ne ii ] , [ ne iii ] , [ fe ii ] , and [ fe iii ] lines and derive the ionic abundances . based on the ionic abundance ratio ( ne iii / h)/(s iii / h ) , we find that the gas observed in the mir is characterized by a higher degree of ionization than the gas observed in the optical spectra .
we compute the elemental abundances of ne , s , ar , and fe .
we find that the @xmath0-elements ne , s , and ar scale with each other .
our determinations agree well with the abundances derived from the optical . the ne / s ratio is higher than the solar value in the three giant h ii regions and points toward a moderate depletion of sulfur on dust grains .
we find that the neon and sulfur abundances display a remarkably small dispersion ( 0.11dex in 15 positions in 30doradus ) , suggesting a relatively homogeneous ism , even though small - scale mixing can not be ruled out . |
there is a large body of evidence for the existence of dark matter , but its basic properties especially its mass and scattering cross section with nucleons remain unknown . assuming dark matter is a thermal relic of the early universe , weakly interacting massive particles are prime candidates , suggested by constraints on the dark matter mass and self - annihilation cross section from the present average mass density @xcite .
however , as this remains unproven , it is important to systematically test the properties of dark matter particles using only late - universe constraints . in 1990
, starkman , gould , esmailzadeh , and dimopoulos @xcite examined the possibility of strongly interacting dark matter , noting that it indeed had not been ruled out .
many authors since have explored further constraints and candidates . in this literature , `` strongly interacting '' denotes cross sections significantly larger than those of the weak interactions ; it does not necessarily mean via the usual strong interactions between hadrons .
we generally consider the constraints in the plane of dark matter mass @xmath2 and spin - independent scattering cross section with nucleons @xmath3 .
figure [ regions ] summarizes astrophysical , high - altitude balloon / rocket / satellite detector , and underground detector constraints in the @xmath4@xmath2 plane .
astrophysical limits such as the stability of the milky way disk constrain very large cross sections @xcite . accompanying and
comparable limits include those from cosmic rays and the cosmic microwave background @xcite .
small cross sections are probed by cdms and other underground detectors @xcite . a dark matter ( dm )
particle can be directly detected if @xmath3 is strong enough to cause a nuclear recoil in the detector , but only if it is weak enough to allow the excluded regions in the @xmath4@xmath2 plane , not yet including the results of this paper .
from top to bottom , these come from astrophysical constraints ( dark - shaded ) @xcite , re - analyses of high - altitude detectors ( medium - shaded ) @xcite , and underground direct dark matter detectors ( light - shaded ) @xcite .
the dark matter number density scales as 1/@xmath2 , and the scattering rates as @xmath4/@xmath2 ; for a fixed scattering rate , the required cross section then scales as @xmath2 .
we will develop a constraint from earth heating by dark matter annihilation to more definitively exclude the window between the astrophysical and underground constraints.,width=312 ] in between the astrophysical and underground limits is the window in which @xmath4 can be relatively large @xcite .
high - altitude detectors in and above the atmosphere have been used to exclude moderate - to - strong values of the cross section in this region @xcite . however , there are still large gaps not excluded .
there also is some doubt associated with these exclusions , as some of the experiments were not specifically designed to look for dm , nor were they always analyzed for this purpose by people associated with the projects .
in fact , the exclusion from the x - ray quantum calorimetry experiment was recently reanalyzed @xcite and it changed substantially from earlier estimates @xcite .
if this intermediate region can be closed , then underground detectors would set the upper limit on @xmath4 .
that would mean that these detectors are generally looking in the right cross section range and that dm - nucleon scattering interactions are indeed totally irrelevant in astrophysics .
we investigate cross sections between the astrophysical and underground limits , and show that @xmath4 is large enough for earth to efficiently capture dm .
incoming dm will scatter off nucleons , lose energy , and become gravitationally captured once below earth s escape velocity ( section 4 ) .
if this capture is maximally efficient , the rate is @xmath5 s@xmath6 .
the gravitationally - captured dm will drift to the bottom of the potential well , earth s core .
self - annihilation results if the dm is its own antiparticle , and we assume standard model final state particles so that these products will deposit nearly all their energy in the core . inside a region in the @xmath4@xmath2 plane that will be defined ,
too much heat would be produced relative to the actual measured value of earth s heat flow .
the maximal heating rate obtained via macroscopic considerations is @xmath7 3330 terawatts ( tw ) , and follows the maximal capture rate , assuming that earth is opaque with a geometric cross section .
note that the flux of dm scales as @xmath8 , while the heat energy from annihilations scales as @xmath2 , _ yielding a heat flow that is independent of dm mass_. the efficient capture we consider leads to a very similar heating rate , though it is based on a realistic calculation of microscopic dm - nucleon scattering , as discussed below .
dm interactions with earth have been previously studied in great detail , e.g. , refs @xcite , but those investigations generally considered only weak cross sections for which capture is inefficient . in our analysis , the @xmath4 exclusion region arises from the captured dm s self - annihilation energy exceeding earth s internal heat flow .
this region is limited below by the efficient capture of dm ( section 4 ) , and above by being weak enough to allow sufficient time for the dm to drift to the core ( section 5 ) .
these two limits define the region in which dm heating occurs .
why is it important ?
earth s received solar energy is large , about @xmath9 tw @xcite , but it is all reflected or re - radiated .
the internal heat flow is much less , about 44 tw ( section 3 ) @xcite . inside this
bounded region for @xmath4 , dm heating would exceed the measured rate by about two orders of magnitude , and therefore is not allowed .
we will show that this appears to close the window noted above in fig .
[ regions ] , up to about @xmath10 gev . in order to be certain of this ,
however , we call for new analyses of the aforementioned constraints , especially the exact region excluded by cdms and other underground detectors .
our emphasis is not on further debate of the details of specific open gaps , but rather on providing a new and independent constraint . in table
i , we summarize the heat values relevant to this paper . while the origin of earth s heat flow is not completely understood , we emphasize that we are not trying to account for any portion of it with heating from dm . while it appears that the window on strongly interacting dm is now closed over a huge mass range , more detailed analyses are needed in order to be absolutely certain .
our calculations are conservative , and the true excluded region is likely to be larger
. it would be especially valuable to have new analyses of the astrophysical limits and the underground detector constraints .
this would give greater certainty that no sliver of the window is still open .
for dm masses in the range 1 gev to 10@xmath1 gev , the upper limits on the dm scattering cross section with nucleons from cdms and other underground experiments have been shown to be true upper limits .
thus the dm does indeed appear to be very weakly interacting , and it will be challenging to detect it .
we are grateful for discussions and advice from nicole bell , eric braaten , matt kistler , jordi miralda - escud , jerry newsom , wendy panero , gary steigman , louie strigari , ralph von frese , terry walker , and especially laura baudis , andy gould , stefano profumo , and hasan yksel .
gdm was supported by the department of energy grant de - fg02 - 91er40690 , jfb by the national science foundation career grant phy-0547102 , and gb during part of this project by the helmholtz association of national research centres under project vh - ng-006 .
g. jungman , m. kamionkowski and k. griest , phys .
* 267 * , 195 ( 1996 ) [ arxiv : hep - ph/9506380 ] ; l. bergstrom , rept .
phys . * 63 * , 793 ( 2000 ) [ arxiv : hep - ph/0002126 ] .
g. bertone , d. hooper and j. silk , phys .
rept . * 405 * , 279 ( 2005 ) [ arxiv : hep - ph/0404175 ] ; g. d. starkman , a. gould , r. esmailzadeh and s. dimopoulos , phys . rev .
d * 41 * , 3594 ( 1990 ) .
x. l. chen , s. hannestad and r. j. scherrer , phys .
d * 65 * , 123515 ( 2002 ) [ arxiv : astro - ph/0202496 ] .
r. h. cyburt , b. d. fields , v. pavlidou and b. d. wandelt , phys .
d * 65 * , 123503 ( 2002 ) [ arxiv : astro - ph/0203240 ] .
d. s. akerib _ et al .
_ , nucl .
instrum .
a * 559 * , 390 ( 2006 ) ; d. s. akerib _ et al . _
[ cdms collaboration ] , phys .
lett . * 96 * , 011302 ( 2006 ) [ arxiv : astro - ph/0509259 ] .
d. s. akerib _ et al .
_ [ cdms collaboration ] , phys . rev .
d * 73 * , 011102 ( 2006 ) [ arxiv : astro - ph/0509269 ] ; i. f. m. albuquerque and l. baudis , phys . rev .
lett . * 90 * , 221301 ( 2003 ) [ arxiv : astro - ph/0301188 ] ; erratum - ibid . *
91 * , 229903 ( 2003 ) .
v. sanglard _ et al . _
[ edelweiss collaboration ] , phys .
d * 71 * , 122002 ( 2005 ) [ arxiv : astro - ph/0503265 ] .
r. bernabei _ et al .
_ , phys .
* 83 * 4918 ( 1999 ) .
j. rich , r. rocchia and m. spiro , phys .
b * 194 * , 173 ( 1987 ) .
b. d. wandelt , r. dave , g. r. farrar , p. c. mcguire , d. n. spergel and p. j. steinhardt , in d. b. cline ( ed . ) , _ sources and detection of dark matter and dark energy in the universe _ ( springer - verlag , new york , 2001 ) [ arxiv : astro - ph/0006344 ] .
a. l. erickcek , p. j. steinhardt , d. mccammon and p. c. mcguire , arxiv:0704.0794 [ astro - ph ] . l. m. krauss , m. srednicki and f. wilczek , phys . rev .
d * 33 * , 2079 ( 1986 ) .
t. k. gaisser , g. steigman and s. tilav , phys .
d * 34 * , 2206 ( 1986 ) .
a. gould , astrophys .
j. , * 321 * 560 ( 1987 ) .
a. gould , astrophys .
j. , * 321 * 571 ( 1987 ) .
a. gould , astrophys .
j. , * 328 * 919 ( 1988 ) .
a. gould , j.a .
frieman , and k. freese , phys .
d. * 39 * , 1029 ( 1989 ) .
a. gould , astrophys .
j. * 368 * 610 ( 1991 ) .
a. gould , astrophys .
j. * 387 * , 21 ( 1992 ) a. gould and s.m.k .
alam , astrophys .
j. , * 549 * 72 ( 2001 ) [ arxiv : astro - ph/9911288 ] .
m. fukugita , p. hut , and d. n. spergel , iassns - ast-88 - 26 ( 1988 ) , unpublished preprint ( available via spires ) .
s. dimopoulos , d. eichler , r. esmailzadeh and g. d. starkman , phys .
d * 41 * , 2388 ( 1990 ) .
m. kawasaki , h. murayama and t. yanagida , prog .
phys . * 87 * , 685 ( 1992 ) .
s. abbas and a. abbas , astropart .
* 8 * , 317 ( 1998 ) [ arxiv : astro - ph/9612214 ] . s. mitra , phys .
d * 70 * , 103517 ( 2004 ) [ arxiv : astro - ph/0408341 ] .
g. zaharijas and g. r. farrar , phys .
d * 72 * , 083502 ( 2005 ) [ arxiv : astro - ph/0406531 ] .
g. r. farrar and g. zaharijas , phys .
lett . * 96 * , 041302 ( 2006 ) [ arxiv : hep - ph/0510079 ] . m. zeilik and s. gregory , _
introductory astronomy and astrophysics , fourth edition _ ( saunder college publishing , new york , 1998 ) .
albuquerque , l. hui , and e.w .
kolb , phys .
d , * 64 * , 083504 ( 2001 ) .
[ arxiv : hep - ph/0009017 ] .
p. crotty , phys .
d * 66 * , 063504 ( 2002 ) [ arxiv : hep - ph/0205116 ] .
i. f. m. albuquerque , j. lamoureux and g. f. smoot , phys .
d * 66 * , 125006 ( 2002 ) [ arxiv : hep - ph/0204301 ] .
s. desai _ et al .
_ [ super - kamiokande collaboration ] , phys . rev .
d * 70 * , 083523 ( 2004 ) [ erratum - ibid .
d * 70 * , 109901 ( 2004 ) ] [ arxiv : hep - ex/0404025 ] .
a. bottino , n. fornengo , g. mignola and l. moscoso , astropart .
phys . * 3 * , 65 ( 1995 ) [ arxiv : hep - ph/9408391 ] .
a. achterberg _ et al .
_ [ amanda collaboration ] , astropart .
* 26 * , 129 ( 2006 ) .
m. cirelli , n. fornengo , t. montaruli , i. sokalski , a. strumia and f. vissani , nucl .
b * 727 * , 99 ( 2005 ) [ arxiv : hep - ph/0506298 ] .
l. bergstrom , j. edsjo and m. kamionkowski , astropart .
* 7 * , 147 ( 1997 ) [ arxiv : astro - ph/9702037 ] .
r. s. chivukula , a. g. cohen , s. dimopoulos and t. p. walker , phys .
lett . * 65 * , 957 ( 1990 ) .
h. nelson , `` searches for dark matter , '' lecture at slac summer institute , 2003 , http://www.slac.stanford.edu/econf/c0307282/lec_notes/ + nelson.html#lec_1 p. c. mcguire , ph.d . thesis ( u. arizona , 1994 ) http://epsci.wustl.edu/mcguire/cvw_pub.html e. k. shirk and p. b. price , astrophys .
j. * 220 * , 719 ( 1978 ) .
d. p. snowden - ifft , s. w. barwick , and p. b. price , astrophys .
j. * 364 * , l25 ( 1990 ) .
j. d. anderson , e. l. lau , t. p. krishner , d. a. dicus , d. c. rosenbaum , and v. l. teplitz , astrophys .
j. * 448 * , 885 ( 1995 ) . h. pollack , s. hurter , and j. johnson , rev .
geophys . * 31 * , 267 ( 1993 ) .
s. karato , _ the dynamic structure of the deep earth _ , ( princeton university press , princeton , 2003 ) .
t. araki _ et al .
_ , nature * 436 * , 499 ( 2005 )
. l. m. krauss , s. l. glashow and d. n. schramm , nature * 310 * , 191 ( 1984 ) . c. g. rothschild ,
m. c. chen and f. p. calaprice , geophys .
lett . * 25 * , 1083 ( 1998 ) [ arxiv : nucl - ex/9710001 ] .
r. s. raghavan , s. schonert , s. enomoto , j. shirai , f. suekane and a. suzuki , phys .
lett . * 80 * , 635 ( 1998 ) .
g. fiorentini , m. lissia , f. mantovani and b. ricci , phys .
b * 629 * , 77 ( 2005 ) [ arxiv : hep - ph/0508048 ] .
g. fiorentini , m. lissia , f. mantovani and r. vannucci , phys .
d * 72 * , 033017 ( 2005 ) [ arxiv : hep - ph/0501111 ] .
b. d. fields and k. a. hochmuth , earth moon planets * 99 * , 155 ( 2006 ) [ arxiv : hep - ph/0406001 ] .
g. l. fogli , e. lisi , a. palazzo and a. m. rotunno , earth moon planets * 99 * , 111 ( 2006 ) [ arxiv : physics/0608025 ] . s. t. dye _ et al .
_ , arxiv : hep - ex/0609041 .
k. griest and d. seckel , nucl .
b. * 283 * , 681 ( 1987 ) [ erratum - ibid * 296 * , 1034 ( 1988 ) ] .
k. griest and m. kamionkowski , phys .
* 64 * ( 1990 ) 615 .
l. hui , phys .
* 86 * , 3467 ( 2001 ) [ arxiv : astro - ph/0102349 ] .
j. f. beacom , n. f. bell and g. d. mack , arxiv : astro - ph/0608090 .
a. gould and s. m. khairul alam , astrophys .
j. * 549 * , 72 ( 2001 ) [ arxiv : astro - ph/9911288 ] . j. lundberg and j. edsjo , phys .
d * 69 * , 123505 ( 2004 ) [ arxiv : astro - ph/0401113 ] .
k. enqvist , a. jokinen , t. multamaki and i. vilja , phys .
b * 526 * , 9 ( 2002 ) [ arxiv : hep - ph/0111348 ] .
t. banks , j. d. mason and d. oneil , phys .
d * 72 * , 043530 ( 2005 ) [ arxiv : hep - ph/0506015 ] .
m. y. khlopov , pisma zh .
teor . fiz .
* 83 * , 3 ( 2006 ) [ jetp lett .
* 83 * , 1 ( 2006 ) ] [ arxiv : astro - ph/0511796 ] ; m. y. khlopov , arxiv : astro - ph/0607048 . j. d. vergados , arxiv : hep - ph/0601064 . l. anchordoqui and f. halzen , annals phys . * 321 * , 2660 ( 2006 ) [ arxiv : hep - ph/0510389 ] .
d. hooper and a. m. taylor , jcap * 0703 * , 017 ( 2007 ) [ arxiv : hep - ph/0607086 ] .
h. baer , a. mustafayev , e. k. park and s. profumo , jhep * 0507 * , 046 ( 2005 ) [ arxiv : hep - ph/0505227 ] .
r. n. mohapatra and v. l. teplitz , phys .
lett . * 81 * , 3079 ( 1998 ) [ arxiv : hep - ph/9804420 ] .
d. javorsek , d. elmore , e. fischbach , d. granger , t. miller , d. oliver and v. teplitz , phys .
d * 65 * , 072003 ( 2002 ) ; d. javorsek , e. fischbach and d. elmore , phys .
d * 67 * , 034015 ( 2003 ) ; v. teplitz , d. javorsek , e. fischbach and r. mohapatra , nucl .
suppl . * 124 * , 205 ( 2003 ) ; b. s. gaudi , s. seager and g. mallen - ornelas , astrophys . j. * 623 * , 472 ( 2005 ) [ arxiv : astro - ph/0409443 ] . g. a. bakos _ et al .
_ , astrophys .
j. * 656 * , 552 ( 2007 ) [ arxiv : astro - ph/0609369 ] .
a. burrows , i. hubeny , j. budaj and w. b. hubbard , astrophys .
j. * 661 * , 502 ( 2007 ) [ arxiv : astro - ph/0612703 ] . | we point out a new and largely model - independent constraint on the dark matter scattering cross section with nucleons , applying when this quantity is larger than for typical weakly interacting dark matter candidates . when the dark matter capture rate in earth is efficient
, the rate of energy deposition by dark matter self - annihilation products would grossly exceed the measured heat flow of earth .
this improves the spin - independent cross section constraints by many orders of magnitude , and closes the window between astrophysical constraints ( at very large cross sections ) and underground detector constraints ( at small cross sections ) . in the applicable mass range , from @xmath0 1 to @xmath0 10@xmath1 gev
, the scattering cross section of dark matter with nucleons is then bounded from _ above _ by the latter constraints , and hence must be truly weak , as usually assumed . |
[ [ assignment - setting ] ] assignment setting + + + + + + + + + + + + + + + + + + an assignment problem is a triple @xmath0 such that @xmath1 is the set of agents , @xmath2 is the set of objects , and the preference profile @xmath3 specifies for each agent @xmath6 his preferences @xmath11 over objects in @xmath5 .
we write @xmath12 to denote that agent @xmath6 values object @xmath13 at least as much as object @xmath14 and use @xmath15 for the strict part of @xmath11 , i.e. , @xmath16 iff @xmath12 but not @xmath17 .
we will assume that the agents have strict preferences and that @xmath18 is represented by a comma separated list as follows : @xmath19 consider an assignment problem in which @xmath20 , @xmath21 and the preferences @xmath22 are as follows .
@xmath23 a random assignment @xmath8 is a @xmath24 matrix @xmath25 $ ] such that @xmath26 $ ] for all @xmath4 , and @xmath27 , ; @xmath28 for all @xmath27 ; and @xmath29 for all @xmath4 .
the value @xmath30 represents the probability of object @xmath31 being allocated to agent @xmath6 .
each row @xmath32 represents the allocation of agent @xmath6 .
the set of columns correspond to the objects @xmath33 .
a feasible random assignment is _ deterministic _ if @xmath34 for all @xmath4 and @xmath35 .
a _ uniform assignment _ is a random assignment in which each agent has probability @xmath10-th of getting each object . for and
assignment problem in which @xmath20 , @xmath21 , the following is an example of a random assigment : @xmath36 in @xmath8 , the probability of agent @xmath37 getting @xmath38 is @xmath39
. given two random assignments @xmath8 and @xmath9 , @xmath40 i.e. , an agent @xmath6 _ sd prefers _ allocation @xmath41 to allocation @xmath42 if @xmath43 an assignment @xmath8 is _ sd - efficient _ is there exists no assignment @xmath9 such that @xmath44 for all @xmath4 and @xmath45 for some @xmath4 .
an assignment is _ ex post efficient _ if it be can represented as a probability distribution over the set of pareto optimal assignments .
we say that a deterministic assignment @xmath9 is _ consistent _ with a random assignment @xmath8 if for each @xmath46 , we have that @xmath47 .
a deterministic assignment can be represented by a permutation matrix in which an entry of one denotes the row agent getting the column object .
a _ decomposition _ of a random assignment @xmath8 is a sum @xmath48 such that @xmath49 $ ] for @xmath50 , @xmath51 , and each @xmath52 is a permutation matrix ( consistent with @xmath8 ) .
consider a random assignment @xmath53 then , the following is a valid decomposition of the assignment . @xmath54
a deterministic assignment is pareto optimal iff it is ex post efficient iff it is sd - efficient .
therefore both sd - efficiency and ex post efficiency are natural generalizations of pareto optimality in the context of random assignments .
the convex hull of pareto optimal discrete assignments will be denoted by @xmath55 .
an efficiency concept @xmath56 is _ combinatorial _ if for any two random assignments @xmath8 and @xmath9 such that @xmath47 if and only if @xmath57 , it holds that @xmath8 is efficient with respect to @xmath56 if and only if @xmath9 is efficient with respect to @xmath56 . [ [ insights - into - ex - post - efficiency ] ] insights into ex post efficiency + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + before we examine ex post efficient random assignment , we review some characterizations of deterministic pareto optimal assignments . the first characterization is with respect to deterministic assignment algorithm called _ serial dictatorship
_ which takes as a parameter a permutation @xmath58 over the set of agents .
serial dictatorship @xmath59 lets the agents in the permutation @xmath58 serially take their most preferred object that has not yet been allocated until each agent has an object .
[ fact : po : sd ] each pareto optimal assignment is an outcome of applying serial dictatorship with respect to some permutation of the agents . it follows that rsd ( random serial dictator ) rule which takes some @xmath58 uniformly at random and then implements serial dictatorship with respect to it is ex post efficient .
the next characterization of pareto optimal is graph - theoretical . for an assignment problem @xmath0 and deterministic assignment @xmath8 , the _ corresponding graph _
@xmath60 is such that @xmath61 and @xmath62 is defined as follows . for all @xmath4 and all @xmath35 ,
* @xmath63 iff @xmath64 and * @xmath65 iff @xmath66 where @xmath67 .
[ fact : po : ttc ] an assignment is pareto optimal if and only if its corresponding graph does not admit a cycle . the cycle is referred to as a trading cycle because it represents a trade of objects in which each agent in the cycle gets the object that he was pointing to ( see e.g. , * ? ? ?
we will use both facts [ fact : po : sd ] and [ fact : po : ttc ] in our arguments .
an ex post efficient assignment can be computed in polynomial time .
an outcome of serial dictatorship is ex post efficient .
if we also require anonymity , then a maximum utility matching for utilities consistent with the ordinal preferences is also sd - efficient and hence ex post efficient .
an ex post random assignment can have multiple pareto optimal decompositions .
consider three agents with identical preferences .
it can even be the case that an ex post efficient lottery can be expressed by a lottery over pareto dominated assignments ( example 2 , * ? ? ?
we consider the problem of testing ex post efficiency of a random assignment .
since we are interested in checking whether a random assignment can be decomposed into pareto optimal assignment , we are reminded of birkhoff s algorithm that can decompose any given random assignment ( represented by a bistochastic matrix ) into a convex combination of at most @xmath68 deterministic assignments ( represented by permutation matrices ) @xcite .
[ [ birkhoffs - algorithm ] ] birkhoff s algorithm + + + + + + + + + + + + + + + + + + + + birkhoff s algorithm works as follows .
we initialize @xmath6 to @xmath37 . for a bistochastic matrix @xmath69 , a permutation matrix @xmath52 with respect to @xmath69
is guaranteed to exist .
@xmath69 is set to @xmath70 where @xmath71 is such that no entry @xmath72 is negative but there is at least one extra zero entry in @xmath73 than in @xmath69 .
index @xmath6 is incremented by one .
the updated @xmath69 is again bistochastic .
the process is repeated ( say @xmath74 times ) until @xmath69 is the zero matrix . then @xmath75 .
when birkhoff s algorithm identifies a permutation matrix @xmath52 , multiplies it by a constant and then subtracts it from the input matrix representing the random assignment , we will refer to such a step as _ decomposing with respect to a permutation matrix_. if a random assignment is sd - efficient , then _ any _ decomposition of the assignment is a decomposition into pareto optimal deterministic assignments . on the other hand ,
if a random assignment is not sd - efficient , it may not admit a decomposition into pareto optimal deterministic assignments .
one may wonder whether we can modify birkhoff s algorithm to check whether a given random assignment is ex post efficient or not .
note that if we decompose with respect to some permutation matrices , then we can decompose in any order .
we first show that checking whether there exists a pareto optimal permutation matrix that can be used to further decompose an assignment is np - complete . checking whether there exists a pareto optimal deterministic assignment consistent with the given random assignment is np - complete .
the following problem is np - complete : serialdictatorshipfeasibility check whether there exists a permutation of agents for which serial dictatorship gives a particular object to an agent @xcite .
we present a reduction from serialdictatorshipfeasibility to the problem of checking whether there exists a pareto optimal deterministic assignment consistent with the given random assignment consider a random assignment @xmath8 in which agent @xmath6 gets @xmath7 with probability one and all other agents gets each object in @xmath76 with non - zero probability .
we argue that serialdictatorshipfeasibility has a yes instance if and only if @xmath8 admits a pareto optimal deterministic assignment consistent with it .
if there exists a pareto optimal deterministic assignment consistent with @xmath8 , then this implies that there exists a pareto optimal assignment in which @xmath6 gets @xmath7 .
this means that there exists some permutation @xmath58 such that @xmath77 .
assume that there exists some permutation @xmath58 such that @xmath77 .
then the deterministic assignment corresponding to @xmath59 is a pareto optimal permutation matrix that is consistent with @xmath8 .
the statement above does not imply that checking whether a random assignment is ex post efficient is np - complete .
it is even not clear whether testing ex post efficiency of a random assignment is in np .
the reason is that the certificate for membership may not in principle be polynomial - sized .
however , we show that verifying an ex post efficient random assignment is in np but the problem is np - complete .
[ mainthm ] testing ex post efficiency of a random assignment is np - complete .
we first show that testing ex post efficiency of a random assignment is in np .
it is sufficient to show that an ex post efficient random assignment admits a polynomial - sized pareto optimal decomposition .
consider the @xmath78 dimensional euclidean space of all @xmath79 matrices .
let @xmath80 denote the minimal polytope containing all determinisitc pareto optimal allocations .
by definition @xmath80 is the set of all ex post efficient allocations .
carathodory s theorem implies that any ex post efficient allocations must be a convex combination of no more than @xmath81 deterministic pareto optimal allocations
( because all vertices in @xmath80 are pareto optimal allocations ) .
dimensional subspace of @xmath82 because its feasible points has to satisfy @xmath83 equality constraints and one of which is redundant . ]
we prove the np - hardness via a reduction from 3-sat . given a 3-sat instance @xmath84 , where @xmath85 is the set of binary variables and @xmath86 is the set of clauses .
let each clause @xmath87 , where for @xmath88 , @xmath89 is either @xmath90 or @xmath91 with @xmath92 .
the assumption that @xmath92 will be crucial in the proof . given a 3-sat instance @xmath93
, we build an assignment problem @xmath94 as follows .
let @xmath95 .
for each agent @xmath96 , let @xmath97 denote the corresponding dummy agent .
let @xmath98 and @xmath99 .
that is , @xmath100 .
we will show that in the decompositions @xmath101 s are `` copies '' of @xmath102 and @xmath103 s are `` copies '' of @xmath104 .
for each @xmath96 , @xmath106 . to define the preferences of the agents , we first introduce the following notation . for any literal @xmath89 in clause @xmath107 , we let @xmath108 denote the item that corresponds to the value of @xmath90 that fails @xmath89 . more precisely , @xmath109 for each @xmath110 and @xmath87 , where @xmath89 is a literal of variable @xmath90 , we let @xmath111 , @xmath112 , and @xmath113 . for any @xmath114 and @xmath110 , if @xmath115 is not defined above then @xmath116 .
moreover , we let @xmath117 .
for example , for @xmath118 , @xmath119 , @xmath120 , and @xmath121 , we have @xmath122 @xmath123 @xmath124 agents preferences are defined in two tables : preferences for @xmath125 s are in table [ tab : x ] and preferences for all @xmath102 s are in table [ tab : c ] . .preferences for @xmath125 s .
[ tab : x ] [ cols=">,<",options="header " , ] clearly for all @xmath110 , @xmath126 is not in any cycle because they have no outgoing edges .
the only possibility of cycles are through @xmath127 , then through truncated figure [ fig : demandx ] , where all @xmath104 and @xmath128 are removed . because @xmath129 satisfies @xmath93 , each potential path from @xmath101 to @xmath102
is blocked by at least one @xmath103 .
meanwhile , there is no cycle involving only @xmath103 s or involving @xmath103 s with different @xmath107 s , because any @xmath103 only has outgoing edges to @xmath115 , which is either ( 1 ) empty , or ( 2 ) contains another @xmath130 with @xmath131 ( recall we assume that @xmath92 for the three literals in @xmath132 ) or @xmath133 .
this proves that @xmath134 is pareto optimal .
@xmath135 is pareto optimal .
the demand graph of @xmath136 in @xmath135 with all outgoing edges of @xmath102 s is illustrated in figure [ fig : demandc ] ( b ) .
clearly @xmath137 and all @xmath101 s are not in any cycle because they have no outgoing edges .
after they are removed , @xmath126 s have no outgoing edges .
@xmath102 can also be removed because none of the remaining agents demand @xmath138 .
the only remaining agents are those in the truncated figure [ fig : demandx ] , where all @xmath104 and @xmath128 are removed , and as in claim [ claim : m1 ] , there is no cycle among them .
checking membership of a point in @xmath55 is np - complete .
since @xmath55 is a convex combination of pareto optimal determistic assignments , it contains all the ex post efficient points .
optimizing a linear functions over @xmath55 is np - complete . since testing membership in @xmath55 , the statements follows from the equivalence between optimizing over a poltope and implementing a separation oracle over over a polytope @xcite .
although we have shown that testing whether a given assignment is ex post efficiency is np - complete , we leave open the case when the assignment is uniform . we now show that in the random assignment problem , ex post efficiency is not combinatorial .
this is already a contrast with a stronger notion of efficiency called _ sd - efficiency _ which depends solely on the support of the random allocations .
a trading cycle of size @xmath139 is _ consistent _ with random assignment @xmath8 if it consists of @xmath140 agents and @xmath140 objects , each objects points to an agent , each object to an agent in the cycle and the cycle satisfies the following constraints : @xmath141 @xmath57 if object @xmath7 points to the agent @xmath6 , and @xmath142 @xmath143 if agent @xmath6 points to object @xmath144 .
( lemma 3 , * ? ? ?
* ) proved that a random assignment is not sd - efficiency iff it admits a trading cycle consistent with it .
the fact that sd - efficiency is combinatorial follows from its characterization .
in contrast , ex post efficiency is not combinatorial .
[ th : expost - not - combinatorial ] ex post efficiency is not combinatorial . consider the following assignment problem .
@xmath145 the following assignment @xmath8 is a result of rsd and hence ex post efficient : @xmath146 now consider the following assignment : @xmath147 note that @xmath9 is a random assignment such that @xmath47 if and only if @xmath57 .
then in each pareto optimal matrix that can be used to decompose the assignment , due to pareto optimality , agents @xmath37 and @xmath148 get an object each from the following sets of objects @xmath149 , @xmath150 , and @xmath151 . agent 1 and 2 can not get the set @xmath152 in a pareto optimal assignment because if it were the case then there exists a serial dictatorship in which agent 3 or 4 take @xmath153 before @xmath154 is allocated . in order for agent @xmath37 to get @xmath39 of @xmath155 , we need to use the @xmath39 time the pareto optimal assignment in which @xmath155 gts @xmath39 . but
this means that agent @xmath148 gets @xmath38 @xmath39 of the time .
but this is not possible since agent @xmath148 gets @xmath38 @xmath156 of time .
in this section , we formalize a new efficiency concept called _ ex post efficiency_. recall that a random assignment is ex post efficient if it can be represented as a convex combination of pareto optimal deterministic assignments .
we say that a random assignment is _ robust ex post efficient _ if any decomposition of the assignment consists of pareto optimal deterministic assignments .
the following is a useful characterization of robust ex post efficiency .
[ th : robust - eff - charac ] an assignment is robust ex post efficient iff it does not admit a non - pareto optimal deterministic assignment consistent with it .
assume that there exists no non - pareto optimal deterministic assignment consistent with the assignment @xmath8 .
then each time , @xmath8 is decomposed with respect to a deterministic assignment , it is with respect to a pareto optimal deterministic assignment . by birkhoff s theorem ,
the updated @xmath8 is still bistochastic after the decomposition .
hence any decomposition of @xmath8 is into pareto optimal deterministic assignment and hence @xmath8 is robust ex post efficient .
now let us assume that that there exists a non - pareto optimal deterministic assignment consistent with the assignment @xmath8 .
we then decompose @xmath8 with respect to such an assignment using birkhoff s algorithm and continue decomposing it . since the decomposition contains at least non pareto optimal deterministic assignment ,
hence @xmath8 is not robust ex post efficient .
we point out that sd - efficiency implies robust ex post efficiency which implies ex post efficiency .
note that in the general domain of voting , ex post efficiency and robust ex post efficiency are equivalent .
sd - efficiency implies robust ex post efficiency which implies ex post efficiency .
we first show that if an assignment is not ex post efficient , then it is not robust ex post efficient .
if an assignment @xmath9 is not ex post efficient , then there does not exist _ any _ decomposition of the assignment into pareto optimal deterministic assignments .
then @xmath9 is not robust ex post efficient .
we now show that if an assignment is not robust ex post efficient , then it is not sd efficient .
if an assignment @xmath9 is not robust ex post efficient , then there exists at least one decomposition of @xmath9 into deterministic assignments in which at least one deterministic assignment is not pareto optimal . but
this implies that there exists at least one deterministic assignment @xmath69 consistent with @xmath9 that is not pareto optimal .
by fact [ fact : po : ttc ] , @xmath69 admits a trading cycle @xmath157 . since @xmath69 is consistent with @xmath9 , each object @xmath7 that points to an agent @xmath6 in the cycle is such that @xmath47 .
now consider a random assignment @xmath8 which is the same as @xmath9 except that each agent @xmath6 gets @xmath158 probability less of the object that was pointing to it and @xmath158 probability more of the object he points to in cycle @xmath157 .
assignment @xmath9 is such that @xmath40 for all @xmath4 and @xmath159 for all @xmath160 .
hence @xmath9 is not sd - efficient .
sd - efficiency is a strictly stronger concept than robust ex post efficiency .
although @xcite did not explicitly define the concept robust ex post efficiency , they showed that a random assignment that has only one decomposition which is a randomization over pareto optimal assignments does not satisfy sd - efficiency .
hence robust ex post efficiency does not imply sd - efficiency .
next , we show that ex post efficiency does not imply robust ex post efficiency so that robust ex post efficiency is a strictly stronger concept than ex post efficiency .
[ remark : expostnotrobust ] consider the assignment problem assignment @xmath8 in the proof of theorem [ th : expost - not - combinatorial ] .
assignment @xmath8 which is the result of random serial dictatorship is ex post efficient .
however , there is a deterministic assignment @xmath69 consistent with @xmath8 that is not pareto optimal . hence @xmath8 is not robust ex post efficient .
hence , we get the following .
the random serial dictatorship mechanism may return an assignment that is not robust ex post efficient .
the simple theorem above strengthens the observation of @xcite that random serial dictatorship is not sd - efficient .
any robust ex post efficient point must lie on a face of the assignment polytope where all extreme points of the face are vectors of pareto optimal assignments .
suppose not , then there is a robust ex post efficient point @xmath125 that is in the interior of the assignment polytope .
then there is a @xmath158 ball centered around @xmath125 that is contained in the polytope .
let @xmath8 be an extreme point that is not pareto optimal .
choose @xmath161 small enough such that @xmath162 lies in the @xmath158 ball .
then one can write @xmath125 as a convex combination of @xmath163 and @xmath8 . moreover , since @xmath163 belongs to the assignment polytope , @xmath163 can in turn be written as a convex combination of vectors of assignments .
hence , we have expressed x as a convex combination of extreme points of the assignment polytope and one of which is not a pareto optimal assignment .
so @xmath125 can not be robust ex post efficient .
hence , any robust ex post efficient must lie on a face of the assignment polytope .
we can then use the same argument to say that all extreme points of the face must correspond to pareto optimal assignments .
we can consider similar computational questions regarding robust ex post efficiency : _ what is the computational complexity of checking whether an assignment is robust ex post efficient ? _
our first observation is that the problem is in conp .
[ remark : inconp ] the problem of checking whether a random assignment is robust ex post efficient is in conp . by theorem [
th : robust - eff - charac ] , any non - pareto optimal deterministic assignment consistent with the random assignment is a witness that the random assignment is not robust ex post efficient . also note that it can be checked in linear time whether a given assignment is pareto optimal ( fact [ fact : po : ttc ] ) .
due to the characterization of robust efficiency in theorem [ th : robust - eff - charac ] , the problem of testing robust ex post efficiency is equivalent to checking whether there exists a constrained non pareto optimal assignment .
previously , it has been shown that checking whether there exists a constrained pareto optimal assignment is np - complete @xcite .
next we give a simple necessary condition for robust ex post efficiency .
if a random assignment is robust ex post efficient , there exists no consistent deterministic assignment in which no agent gets his most preferred object .
the argument is as follows . for a random assignment @xmath8
, it is sufficient to show that if there exists a consistent deterministic assignment @xmath9 in which no agent gets his most preferred object , then that assignment is not pareto optimal and hence @xmath8 is not robust ex post efficient .
if no agent gets their most preferred object , then there exists no permutation over the agents under serial dictatorship returns @xmath8 . therefore ,
by fact [ fact : po : sd ] , @xmath9 is not pareto optimal .
hence by theorem [ th : robust - eff - charac ] , @xmath8 is not robust ex post efficient .
next , we show that robust ex post efficiency is combinatorial .
[ th : robust - is - combinatorial ] robust ex post efficiency is combinatorial .
if @xmath8 is not robust ex post efficient , then there exists a deterministic assignment @xmath164 where @xmath57 for @xmath165 and @xmath164 is not pareto optimal .
hence for deterministic assignment @xmath164 , @xmath47 for @xmath165 .
thus @xmath164 is not robust ex post efficient because it admits a pareto dominated deterministic assignment that is consistent with it .
the same argument also shows that is @xmath9 is not robust ex post efficient , then @xmath8 is not robust ex post efficient . in the previous section
, we mentioned that the complexity of testing whether the uniform assignment is ex post efficient is still open . on the other hand
, it can be easily checked whether the uniform assignment is robust ex post efficient .
the uniform assignment is robust ex post efficient if and only if the preferences are unanimous .
the arguments is as follows .
if preferences are unanimous then every assignment is pareto optimal .
if preferences are not unanimous , then for some two objects @xmath7 and @xmath144 , at least two agents have opposite preferences over them . in this case ,
any assignment that gives each of @xmath7 and @xmath144 to one of the two agents who prefers it less is not pareto optimal .
two agents are said to of the same _ type _ if they have identical preferences .
we show that if there are a constant number agents , types , then robust ex post efficiency can be checked in polynomial time .
[ lemma : agent - types ] if there is a trading cycle consistent with random assignment that contain multiple agents of the same type , then there also exists a pareto cycle consistent with the random assignment , in which there is at most one agent of the same type .
we show that if there exists a pareto cycle also containing @xmath102 agents of the same type , then there exists a pareto cycle containing @xmath166 agents of the same type .
consider the pareto cycle in which the @xmath102 agents of the same type are @xmath167 where agent @xmath168 is the agent who has the least preferred object pointing to it among all the objects that point towards the @xmath102 agents .
then agent @xmath168 can point directly to the object that agent @xmath169 points to thereby forming a smaller cycle only including @xmath170 .
if there are a constant number of agent types , robust ex post efficiency can be checked in polynomial time .
we will use the characterization of robust efficiency in theorem [ th : robust - eff - charac ] to propose an algorithm . since we will check for a pareto dominated assignment , by fact [ fact : po : ttc ] ,
such an assignment admits a trading cycle consistent with the random assignment . from lemma [ lemma : agent - types ] , we know that if there is a trading cycle consistent with the random assignment then there is also a trading cycle consistent with the assignment in which there is at most one agent of one type . in order to check for robust efficiency ,
we need to check whether there is a trading cycle for which the agents outside the trading cycle are assigned an object each from outside the trading cycle consistent with the random assignment .
we will use the fact that if there exists a trading cycle consistent with the assignment which contains multiple agents of the same type and if the agents outside the trading cycle are also perfectly matched to an object consistent with the random assignment , then there exists a trading cycle consistent with the assignment which contains at most one agent of each type and for which the agents outside the trading cycle are also perfectly matched to an object consistent with the random assignment .
note that are constant number @xmath171 of possible orderings in which agent types are present in a pareto cycle in which there is at most one agent of the same type . for each of the agents types in the cycle , there may be @xmath172 options of objects that point to him .
for each agent type there are @xmath172 agents that could be used for that type .
therefore , the number of pareto cycles to be considered is at most @xmath173 where in each pareto cycle there are at most @xmath140 agents and @xmath140 objects . for each of the ( short ) pareto cycles considered that are not more than @xmath173
, we need to check whether the other agents can be perfectly matched to the unallocated objects .
this can again be checked in polynomial time via the algorithm to check whether a perfect matching exists for agents _ not _ in the pareto cycle .
it will be interesting to check whether there exists a fixed parametrized algorithm with parameter number of agent types .
we examined different aspects of ex post efficiency of random assignments .
one of the most important technical result in the paper is that testing ex post efficiency is np - complete .
the result contrast with the followings facts ( 1 ) ex post stability in the two sided marriage setting can be tested in polynomial time via an lp @xcite ; ( 2 ) sd - efficiency can be tested in polynomial time ; and ( 3 ) an ex post efficient assignment can be computed in polynomial time .
one implication of the np - completeness result is that the set of ex post efficient assignments can not be characterized compactly .
unless p = np , there is no polynomial - time separation oracle for the convex hull of pareto optimal assignments . due to the well - known equivalence of optimization and separation
, it follows that optimizing a linear function over the convex hull of pareto optimal assignments is np - complete as well .
another corollary is that for the marriage market model , testing ex post effiency is np - complete because the assignment setting can be viewed as a marriage market in which one side of the market is completely indifferent .
we also showed that in the random assignment problem , robust ex post efficiency is combinatorial whereas ex post efficiency is not .
the finding that ex post efficiency is not combinatorial contrasts with the fact that in randomized voting , ex post efficiency of a lottery simply depends on its support .
a number of open problems arise as a result of this study .
the complexity of checking whether a random assignment is robust ex post efficient is open .
similarly , for a constant number of agent types , the complexity of checking whether a random assignment is ex post efficient is also open .
computational aspects of pareto optimal deterministic assignments have been studied in great depth in recent years @xcite .
the more general randomized resource allocation settings provide a suitable ground for further developments in the algorithmic aspects of matching under preferences .
nicta is funded by the australian government through the department of communications and the australian research council through the ict centre of excellence program .
abraham , d. j. , cechlrov , k. , manlove , d. , mehlhorn , k. , 2005 .
pareto optimality in house allocation problems . in : proceedings of the 16th international symposium on algorithms and computation ( isaac ) .
3341 of lecture notes in computer science ( lncs ) .
11631175 .
aziz , h. , brandl , f. , brandt , f. , 2014 .
universal dominance and welfare for plausible utility functions . in : conitzer , v. , easley , d. ( eds . ) ,
proceedings of the 15th acm conference on economics and computation ( acm - ec ) .
forthcoming .
roth , a. e. , rothblum , u. g. , vande vate , j. h. , 1993 .
stable matchings , optimal assignments , and linear programming .
mathematics of operations research 18 ( 4 ) , 803828 803828 803828 803828 803828 .
saban , d. , sethuraman , j. , 2013 .
the complexity of computing the random priority allocation matrix . in : proceedings of the 9th international workshop on internet and network economics ( wine ) .
lecture notes in computer science ( lncs ) .
js1353/pubs / rpcomplexity.pdf . | in the random assignment problem , objects are randomly assigned to agents keeping in view the agents preferences over objects .
a random assignment specifies the probability of an agent getting an object .
we examine the structural and computational aspects of ex post efficiency of random assignments .
we first show that whereas an ex post efficient assignment can be computed easily , checking whether a given random assignment is ex post efficient is np - complete . hence implementing a given random assignment via deterministic pareto optimal assignments
is np - hard .
we then formalize another concept of efficiency called robust ex post efficiency that is weaker than stochastic dominance efficiency but stronger than ex post efficiency .
we present a characterization of robust ex post efficiency and show that it can be tested in polynomial time if there are a constant number of agent types .
it is shown that the well - known random serial dictatorship rule is not robust ex post efficient .
finally , we show that whereas robust ex post efficiency depends solely on which entries of the assignment matrix are zero / non - zero , ex post efficiency of an assignment depends on the actual values . random assignment , efficiency + _ jel _ : c62 , c63 , and c78 pareto optimality has been termed the _
`` single most important tool of normative economic analysis '' _
@xcite .
it appeals to the idea that there should not exist another possible outcome different from the social outcome which all the agents prefer .
we consider pareto optimality in the _ random assignment problem _ which is a fundamental and widely applicable setting in computer science and economics ( see e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in a
_ random assignment problem _ @xmath0 , there is a set of agents @xmath1 , a set of objects @xmath2 , and a preference profile @xmath3 that specifies for each agent @xmath4 his _ strict _ preferences over objects in @xmath5 . the goal is to find a desirable assignment keeping in view the preferences of the agents .
a _ random assignment _ which we will simply refer to as assignment assigns the probability of agents getting objects .
a random assignment can be represented as a bistochastic matrix in which each entry denotes the probability of an agent getting an object . since both
the probability of an agent getting some object and the probability that an object is allocated to some agent is one , each column and row of the random assignment matrix sums up to one .
a _ deterministic assignment _ is a random assignment in which the probability of an agent getting an object is either one or zero .
the advantage of using random assignments instead of deterministic assignments is that they can allow for better ex ante fairness .
it is well - known that any random assignment can be a result of a probability distribution over deterministic assignments @xcite .
a random assignment also a useful time - sharing interpretation whereby the probability of agent @xmath6 getting object @xmath7 is the fraction of time he will be matched to object @xmath7 ( see e.g. , * ? ? ?
* ; * ? ? ?
* ) . in this paper , we focus on efficiency of random assignments . a deterministic assignment @xmath8 is _ pareto optimal _ if there exists no other deterministic assignment @xmath9 such that each agent weakly prefers his object allocated in assignment @xmath9 and at least one agent strictly prefers his object allocated in assignment @xmath9 .
when the assignment is random , pareto optimality can be generalized to two well - studied efficiency concepts ex post efficiency and stochastic dominance ( sd ) efficiency .
a random assignment is _ ex post efficient _ if it can be represented as a convex combination of pareto optimal deterministic assignments .
a random assignment is sd - efficient is there exists no other random assignment which each agent weakly prefers and some agent strictly prefers with respect to the stochastic dominance relation .
ex post efficiency is a weaker requirement than _ stochastic dominance ( sd ) efficiency _ @xcite .
the main research problem in this paper is to understand _ the structure and complexity of efficient assignments in particular ex post efficiency assignments_. we not only consider ex post efficiency and sd - efficiency but also introduce an intermediate notion called _ robust ex post efficiency _ that is weaker than sd - efficiency and stronger than ex post efficiency .
we seek to understand the geometry of the ex post efficient polytope and where the robust ex post efficient and sd - efficient points lie within the ex post efficient polytope or the assignment polytope .
an efficiency concept is deemed _ combinatorial _ if the efficiency of an assignment solely depends on which entries of the assignment matrix are zero or non - zero .
we explore which of the efficiency concepts are combinatorial .
we also consider natural computational problems related to efficiency of random assignments .
previously , computational aspects of pareto optimal deterministic assignments have been studied in great depth in recent years @xcite .
similar analysis has been done for sd - efficient assignments where it has been shown that not only can an sd - efficient random assignment be computed efficiently @xcite , a linear programming formulation can be used to check whether an assignment is sd - efficient or not @xcite . however , to the best of our knowledge , the complexity of testing ex post efficiency has not been settled . testing
ex post efficiency is also closely related to implemeting a random assignment with respect to discrete pareto optimal assignments .
if one is able to compute an sd - efficient assignment @xcite , then the question arises that why should we bother with a less demanding notion of efficiency ?
there are a number of reasons why implementation of ex post assignments and testing ex post efficiency is important .
firstly , the algorithm to test sd - efficiency of a random assignment can not be used to test weaker notions of efficiency . secondly ,
in many scenarios , a random assignment may be given _ a priori _ because of various constraints and hence may not be sd - efficient . for example , the random assignment could be a result of an already decided time sharing agreement .
such a random assignment may need to be implemented in any case and would preferably be implemented via pareto optimal deterministic assignments . for example , agents may already have a time sharing assignment in place and one may want to know whether it can be achieved by randomizing over deterministic pareto optimal assignments .
thirdly , there may already be simple strategyproof method such as the uniform assignment rule in place where each agent gets @xmath10 of each object .
one may want to implement the uniform assignment via a convex combination of pareto optimal assignments even if it may not be sd - efficient .
we also note that sd - efficiency is incompatible with strategyproofness when also requiring anonymity @xcite .
finally , the convex hull of deterministic pareto optimal assignments is an interesting mathematical object and testing ex post efficiency of a random assignment is equivalent to checking whether a given assignment is in the convex hull .
the problem has important connections with optimizing linear functions over this convex hull .
[ [ contributions ] ] contributions + + + + + + + + + + + + + we first examine the problem of checking whether a given random assignment is ex post efficient and obtain insights into why the problem may be computationally challenging .
we show that whereas computing an ex post efficient assignment is easy , checking whether a given random assignment is ex post efficient is np - complete . hence implementing a given random assignment via deterministic pareto optimal assignments
is np - hard .
even if it is known that a random assignment is ex post efficient , finding its pareto optimal decomposition is np - hard .
our result also implies that optimizing over the convex hull of pareto optimal assignments is np - complete .
we formalize a new efficiency concept called _ robust ex post efficiency _ that is weaker than sd - efficiency but stronger than ex post efficiency .
a characterization of robust ex post efficiency is also presented .
previously , characterizing sd - efficiency has already attracted considerable interest ( see e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) .
we show that robust efficiency can be checked in polynomial time if there are a constant number of agent types .
we highlight that the well - known random serial dictatorship mechanism @xcite is not robust ex post efficient .
our finding strengthens the observation of @xcite that random serial dictatorship is not sd - efficient .
we show that whereas robust ex post efficiency is combinatorial , ex post efficiency is not .
the finding that ex post efficiency is not combinatorial also contrasts with the fact that in randomized voting , ex post efficiency of a lottery simply depends on its support .
table [ table : summary - expost ] summarizes some of the results . |
the study of proximity effect in superconductor ( s ) - ferromagnet ( f ) hybrid structures has its long history following the first theoretical proposal of the so - called @xmath0 state in a mesoscopic ring containing sfs josephson junction .
the penetration depth of cooper pairs into a diffusive normal metal ( n ) is characterized by the length scale @xmath1 , whereas in a ferromagnet this length scale is considerably shorter and is given by @xmath2 . here
@xmath3 is temperature , @xmath4 is the diffusion constant , and @xmath5 is the magnitude of the exchange potential in the ferromagnet . since the exchange field affects differently electrons with opposite spins , spin - singlet cooper pairs are fragile under the exchange potential .
in addition to a short penetration length , the pairing function of spin - singlet pair spatially oscillates with changing its sign under the exchange potential , @xcite which enables the @xmath0 states in sfs junctions .
@xcite although the @xmath0 state was predicted theoretically in 1970 s , it has been confirmed experimentally only recently .
@xcite the details of research progress have been summarized in several review papers .
@xcite bergeret , volkov and efetov @xcite proposed theoretically new type of proximity - induced superconducting state in ferromagnets , the so called long - range spin - triplet pairing state .
inhomogeneous magnetic structures near the sf interface ( see fig . 1 ) rotate the spin direction of an electron , which induces equal spin - triplet @xmath6-wave cooper pairs in ferromagnets .
@xcite such cooper pairs should have the odd - frequency symmetry to satisfy the requirement from the fermi - dirac statistics of electrons .
since equal spin - triplet pairs are not suppressed by the exchange potential , they have the long range penetration length into ferromagnets characterized by @xmath7 .
experimentally such effect has been observed first as the long - range josephson coupling in sfs junctions , @xcite where the ferromagnet is a half - metallic cro@xmath8 compound . in clean half - metallic sfs junctions
the equal - spin triplet pairs can have the odd - parity even - frequency symmetry , @xcite whereas in the experimentally relevant dirty limit case the equal - spin triplet even - parity @xmath6-wave odd - frequency symmetry is the only possible choice .
the experiment @xcite as well as anomalous conductance oscillations observed in sf hybrids @xcite stimulated much theoretical @xcite and experimental work @xcite . as a result
, a number of signatures of triplet correlations have been observed .
however , no experiment so far has reported direct and unequivocal evidences of long - range spin - triplet odd - frequency pairs . ) or noncollinear ( @xmath9 ) . ,
width=321 ] existing theories @xcite predict that the presence of odd - frequency pairs causes the enhancement of the zero - energy local density of states ( ldos ) .
@xcite according to a number of theoretical papers on the proximity effect in various sf hybrid structures , @xcite the relative fraction of odd - frequency pairs to even - frequency pairs depends sensitively on junction parameters such as the resistivity of f , the transparency of sf interface , the amplitude of the exchange energy @xmath5 , and the geometry of junctions . to observe clear evidence of spin - triplet odd - frequency pairs in experiments
, theoretical studies should show a way how to optimize the fraction of odd - frequency spin - triplet pairs in wide parameter range tunable in actual experiments . in this paper , by solving full - spin usadel equation @xcite in a wide parameter range ,
we systematically calculate the ldos at a surface of a diffusive ferromagnet connected with a metallic superconductor .
in particular , we focus on a relation between the magnitude of the zero - energy peak ( zep ) in the ldos and the fraction of triplet odd - frequency pairs and show the robustness of the presence of the zep . it should be noted that zep formation has been reported in measured tunneling conductance spectra in oxide - based sf hetero - structures with nonuniform ferromagnets .
@xcite however the physical origin of zep in these structures is still unclear .
therefore we propose an experimental method to explicitly detect a signature of spin - triplet paring by measuring zep .
this paper is organized as follows . in sec .
ii , we present a model of an sf junction and describe the numerical method to solve the nonlinear usadel equation in such system .
the numerical results of the ldos for various parameters and discussion of the robustness of zep are presented in sec .
iii . in sec .
iv , the summary of our results is presented . throughout the paper
we confine ourselves to the regime of zero temperature and put @xmath10 .
in this section , we will introduce a model of an sf junction and describe how to numerically solve the nonlinear usadel equation with inhomogeneous magnetization in a ferromagnet f. the system we consider is an one - dimensional diffusive sf junction where magnetization in the f layer is either homogeneous or rotating like illustrated in fig . 1 .
the rotation is introduced by dividing the f layer with thickness @xmath11 into f1 and f2 layers where magnetization are in general non - collinear .
the thicknesses of f1 and f2 layers , the sf barrier resistance , and the f layer resistance are respectively defined by @xmath12 , @xmath13 , @xmath14 , and @xmath15 .
the exchange field in the f2 layer is aligned along the @xmath16 axis , while that in the f1 layer is misaligned by angle @xmath17 in the @xmath18-@xmath16 plane .
importantly , the misorientation angle @xmath17 in exchange - spring ferromagnets @xcite can be experimentally controlled by applying an in - plane magnetic field . in our calculation
we have assumed that the exchange energy @xmath5 in the f1 and f2 layers are the same , and the f1 and f2 interface is transparent .
because of the spin - rotation of electrons in ferromagnets , we have to solve @xmath19 matrix usadel equation @xcite given by , @xmath20=0 , \label{usadel0}\end{aligned}\ ] ] where @xmath4 is the diffusion constant in f. here hamiltonian @xmath21 and the green function @xmath22 are respectively defined by @xmath23 , \\
\check{g}(x , e ) = & \left [ \begin{array}{cc } \hat{g}(x , e ) & \hat{f}(x , e ) \\ \undertilde{\hat{f}}(x ,
e ) & \undertilde{\hat{g}}(x , e ) \end{array } \right],\end{aligned}\ ] ] where @xmath24 here @xmath25 with @xmath26 are pauli matrices and @xmath27 is the @xmath28 unit matrix .
the magnetic moment @xmath29 in a ferromagnet is defined as @xmath30 throughout this paper , @xmath31 and @xmath32 indicate @xmath33 and @xmath28 matrices , respectively . in what follows ,
we only consider the @xmath6-wave spin - singlet pair potential in a superconductor , ( i.e. , @xmath34 ) .
the particle - hole symmetry results in @xmath35 to solve eq .
( [ usadel0 ] ) , we use the riccati parameterization @xcite for the green function , @xmath36 @xmath37 = \left [ \begin{array}{cc } \hat{n } & \hat{0 } \\ { \hat{0 } } & \undertilde{\hat{n } } \end{array}\right ] \left [ \begin{array}{cc } \hat{1 } - \hat{\gamma } \undertilde{\hat{\gamma } } & 2\hat{\gamma } \\ 2\undertilde{\hat{\gamma } } & - ( \hat{1 } - \undertilde{\hat{\gamma } } \hat{\gamma } ) \end{array}\right ] , \end{aligned}\ ] ] where @xmath38 here riccati parameters @xmath39 and @xmath40 are independent from each other .
the normalization condition of the green function is automatically satisfied under the parameterization , @xmath36 @xmath41 .
the derivative of the inverse matrix @xmath42 can be done as follows @xmath43 where @xmath44 which is obtained from an identity @xmath45 .
finally , the usadel equation ( [ usadel0 ] ) is reduced to two partial differential equations for @xmath46 and @xmath47 , @xmath48 - \hat{e } \hat{\gamma } + \hat{\gamma } \undertilde{\hat{e } } + \hat{\delta } - \hat{\gamma } \undertilde{\hat{\delta } } \hat{\gamma}=0 , \label{usadel1 } \\ & -i\hbar d\left [ \partial_x^2 \undertilde{\hat{\gamma } } -(\partial_x\undertilde{\hat{\gamma } } ) \ , { \hat{f}}\ , ( \partial_x\undertilde{\hat{\gamma } } ) \right ] - \undertilde{\hat{e } } \undertilde{\hat{\gamma } } + \undertilde{\hat{\gamma } } { \hat{e } } -\undertilde{\hat{\delta } } - \undertilde{\hat{\gamma } } { \hat{\delta } } \undertilde{\hat{\gamma}}=0 \label{usadel2 } .\end{aligned}\ ] ] after taking the complex conjugation and @xmath49 in the above equations , we find that @xmath50 and @xmath51 obeys the same equation .
thus we conclude that @xmath52 at the interface between a ferromagnet and a superconductor , the kupriyanov - lukichev boundary condition @xcite connects green functions on both sides , @xmath53 , \label{kl0}\end{aligned}\ ] ] where @xmath54 , \\ g_s = & \frac{e}{\sqrt{e^2-\delta^2}},\\ f_s = & \frac{i\delta}{\sqrt{e^2-\delta^2 } } .\end{aligned}\ ] ] here @xmath55 with @xmath56 , where @xmath57 ( @xmath58 is the superconducting transition temperature ) . in the calculation , we have assumed the retarded causality in the green function . by using eq .
( [ derivative1 ] ) , we obtain two boundary conditions @xmath59 since @xmath60 and @xmath61 , @xmath50 and @xmath51 satisfy the same boundary condition . by numerically solving nonlinear differential equations ( [ usadel1 ] ) and ( [ usadel2 ] ) together with the boundary conditions ( [ kl1 ] ) and ( [ kl2 ] )
, we will calculate local density of state ( ldos ) @xmath62 , \end{aligned}\ ] ] and the pair function matrix @xmath63 \vec{\sigma}_2 \nonumber \\ = & \left[\begin{array}{cc } i f_1(e ) + f_2(e ) & - i f_3(e ) - i f_0(e ) \\ - i f_3(e ) + i f_0(e ) & - i f_1(e ) + f_2(e ) \end{array}\right ] , \end{aligned}\ ] ] where @xmath64 is the normal dos and @xmath65 . here
the components @xmath66 , @xmath67 , @xmath68 , and @xmath69 correspond , respectively , to spin singlet [ @xmath70 , equal - spin triplets [ @xmath71 , [ @xmath72 ) ] , and hetero - spin triplet paring [ @xmath73 . since the diffusive limit is considered , the singlet component @xmath66 has an even - frequency symmetry while the triplet components @xmath74 ( @xmath75 ) have an odd - frequency symmetry .
it should be noted that the @xmath76 component of the magnetic moment @xmath29 is zero in this paper [ see eq .
( [ mag ] ) ] . as a consequence
the equal - spin triplet component @xmath68 is always zero .
in this section we study the ldos and the zero - energy peak ( zep ) in ldos for the cases of the uniform and nonuniform magnetization in a systematic manner . in order to show the robustness of the zep induced by spin - triplet odd - frequency paring ,
we calculate phase diagrams of the zep and the pair amplitudes as functions of several variables .
we also show how to detect signatures of the long - range triplet - paring experimentally .
let us first discuss the ldos for the uniform magnetization case , @xmath77 @xmath78 . in the calculation
, we set @xmath79 and @xmath80 ( a moderate proximity - effect regime ) . we show the f layer thickness @xmath11 dependence of the ldos @xmath81 at the edge of f2 layer , namely at @xmath82 in figs .
2(a)@xmath83(c ) .
when @xmath84 is very small or the thouless energy @xmath85 is much larger than @xmath86 , @xmath77 @xmath87 , the mini gap is formed due to the proximity effect . in this case
we can neglect magnetic effects and then the magnitude of the mini gap is approximately given by @xcite @xmath88 as in the case of diffusive s / n(normal metal ) junctions . by increasing @xmath11 ,
@xmath89 is decreased .
eventually , if a resonant condition @xmath90 is satisfied , the mini gap is closed completely [ see figs .
2(a ) and ( c ) ] .
@xcite as a consequence , the zero - energy peak ( zep ) is developed .
however , if we increase @xmath11 or decrease @xmath85 further , the ldos profile near @xmath91 becomes almost flat because the mini - gap edges move outwards toward the superconducting - gap edge .
therefore , the zep can be realized only near the resonant condition @xmath90 .
@xcite in order to see more clearly this point , we have plotted the zero - energy density of state @xmath92 and the amplitude of the pair functions @xmath93 ( @xmath94 ) as a function of @xmath11 in fig .
3(a ) . in the uniform - magnetization case ,
long - range triplet components @xmath67 and @xmath68 are completely absent .
@xcite moreover near the resonant condition @xmath90 , although the fraction of the short - range triplet component @xmath95 is larger than the singlet one @xmath96 , both components coexist with each other .
these observations are consistent with previous result .
@xcite on the other hand , in the case of a nonuniform magnetization ( @xmath9 ) , the behavior of @xmath92 is drastically different from the uniform case due to the appearance of long - range spin - triplet paring .
the ldos at the edge of @xmath97 layer as a function of @xmath11 is shown in figs .
2(d)@xmath83(f ) . as clearly seen from figs .
2(d ) and ( e ) , not only near the resonant condition but also a wide range of @xmath11 , the zep is developed . in order to understand the physical origin of the zep , we have plotted the zero - energy ldos @xmath92 and the pair functions @xmath98 at the edge of @xmath97 layer ( @xmath82 ) as a function of @xmath11 in fig .
if the long - range triplet component @xmath99 is largely developed in comparison with short - range components ( @xmath66 and @xmath69 ) , the zep is highly enhanced .
although short range components show an exponentially - damped oscillation as a function of @xmath11 , the long - range one decays considerably slowly with increasing @xmath11 .
this is due to the fact that long - range pairs ( @xmath100 ) have zero center - of - mass momentum as in the case of singlet - pairs in conventional sn junctions .
therefore in the case of the long @xmath97 layer _
i. e. _ , @xmath101 , one can have almost pure long - range triplet component near the edge of f@xmath102 layer .
it is important to note that the dependences of @xmath92 and @xmath99 on @xmath11 in the regime @xmath103 are closely related to each other as demonstrated in fig .
this can be explained as follows .
if the long - range component @xmath99 is fully developed , the zero - energy ldos is approximately given as @xmath104 \approx \sqrt{1 - \left| f_1(0 ) \right|^2 } , \ ] ] as can be obtained by using the normalization condition @xmath105 thus the @xmath11 dependence of zep is closely related to that of the long - range triplet component @xmath99 .
@xcite therefore we can conclude that a systematic zep measurement by changing @xmath11 gives a strong evidence for the signature of the long - range spin - triplet correlations . in sec .
c we will discuss in more detail the ways to discriminate experimentally between the short- and long - range triplet components .
heretofore only limiting cases , e.g. , a very weak exchange field , @xcite a very strong exchange field ( like half metals ) , @xcite and a very thin f layer thickness ( @xmath106 ) , @xcite have been explored in previous theoretical studies on the zep in sf junctions . moreover in order to validate the linearization approximation for the usadel equation , only the weak proximity - effect regime ( equivalently @xmath107 ) has been explicitly investigated .
therefore the natural question to ask is _ how robust the presence of the zep induced by spin - triplet odd - frequency paring is in reality ? _ to solve this crucial issue
, we will calculate the zero - energy ldos @xmath92 and the pair amplitudes @xmath98 by systematically varying several parameters , @xmath108 @xmath109 ( 1 ) the exchange energy @xmath5 and ( 2 ) the barrier resistance @xmath14 which is a control parameter for the proximity effect @xmath110 the linearized approximation .
then we would like to show the robustness for the presence of zep induced by spin - triplet odd - frequency paring . in fig .
4 we show the phase diagram of @xmath111 and a most dominant component of pair amplitudes for an sf junction with the uniform [ figs .
4(a ) and ( b ) ] and the nonuniform magnetization - configuration [ figs .
4(c ) and ( d ) ] as a function of @xmath112 and @xmath113 . in the calculation
, we have assumed that @xmath114 and @xmath115 . in the zero - energy ldos phase diagram ,
following three phases have been realized , @xmath116 , ( i ) the mini - gap phase with @xmath117 , ( ii ) the zep phase with @xmath118 , and ( iii ) the flat phase with @xmath119 . in the calculation
we have defined the zep phase as regions with @xmath120 and the flat phase with @xmath121 as a matter of practical convenience . as was already discussed in sec .
a , in the case of the uniform magnetization , only the short - range components @xmath66 and @xmath69 exist . as a consequence , only in the vicinity of the resonant condition the zep phase
is developed [ see the dotted line in figs .
4(a ) and ( b ) ] .
in contrast , if we consider a nonuniform magnetization ( @xmath9 ) , the zep phase is appeared in a wide range of the parameter region for @xmath113 and @xmath112 as clearly shown in fig .
this can be attributed to the appearance of the long - range triplet component @xmath99 [ see fig .
4(d ) ] . from above results
, one can conclude that in the sf junction with nonuniform magnetizations , _ the existence of the zep induced by spin - triplet paring is very robust irrespective of the device and material parameters _ as long as @xmath122 .
therefore the experimental observation of the zep gives a unequivocal evidence of the spin - triplet paring .
this is one of the important findings in our work . from the normal value @xmath64 , _
i. e. , _ @xmath123 at as a function of the f layer thickness @xmath11 for an sf junction with @xmath78 ( the black dashed line ) and @xmath9 ( the red solid and blue dotted line ) for different values of @xmath124 .
the ldos is evaluated at @xmath82 .
parameters are @xmath114 and @xmath79 .
, width=321 ] in this section we will investigate the zep structure in more detail and propose an experimental method to explicitly detect a signature of long - range triplet paring through the zep measurement , so to say _ the zep spectroscopy_. moreover the possibility for observing _ the singlet to triplet crossover _ via the zep spectroscopy is discussed as well .
we have calculated the deviation of the zero energy ldos @xmath92 from the normal value @xmath125 , _ i. e. , _ @xmath126 at @xmath82 by changing @xmath11 . in the calculation
we have used @xmath114 and @xmath79 as in fig .
3 . in fig .
5 we show a log plot of @xmath127 as a function of @xmath11 in the case of a moderate proximity regime ( @xmath80 ) and a strong proximity regime ( @xmath128 ) .
it should be noted that in the strong proximity regime , the well - used linearized usadel approach is not justified at all . in the case of an uniform magnetization ( @xmath78 )
, @xmath127 shows the oscillatory damped behavior .
this means that @xmath129 changes its sign almost periodically with increasing @xmath11 .
such a behavior is consistent with previous theoretical predictions @xcite and also with experimental results .
@xcite from the normal value @xmath64 , _
i. e. , _ @xmath130 at @xmath82 as a function of the f layer thickness @xmath11 for an sf junction with a non - uniform magnetization ( @xmath9 ) in a strong proximity regime ( @xmath128 ) .
panel ( b ) shows the absolute value of pair functions @xmath98 at @xmath82 as a function of @xmath11 .
@xmath66 ( black dashed line ) , @xmath69 ( blue dotted line ) , and @xmath67 ( red solid line ) are the shot - range singlet , the short - range triplet , and the long - range triplet component , respectively .
parameters are @xmath114 and @xmath79 .
the gay , blue and red region are corresponding to singlet- , short - range triplet- , and long - range triplet - dominant phases .
the upper table shows paring symmetries for a most dominant component and the sign of @xmath131 . , width=321 ] in contrast , the behavior of @xmath127 for the case of inhomogeneous magnetization ( @xmath9 ) is drastically different from that for the uniform one . due to the development of the long - range triplet component @xmath99 , @xmath127 decay very slowly as a function of @xmath11 . and
apparently the sign of @xmath129 is always @xmath132 as long as @xmath133 [ see also fig .
3(b ) ] . in this case
, the typical decay length is given by @xmath134 $ ] , where @xmath135 is the imaginary part of the energy which accounts for inelastic process , for instance , the thermal excitation or the spin - orbit interaction . throughout this paper
we set @xmath136 .
from above results , we can conclude that by systematically measuring the zero energy ldos as a function of @xmath11 , we can explicitly distinguish between long - range ( @xmath67 ) and short - range parings ( @xmath66 and @xmath69 ) .
more interestingly , we have found a following peculiar feature in the strong proximity regime ( @xmath128 ) . as clearly seen from the blue dotted curve in fig . 5 , a @xmath137 from the short - range ( @xmath138 ) to the long - range behavior ( @xmath139 ) appears . in order to see more explicitly this crossover ,
we have also calculated pair functions as a function of @xmath11 in fig .
6 and found that dip positions of @xmath127 are almost identical with crossover points between singlet- and triplet - dominant phases .
it is important to note that such a singlet - to - triplet paring crossover can be regarded as an even - to - odd frequency paring one . a physical origin of above remarkable phenomena
can be explained as follows .
if all the components @xmath74(@xmath140 ) coexist with each other , @xmath129 can be expressed from ( [ normalization ] ) as @xcite @xmath141 by assuming @xmath142 , and using the facts that @xmath143 for the singlet component and @xmath144 for triplet components ( @xmath75 ) .
@xcite therefore , singlet and triplet components respectively give a negative and positive contribution to @xmath129 .
so if the dominant component is changed from singlet to triplet ( or from triplet to singlet ) by increasing @xmath11 , @xmath129 changes its sign .
this gives rise to a dip structure in the @xmath145 vs @xmath11 curve as shown in fig .
it is important to note that even in the weak or moderate proximity regime ( @xmath146 ) such a singlet - triplet crossover phenomena can be realized . as was demonstrated in fig .
3(b ) , the long - range component @xmath99 is largely enhanced in compared with strong proximity cases [ fig . 6(b ) ] .
therefore single crossover ( from a short - range singlet to short - range triplet ) is realized only at around the resonant condition .
similarly we can regard well known zero - energy ldos oscillation as a function @xmath11 in an @xmath147 sf junction as a multiple ( short - range ) singlet to ( short - range ) triplet crossover [ see figs .
3(a ) and 5 ] .
therefore we can explicitly conclude that _ we can identify crossover points of the paring symmetry from dip positions of @xmath127_. this remarkable feature has never been recognized in previous studies and is one of most important findings in our paper . from above results
it is clear that the systematic ldos measurement by changing the exchange field @xmath5 , the f layer thickness @xmath11 , and the barrier resistance @xmath14 gives a clear and unambiguous evidence of the novel long - range triplet odd - frequency paring .
dependence of @xmath148 and pair functions @xmath98 ( @xmath94 ) in the f layer , where @xmath149 is corresponding to the sign of @xmath129 .
@xmath66 ( black dashed line ) , @xmath69 ( blue dotted line ) , and @xmath67 ( red solid line ) are the shot - range singlet , the short - range triplet , and the long - range triplet component , respectively .
parameters are @xmath9 , @xmath114 , @xmath150 , @xmath151 , and @xmath79 . ,
width=321 ] finally we would like to propose a systematic method to detect the signature of odd - frequency paring and observe the singlet - to - triplet crossover experimentally .
7(a ) shows a scheme of an sf junction in contact with an stm tip to measure the differential conductance or the ldos in the f layer .
the spatial dependence of the ldos of f in the strong proximity regime is plotted in fig .
7(b ) . as clearly seen from fig .
7(b ) , the measurement of the position dependence of the zero - energy ldos would enable clear identification of long - range odd - frequency paring as well as the the singlet - to - triplet crossover .
it should be noted that the spatial dependence of the ldos of an inhomogeneous sf junctions for a @xmath152 proximity and small exchange field case has been investigated by cottet .
to conclude we have systematically investigated the superconducting proximity effect in sf junctions with an uniform and a nonuniform magnetization in terms of spin - triplet odd - frequency paring . by solving the nonlinear usadel equation fully numerically , we have calculated the ldos in a ferromagnet and found following remarkable results .
( 1)in contrast to the case of the uniform magnetization , @xcite the ldos in sf junctions with a nonuniform magnetization has a zero energy peak in a wide range of parameters , indicating _ the robust presence of the zep induced by the spin - triplet odd - frequency paring_. ( 2)the zep height is damped very slowly with increasing @xmath11 due to the development of long - range spin - triplet paring .
this behavior is very contrast to uniform magnetization cases in which the zero energy ldos shows exponentially damped - oscillation as a function of @xmath11 .
@xcite ( 3)the dip position of @xmath127 is corresponding to the crossover - point between singlet and triplet or even and odd paring .
this means that the zep spectroscopy gives us clear information on the paring - symmetry of cooper pairs . above remarkable results clearly
indicate that the experimental observation of the zep for sf junctions with a non - uniform magnetization provides a smoking gun for the existence of the novel spin - triplet odd - frequency paring . in this paper
, we have discussed proximity effect assuming spin - singlet @xmath6-wave superconductor as a bulk state of s. an extension to unconventional superconductors is possible based on more general boundary condition @xcite taking account of the andreev bound state .
proximity effect in spin - triplet @xmath153-wave superconductors is interesting @xcite since odd - frequency pairing is also induced from bulk superconductor without the exchange energy .
@xcite in addition , we have especially focused on the ldos .
it is interesting to discuss the anomalous meissner effect @xcite and the surface impedance @xcite due to the proximity effect by odd - frequency pairing .
we would like to thank t. akazaki , a. cottet , n. birge , m. blamire , s. jiang , s. kashiwaya , a. s. vasenko , and t. yokoyama for useful discussions and comments .
one of authors ( s. k. ) would like to thank theory group in institut laue - langevin for hospitality during the course of this work .
this work was supported by the topological quantum phenomena ( no.22103002 ) kakenhi on innovative areas , a grant - in - aid for scientific research ( no .
22710096 ) from mext of japan , and the jsps institutional program for young researcher overseas visits .
99 l. n. bulaevskii , v. v. kuzii , and a. a. sobyanin , jetp lett . *
25 * , 291 ( 1977 ) .
p. fulde and r. a. ferrell , phys .
rev . * 135 * , a550 ( 1964 ) .
a. i. larkin and y. n. ovchinnikov , sov .
jetp * 20 * , 762 ( 1965 ) .
a. i. buzdin , l. n. bulaevskii , and s. v. panyukov , jetp lett . * 35 * , 179 ( 1982 ) .
v. v. ryazanov , v. a. oboznov , a. yu . rusanov , a. v. veretennikov , a. a. golubov , and j. aarts , phys .
lett . * 86 * , 2427 ( 2001 ) .
t. kontos , m. aprili , j. lesueur , f. genet , b. stephanidis , and r. boursier , phys .
lett . * 89 * , 137007 ( 2002 ) .
j. w. a. robinson , s. piano , g. burnell , c. bell , and m. g. blamire , phys .
lett . * 97 * , 177003 ( 2006 ) .
a. a. golubov , m. yu . kupriyanov , and e. ilichev , rev .
phys . * 76 * , 411 ( 2004 ) .
a. i. buzdin , rev .
phys . * 77 * , 935 ( 2005 ) .
f. s. bergeret , a. f. volkov , and k. b. efetov , rev .
phys . * 77 * , 1321 ( 2005 ) . f. s. bergeret , a. f. volkov , and k. b. efetov , phys . rev
* 86 * , 4096 ( 2001 ) .
a. f. volkov , f. s. bergeret , and k. b. efetov , phys .
lett . * 90 * , 117006 ( 2003 ) .
m. eschrig , phys . today * 64 * , 43 ( 2011 ) .
a. kadigrobov , r. i. shekhter , and m. jonson , europhys . lett . * 54 * , 394 ( 2001 ) .
r. s. keizer , s. t. b. goennenwein , t. m. klapwijk , g. miao , g. xiao , a. gupta , nature * 439 * , 825 ( 2006 ) .
m. eschrig , j. kopu , j. c. cuevas , and g. schn , phys .
90 * , 137003 ( 2003 ) .
y. asano , y. tanaka , and a. a. golubov , phys .
lett . * 98 * , 107002 ( 2007 ) .
y. asano , y. sawa , y. tanaka , and a. a. golubov , phys .
b * 76 * ( 2007 ) 224525 . v. braude and y. v. nazarov , phys .
lett . * 98 * , 077003 ( 2007 ) .
m. eschrig and t. lfwander , nature phys . * 4 * , 138 ( 2008 ) .
j. w. a. robinson , j. d. s. witt , and m. g. blamire , science * 329 * , 59 ( 2010 ) .
t. s. khaire , m. a. khasawneh , w. p. pratt , jr . , and n. o. birge , phys .
lett . * 104 * , 137002 ( 2010 ) .
d. sprungmann , k. westerholt , h. zabel , m. weides , and h. kohlstedt , phys .
b * 82 * , 060505(r ) ( 2010 ) .
m. s. anwar , f. czeschka , m. hesselberth , m. porcu , j. aarts , phys .
b * 82 * , 100501(r ) ( 2010 ) . c. klose , t. s. khaire , y. wang , w. p. pratt , jr . , n. o. birge , b. j. mcmorran , t. p. ginley , j. a. borchers , b. j. kirby , b. b. maranville , and j. unguris , phys .
108 * , 127002 ( 2012 ) .
y. wang , w. p. pratt , jr . , and n. o. birge , phys .
b * 85 * , 214522 ( 2012 ) .
i. sosnin , h. cho , v. t. petrashov , and a. f. volkov , phys .
lett . * 96 * , 157002 ( 2006 ) . ya .
v. fominov , a. f. volkov , and k. b. efetov , phys .
b * 75 * , 104509 ( 2007 ) .
t. yokoyama , y. tanaka , and a. a. golubov , phys . rev .
b * 75 * , 094514 ( 2007 ) .
y. sawa , t. yokoyama , y. tanaka , and a. a. golubov , phys .
b * 75 * , 134508 ( 2007 ) .
k. halterman , p. h. barsic , and o. t. valls , phys .
* 99 * , 127002 ( 2007 ) .
m. houzet and a. i. buzdin , phys .
b * 76 * , 060504(r ) ( 2007 ) .
a. cottet , phys .
b * 76 * , 224505 ( 2007 ) .
j. linder , t. yokoyama , and a. sudb , phys .
b * 77 * , 174507 ( 2008 ) .
j. linder , t. yokoyama , y. tanaka , y. asano , and a. sudb , phys .
b * 77 * , 174505 ( 2008 ) . j. linder , t. yokoyama , a. sudb , and m. eschrig , phys . rev
. lett . * 102 * , 107008 ( 2009 ) .
m. a. silaev , phys .
b * 79 * , 184505 ( 2009 ) .
m. eschrig , phys .
b * 80 * , 134511 ( 2009 ) .
j. linder , m. cuoco , a. sudb , phys .
b * 81 * , 174526 ( 2010 ) .
y. v. fominov , a. a. golubov , t. y. karminskaya , m. y. kupriyanov , r. g. deminov , and l. r. tagirov , jetp lett .
* 91 * , 308 ( 2010 ) .
l. trifunovic and z. radovi , phys .
b * 82 * , 020505(r ) ( 2010 ) . i. v. bobkova , and a. m. bobkov , phys . rev .
b * 82 * , 024515 ( 2010 ) .
m. s. kalenkov , a. d. zaikin , and v. t. petrashov , phys .
* 107 * , 087003 ( 2011 ) .
a. cottet , phys .
107 * , 177001 ( 2011 ) .
g. annunziata , m. cuoco , c. noce , a. sudb , and j. linder , phys .
b * 83 * , 060508(r ) ( 2011 ) . a. i. buzdin , a. s. melnikov , and n. g. pugach , phys . rev .
144515 ( 2011 ) . l. trifunovic , z. popovi , and z. radovi , phys .
b * 84 * , 064511 ( 2011 )
. t. y. karminskaya , a. a. golubov , and m. y. kupriyanov , phys .
b * 84 * , 064531 ( 2011 )
. m. knezevi , l. trifunovic , and z. radovi , phys .
b * 85 * , 094517 ( 2012 ) .
a. ozaeta , a. s. vasenko , f. w. j. hekking , and f. s. bergeret , phys .
b * 85 * , 174518 ( 2012 ) . c. t. wu , o. t. valls , and k. halterman , phys . rev .
b * 86 * , 014523 ( 2012 ) .
a. s. melnikov , a. v. samokhvalov , s. m. kuznetsova , and a. i. buzdin , phys .
109 * , 237006 ( 2012 ) .
a. s. vasenko , a. ozaeta , s. kawabata , f. w. j. hekking , and f. s. bergeret , to appear in j. supercond .
( 2013 ) doi:10.1007/s10948 - 012 - 2044 - 9 .
k. usadel , phys .
lett . * 25 * , 507 ( 1970 ) .
w. belzig , f. k. wilhelm , c. bruder , g. schn , and a. d. zaikin , superlattices and microstructures * 25 * , 1251 ( 1999 ) .
k. dybko , k. werner - malento , p. aleshkevych , m. wojcik , m. sawicki , and p. przyslupski , phys .
b * 80 * , 144504 ( 2009 ) . y. kalcheim , t. kirzhner , g. koren , and o. millo , phys .
b * 83 * , 064510 ( 2011 ) .
i. fridman , l. gunawan , g. a. botton , and j. y. t. wei , phys .
b * 84 * , 104522 ( 2011 ) . c. visani , z. sefrioui , j. tornos , c. leon , j. briatico , m. bibes , a. barthlmy , j. santamara , and j. e. villegas , nature phys . * 8 * , 539 ( 2012 ) .
y. kalcheim , o. millo , m. egilmez , j. w. a. robinson , and m. g. blamire , phys .
b * 85 * , 104504 ( 2012 ) . e. f. kneller and r. hawig , ieee trans . mag .
* 27 * , 3588 ( 1991 ) .
e. e. fullerton , j. s. jiang , m. grimsditch , c. h. sowers , and s. d. bader , phys .
b * 58 * , 12193 ( 1998 ) .
j. y. gu , j. kusnadi , and c. y. you , phys .
b * 81 * , 214435 ( 2010 ) .
n. schopohl and k. maki , phys . rev.b
* 52 * , 490 ( 1995 ) .
n. schopohl , _ quasiclassical methods in superconductivity and superfluidity _ , ( springer , heidelberg , 1998 ) pp
m. eschrig , phys . rev .
b * 61 * , 9061 ( 2000 ) .
j. linder , a. sudb , t. yokoyama , r. grein , and m. eschrig , phys .
b * 81 * , 214504 ( 2010 ) . m. y. kupriyanov and v. f. lukichev , sov .
jetp * 67 * , 1163 ( 1988 ) .
a. a. golubov , m. y. kupriyanov , y. v. fominov , jetp lett . * 75 * , 190 ( 2002 ) .
t. yokoyama , y. tanaka , and a. a. golubov , phys .
b * 72 * , 052512 ( 2005 ) .
t. yokoyama , y. tanaka , and a. a. golubov , phys .
b * 73 * , 094501 ( 2006 ) . a. i. buzdin , phys . rev .
b * 62 * , 11377 ( 2000 ) .
a. s. vasenko , a. a. golubov , m. y. kupriyanov , and m. weides , phys .
b * 77 * , 134507 ( 2008 ) .
a. s. vasenko , s. kawabata , a. a. golubov , m. y. kupriyanov , c. lacroix , f. s. bergeret , and f. w. j. hekking , phys .
b * 84 * , 024524 ( 2011 ) .
t. kontos , m. aprili , j. lesueur , and x. grison , phys .
lett . * 86 * , 304 ( 2001 ) .
k. m. boden , w. p. pratt , jr . , and n. o. birge , phys .
b * 84 * , 020510 ( 2011 ) .
y. tanaka , y. v. nazarov , and s. kashiwaya , phys .
90 * , 167003 ( 2003 ) .
y. tanaka , y. v. nazarov , a. a. golubov , and s. kashiwaya , phys .
b * 69 * , 144519 ( 2004 ) .
y. tanaka and s. kashiwaya , phys .
b * 70 * , 012507 ( 2004 ) .
y. tanaka , s. kashiwaya , and t. yokoyama , phys .
b * 71 * , 094513 ( 2005 ) .
y. asano , y. tanaka , and s. kashiwaya , phys .
* 96 * , 097007 ( 2006 ) .
m. r. brydon , b. kastening , d. k. morr , d. manske , phys .
b * 77 * , 104504 ( 2008 ) . p. m. r. brydon , phys .
b * 80 * , 224520 ( 2009 ) . y. tanaka and a. a. golubov , phys
lett . * 98 * , 037003 ( 2007 ) .
y. tanaka , y. asano , a. a. golubov and s. kashiwaya , phys .
b * 72 * , 140503 ( 2005 ) .
t. yokoyama , y. tanaka , and n. nagaosa , phys .
lett . * 106 * , 246601 ( 2011 ) .
y. asano , a. a. golubov , y. v. fominov , and y. tanaka , phys .
lett . * 107 * , 087001 ( 2011 ) . | we have investigated the proximity effect in superconductor / ferromagnet junctions in a systematic manner to discuss the relation between the zero energy peak ( zep ) of the local density of states ( ldos ) and spin - triplet odd - frequency paring . by exactly solving the nonlinear usadel equations , we have found that the zep is realized in a wide range of geometrical and material parameters in the case of the noncollinear magnetization .
this strongly suggests the robustness of the zep induced by spin - triplet odd - frequency paring in such systems .
we also found that the crossover from singlet to triplet paring can be detected by measuring the f layer thickness dependence of the zep height .
further , we show how to observe signatures of spin - triplet odd - frequency paring and the paring crossover by the ldos measurements .
our results provide a direct way to experimentally detect signatures of odd - frequency paring state . |
the experimental analyses of the differential elastic cross section usually rely on the following assumption : @xmath15 let us note that in the fitting procedure of @xmath9 , one must vary at least four parameters , @xmath0 , @xmath16 , @xmath17 and @xmath18 , the normalisation coefficient reflecting the systematic errors ( the fitted luminosity is @xmath18 times the true one ) .
( [ exp ] ) works reasonably well at low energies . at high energies ,
the elastic amplitude becomes a sum of many different diagrams which unitarize it and this simple behaviour changes . in some models @xcite , which describe the differential cross section in a wide region of @xmath19 and @xmath10
, the amplitude reaches the black disk limit ( bdl ) .
this leads to a significant change of the @xmath10 dependence of @xmath1 and @xmath12 , as seen in fig .
1.a and @xcite .
in this case , the standard fitting procedure by an exponential will give bad results . to see this , we can simulate data at @xmath20tev from the model of ref .
@xcite , and then analyse them using an exponential form for the scattering amplitude . in fig . 1.b
, we show the result : large differences in @xmath1 and @xmath18 can not be detected by a change in @xmath21 . as a consequence
, one can obtain very wide variations of @xmath6 depending on the input @xmath1 , as shown in fig .
but the value of @xmath1 at @xmath20tev strongly depends on the model , so it must be included in the fitting procedure . in the dubna dynamical model @xcite at @xmath22gev ( plain curve ) , @xmath23gev (
long dashes ) , @xmath24tev ( short dashes ) and @xmath20tev ( dash - dotted curve ) ; ( b ) the dependence on @xmath1 of @xmath25 and of the normalisation coefficient @xmath18 obtained in the fitting procedure of the simulated data at @xmath20tev.,title="fig : " ] in the dubna dynamical model @xcite at @xmath22gev ( plain curve ) , @xmath23gev ( long dashes ) , @xmath24tev ( short dashes ) and @xmath20tev ( dash - dotted curve ) ; ( b ) the dependence on @xmath1 of @xmath25 and of the normalisation coefficient @xmath18 obtained in the fitting procedure of the simulated data at @xmath20tev.,title="fig : " ] note also that the slope of the differential cross sections does not necessarily grow with @xmath19 @xcite .
some models predict a non - vanishing spin - flip amplitude at high energies .
such a contribution to the differential cross section will lead to a decrease of the slope as @xmath26 . indeed , let us assume that the slope of the spin - flip amplitude , @xmath27 , is @xmath28 larger than the slope @xmath29 of the spin - non - flip amplitude . in this case , we have for the total slope @xcite @xmath30 and for the cross section : @xmath31 obtained from the standard fitting procedure using an exponential form of the amplitude and a fixed @xmath1 at @xmath20tev with free normalization ( @xmath18 ) ; the diamond gives the input @xmath6 and @xmath1 ; ( b ) @xmath32 obtained from the simulated data at @xmath20tev analysed with an exponential form of the scattering amplitude(points with errors ) ; the curve gives the model prediction with @xmath33 .
, title="fig : " ] obtained from the standard fitting procedure using an exponential form of the amplitude and a fixed @xmath1 at @xmath20tev with free normalization ( @xmath18 ) ; the diamond gives the input @xmath6 and @xmath1 ; ( b ) @xmath32 obtained from the simulated data at @xmath20tev analysed with an exponential form of the scattering amplitude(points with errors ) ; the curve gives the model prediction with @xmath33 .
, title="fig : " ]
in order to measure reliably @xmath6 , we need to find additional ways to determine @xmath17 .
one possibility was given in @xcite , which uses a particular property of the proton - proton scattering amplitude at high energies and small @xmath10 : at a specific value @xmath34 , the real part of the coulombic amplitude cancels the real part of the hadronic amplitude @xmath35 .
so , the value @xmath36 will have a minimum at this special point , and the value of @xmath37 may be possible to extract from the experimental data .
note that this is a generic property that does not rely on a given model .
one has in general gev ; ( b ) the statistical sums ( upper and lower curves ) of the even and odd intervals coincides with possible oscillations ; the middle curves are the same sums but the beginning of the intervals was shifted by half an interval .
, title="fig : " ] gev ; ( b ) the statistical sums ( upper and lower curves ) of the even and odd intervals coincides with possible oscillations ; the middle curves are the same sums but the beginning of the intervals was shifted by half an interval .
, title="fig : " ] there can be some additional effects which change the form of @xmath9 .
for example , there can be some small oscillations in the scattering amplitude which are predicted by several models ( see @xcite for a review ) .
one such possible oscillation with period proportional to @xmath40 was analyzed in @xcite .
it was shown that such oscillations may exist in the experimental data obtained at @xmath41gev , as shown in fig
however in the standard fitting procedure , if we do not now the form and @xmath10-dependence of such oscillations , they are hidden in the statistical noise . in @xcite , a new method was proposed , based on statistically independent choices .
the experimental interval in @xmath42 is divided into small equal intervals @xmath43 .
one then calculates sums of the differences between the experimental data and the theoretical curve normalised by the experimental error , for @xmath44 odd or even : @xmath45 the corresponding curves are shown in fig .
the statistical analysis has shown the existence of oscillations with a period @xmath46 at a significance level of 3 @xmath47 .
we can conclude that some additional research is needed .
it is likely that the bdl regime will be reached at lhc energies .
its effects will have to be taken into account and one will have to fit simultaneously all 4 parameters @xmath18 , @xmath6 , @xmath16 and @xmath17 . to investigate the non - linear behaviour of the parameters of the scattering amplitude , one will probably need to develop new methods . +
* acknowledgments : * o.v .
selyugin is grateful to the organizing committee of the conference for the financial support and the warm hospitality , and also wishes to acknowledge support from fnrs and from the universit de lige where part of this work was done .
99 r. fiore et al . , arxiv:0810.2902 [ hep - ph ] .
j. r. cudell and o. v. selyugin , czech .
j. phys .
* 54 * ( 2004 ) a441 [ arxiv : hep - ph/0309194 ] .
j. r. cudell and o. v. selyugin , phys .
b * 662 * ( 2008 ) 417 [ arxiv : hep - ph/0612046 ] .
s. v. goloskokov , s. p. kuleshov and o. v. selyugin , z. phys .
c * 50 * ( 1991 ) 455 ; sov . j. part .
nucl . * 18 * ( 1987 ) 14 .
o. v. selyugin , phys .
b * 333 * ( 1994 ) 245 [ arxiv : hep - ph/9312305 ] .
o. v. selyugin , mod .
phys . lett .
a * 14 * ( 1999 ) 223 .
p. gauron , b. nicolescu and o. v. selyugin , phys .
b * 629 * ( 2005 ) 83 [ arxiv : hep - ph/0406177 ] .
o. v. selyugin , in the proceedings of the international workshop on high - energy physics : diffraction 95 , novy svet , ukraine , 6 - 11 sep 1995 p. 65 . ; ukr .
j. of phys .
* 3 * ( 1996 ) 296 ; p. gauron , b. nicolescu and o. v. selyugin , phys .
b * 397 * ( 1997 ) 305 . | the problems linked with the extraction of the basic parameters of the hadron elastic scattering amplitude at the lhc are explored .
it is shown that one should take into account the saturation regime which will lead to new effects at the lhc .
address= bogoliubov laboratory of theoretical physics , jinr , 141980 , dubna , russia address = ifpa , dept .
ago , bt .
b5a , universit de lige , b4000 lige , belgium in the nearest future , new experiments will measure the elastic scattering amplitude at the lhc .
we hope that they will improve the theoretical understanding of this basic object of hadronic physics .
now models predict a wide range of possibilities , for example for the growth of @xmath0 .
however , we had such a situation a long time ago . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ with the construction of large accelerators , it is hoped that the mysteries of high - energy scattering will unfold in the near future . "
+ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ long after , we met the problems linked with different results for the size of @xmath1 , the ratio of the real part to the imaginary part of the scattering amplitude [ ua4 measured @xmath2 and ua4/2 @xmath3 and then , at the tevatron , @xmath0 is ill - determined [ @xmath4 mb , the results of e811 experiment or @xmath5 mb according to cdf . ] from the experimental data through the luminosity - independent method . ]
the theoretical predictions for @xmath6 at the lhc cover a wide range , from @xmath7 mb up to @xmath8 mb . for elastic scattering ,
several models predict very different behaviours , such as , for example , oscillations , or a non - exponential dependence of @xmath9 on @xmath10 @xcite .
these uncertainties are usually connected with the interaction at large distances .
furthermore , it was shown @xcite that , at small impact parameter @xmath11 , the overlapping function grows and may reach a saturation regime at the lhc , which leads to essential changes in the @xmath10-dependence of @xmath1 and of the slope @xmath12 .
hence , the measurements of forward elastic scattering will give us information on hadron interactions at small @xmath11 and large distances . |
the _ spitzer space telescope _ @xcite , with its superb sensitivity and resolution at infrared ( ir ) wavelengths , provides an excellent opportunity to study planetary debris disks around stars @xcite .
for instance , a comprehensive _ spitzer _ 24 survey of main - sequence a - type stars has shown that up to @xmath150% of young ( @xmath230 myr ) stars have little or no 24 excess emission from debris disks , large debris - disk excesses decrease significantly at ages of @xmath1150 myr , and much of the dust detected may be generated episodically by collisions of large planetesimals @xcite .
the dust in these debris disks would have dissipated long before the stars evolve off the main sequence .
dust can be replenished during late evolutionary stages . as a low- or intermediate - mass star loses a significant fraction of its initial mass to become a white dwarf ( wd ) , its planetary system expands .
sub - planetary objects , such as asteroids and comets , can be injected to very small radii and be tidally pulverized by the wd , while orbital resonances with giant planets can raise the collision rates among sub - planetary objects and generate dust @xcite .
this freshly produced dust can be detected through ir excesses and allows us to peer into the late evolution of planetary systems .
the first two wds reported to possess dust disks were g29 - 38 and gd362 , both exhibiting near - ir excesses that were confirmed spectroscopically to be dust continuum emission @xcite .
a subsequent _ spitzer _ survey of 124 wds at 4.5 and 8.0 @xmath3 m found one additional dust disk , around wd2115@xmath4560 @xcite .
as g29 - 38 , gd362 , and wd2115@xmath4560 are wds with photospheric absorption lines of heavy elements , searches for dust disks have been conducted for daz and dbz wds , and indeed many more dust disks were discovered .
for example , a near - ir spectroscopic survey of 20 daz wds found a dust disk around gd56 @xcite , the presence of a dust disk around the daz wd1150@xmath4153 was diagnosed by @xmath5-band excess and confirmed spectroscopically @xcite , and _ spitzer _
photometric observations of 9 daz / dbz wds revealed dust disks around gd40 , gd133 , and pg1015 + 161 @xcite
. one common characteristic of these dust disks around daz / dbz wds is that they are all small , with outer radii @xmath60.01 au . as the dust disks are completely within the roche limits of the wds and the dust mass
is estimated to be only @xmath110@xmath7 g , it is suggested that tidally disrupted asteroids produce the dust disks and that the accretion of this dust enriches refractory metals , such as ca , mg , fe , and ti , in the wd atmospheres @xcite . to date , @xmath120 dust disks around daz / dbz wds have been reported , all consistent with this suggested origin of tidal disruption of asteroids @xcite .
more dust disks of this type are being found from the _ wide - field infrared survey explorer _ observations of wds ( e.g. , * ? ? ?
an entirely different kind of dust disk has been discovered around the central wd of the helix planetary nebula ( pn ) , wd2226@xmath4210 @xcite .
_ spitzer _ observations of the helix nebula show a bright compact source coincident with the central wd in the 24 and 70 bands , and follow - up infrared spectrograph ( irs ; * ? ? ?
* ) observations have verified that the mid - ir emission originates from a dust continuum .
the spectral energy distribution ( sed ) of this ir - emitter indicates a blackbody temperature of 90 - 130 k , and its luminosity , 5 - 11@xmath810@xmath9 ergs s@xmath10 , requires an emitting area of 3.8 - 38 au@xmath11 .
these properties can only be explained by the presence of a dust cloud ; furthermore , little extinction exists toward the wd , so the dust cloud must be flattened with a disk geometry . adopting a stellar effective temperature of 110,000 k for wd2226@xmath4210 @xcite and assuming astronomical silicates with a power - law size distribution and a maximum grain radius of 1000 for the dust grains , models of the sed
indicate that the dust disk extends between 35 and 150 au from the wd and has a mass of @xmath10.13 @xmath12 .
since this dust must have been generated recently and since the radial location of the dust disk corresponds to that of the kuiper belt in the solar system , @xcite suggest that the dust disk around wd2226@xmath4210 was produced by collisions of kuiper belt - like objects ( kbos ) or the break - up of comets . to simulate the dust disk of wd2226@xmath4210 ,
the dynamic evolution of a debris disk around a 3 @xmath13 star has been modeled from the main sequence ( corresponding to @xmath1a0v ) to the wd stage , and it is found that collisions among kbos may produce the amount of dust observed @xcite .
if the dusk disk around the helix central star is indeed produced by collisions of kbos , similar dust disks should be found around other wds and a survey would allow us to assess their frequency of occurrence and physical properties .
these results can then be compared with models of debris disks evolution @xcite for implications on their planetary systems . to search for dust disks similar to that around the central wd of the helix nebula
, we have conducted three surveys using _ spitzer _ observations : ( 1 ) 24 survey of hot wds and pre - wds ( this paper ) , ( 2 ) archival survey of ir excesses of wds @xcite , and ( 3 ) archival survey of ir excesses of central stars of pne ( cspns ; * ? ? ?
the combined results from these three surveys will provide a comprehensive picture of post - main sequence dust production and dynamic evolution of debris disks .
this paper reports the results of the _ spitzer _ 24 survey of hot wds as well as follow - up spectroscopic and imaging observations for a subset of hot wds with 24 excesses .
section 2 describes the target selection and observations , section 3 reports the results , and section 4 discusses the implications .
a summary is given in section 5 .
the large dust disk around the central wd of the helix nebula is detectable at 24 because wd2226@xmath4210 has a high temperature and thus high luminosity .
stellar luminosity is a steeper function of temperature than of radius , and wds span a small range of radii ; furthermore , stellar effective temperatures are more readily available in the literature than distances and luminosities .
therefore , we use stellar effective temperature as our main target selection criterion and selected targets from two sources for our 24 survey .
first , we use the web version of the @xcite wd catalog to search for wds whose spectral classifications indicate temperatures near or higher than 100,000 k ; 58 such hot wds are found .
hot wds are the youngest wds and often surrounded by evolved pne ; 22 of the 58 wds selected are in known pne . to probe
whether dust disks are present around pre - wds , we have selected 14 hot cspns @xcite whose effective temperatures are @xmath14100,000 k but whose surface gravity is not yet high enough to be classified as wds . although pre - wds evolve along tracks of almost constant luminosities in the hr diagram , we do not select cspns with lower effective temperatures as they are usually surrounded by smaller and brighter pne , making it difficult to obtain accurate photometric measurements of the cspns . these 58 wds and 14 pre - wds form the target list of our _ spitzer _ 24 survey for dust disks . among the 72 targets ,
`` wd 0123@xmath4842 '' was a misnomer from wrong coordinates , and has been removed from the web version of the mccook & sion wd catalog ( e. sion , personal communication ) ; thus , this spurious object will not be discussed further in this paper .
the 71 valid targets are listed in table 1 .
the coordinates in columns 2 - 3 are measured from the digitized sky survey 2 ( dss2 ) , and in many cases represent an improved accuracy when compared with those given by mccook & sion or simbad . unless noted , the effective temperatures ( @xmath15 ) in column 4 are taken from @xcite or converted from the spectral classification in column 5 taken from mccook & sion s wd catalog . the common name and png number of associated pne
are given in columns 6 - 7 .
the _ spitzer _ observations reported in this paper were obtained from three programs .
the main program is our 24 survey of the hot wds or pre - wds ( program 40953 ) using the multiband imaging photometer for _ spitzer _ ( mips ; * ? ? ?
* ) . to assess the nature of hot wds 24 excesses , we have further obtained
_ spitzer _
irs , infrared array camera ( irac ; * ? ? ?
* ) , and mips 70 observations ( program 50629 ) . as the mips
24 survey was not completed before the proposal deadline for _ spitzer _ cycle 5 , the last cryogenic cycle of the _ spitzer _ mission , irs and mips 70 observations were obtained for only a subset of hot wds with 24 excesses : cspn k1 - 22 , wd0103 + 732 ( cspn egb1 ) , wd0127 + 581 ( cspn sh2 - 188 ) , and wd0439 + 466 ( cspn sh2 - 216 ) . among these four targets ,
archival irac observations are available for wd0439 + 466 ; we have thus obtained new irac observations for only the other three .
finally , we have obtained a mips 70 observation of kpd0005 + 5106 (= wd0005 + 511 ) through a _ chandra x - ray observatory _ observing program ( go8 - 9026 ) .
this mips observation was merged with our _ spitzer _ program 40953 for convenience . in the mips
24 survey of hot wds , each of the 71 objects in table 1 was imaged using the small - field photometry mode which obtained a sequence of 14 dithered exposures in a preset pattern ( see the _
spitzer space telescope _
observer s manual for more details ) .
the observations used an exposure time of 10 s and cycled through the pattern 3 times , yielding a total exposure time of 420 s in the central 32@xmath832 region . in a typical median background condition ,
the depth of the survey will reach a 1-@xmath16 point - source sensitivity of 33.6 @xmath3jy .
the raw data were processed using the data analysis tool ( dat ; * ? ? ?
* ) for basic reduction ( e.g. , dark subtraction and flat fielding ) .
all individual exposures were first corrected with a scan - mirror - dependent flat to correct for the dark spots due to particulate matter on the pick - off mirror . for targets that have no large - scale extended emission ( from either the surrounding pne or the background galactic cirrus ) , a second flat field generated from the data itself was applied to correct the possible dark latency and scattered light gradient in order to enhance photometric sensitivity .
details of these processing procedures can be found in @xcite .
the final mosaics were then constructed with a pixel size of 1245 ( half the size of the physical pixel ) for photometry measurements . the calibration factor
4.54@xmath810@xmath17 mjy sr@xmath10 ( dn s@xmath10)@xmath10 was used to convert the instrumental unit to physical units @xcite .
all final mosaics were astrometrically calibrated by comparisons with the two micron all sky survey ( 2mass ; * ? ? ?
* ) sources in the field to establish the world coordinate system ( wcs ) , the accuracy of which is generally better than @xmath11 .
we first performed source extraction using _ starfinder _ @xcite with a smoothed theoretical point spread function ( psf ) generated by the stinytim program ( krist 2006 ) .
we consider the wd detected if the extracted source position is within 15 from the given wd position .
in other words , this method ensures a point - source - like object is required to be coincident with the given wd position . for sources that are not detected
, we estimate their `` observed '' point source flux by fixing the psf at the given wd position on the psf - subtracted ( source - free ) image and using the minimum @xmath18 technique . in some cases where the nebular emission is bright ,
this psf flux is totally dominated by the background ( e.g. , cspn ic 289 , cspn mewe 1 - 3 , and wd1958 + 015 in ngc6852 ) .
these psf fluxes ( non - color corrected ) are listed in column 3 of table 2 .
we have estimated the 1-@xmath16 point source sensitivity of the observation using the pixel - to - pixel variation inside a radius of 44 centered at the source position on the source - free image ( i.e. , the systematic background noise ) .
the final photometry error includes ( 1 ) the systematic error of the observation , estimated from pixel - to - pixel variations ; ( 2 ) 24 @xmath3 m detector repeatability , @xmath11% of the measured flux ; and ( 3 ) 24 @xmath3 m confusion noise , @xmath115 @xmath3jy , a median value estimated from the extragalactic source counts @xcite .
these error contributions are summed in quadrature and listed in column 4 of table 2 as `` uncertainty '' .
this `` uncertainty '' has no direct relationship to the detectability of the source .
it simply reflects the photometric accuracy if a point source was coincident with the given wd position .
for non - detections , we also list 3-@xmath16 upper limits in column 5 of table 2 , computed as the three times of the uncertainty plus the psf flux . in most cases , these 3-@xmath16 upper limits are totally dominant by the surrounding bright nebular emission as remarked in column 8 in table 2 . for detected sources ,
we have also carried out aperture photometry using a source aperture of 6225 radius , a sky annulus of radii 1992 2988 , and an aperture correction of 1.7 .
these results are compared with those determined from the psf - fitting method .
when the two measurements for a source are discrepant , we examine the field for nearby objects or background structures that might compromise the photometry , and adopt the measurement that is less compromised .
the aperture photometry result is adopted for wd0439 + 466 , and the psf - fitting results are adopted for the other eight cases of detections .
when irs observations are available , we use the background - subtracted flux densities near 24 to further constrain the mips 24 photometric measurements .
four targets were included in the mips 70 observations as part of our program 50629 .
the three original targets , cspn k1 - 22 , wd0103@xmath19732 , and wd0439@xmath19466 , were observed in the photometry raster map mode with 3@xmath81 maps in the array column and row direction and a step size of 1/8 array .
this yields a uniformly covered area within a diameter of 32 centered at the target .
each map was repeated 4 times ( cycles ) with 10 s integration per frame , resulting in a total effective integration of 960 s. wd0005 + 511 was a target from the _ chandra _ program go8 - 9026 .
its 70 observations were made in the default photometry mode , with 10 s integration per frame and three cycles , and the total integration time was 240 s. the basic reduction ( dark subtraction , illumination correction ) of the 70 data was processed using dat , similar to that of the 24 data .
the known transient behavior in the 70 array was removed by masking out the sources in the field of view and time filtering the data @xcite . for the objects in pne with extended nebulosity seen at 70 , the size of the masked area was adjusted to cover most of the nebulosity accordingly .
the final mosaics were combined using the wcs with a subpixel size of 493 .
figure 1 shows the 70 images alongside the 24 images of k1 - 22 , wd0103 + 732 , and wd0439 + 466 .
the 70 image of wd0103 + 732 shows bright extended emission with a central peak at the expected wd position .
the diffuse emission at 70 , appearing more extended than that in the 24 band , is most likely dominated by bright emission lines such as the [ ] 88 line . to minimize the nebular contamination , we performed aperture photometry with a very small ( 16 ) aperture and sky annulus of 1839 .
we have also used psf fitting to estimate the source brightness .
both methods give a point source flux of @xmath155 mjy .
the pixel - to - pixel variations in the data suggest 1-@xmath16 point source sensitivity of 8.4 mjy .
wd0439 + 466 is coincident with a very faint 70 source .
the source is superposed on diffuse emission , similar to the case of wd0103 + 732 .
using the same approach as for wd0103 + 732 , we estimate the source to have a 70 flux density of [email protected] mjy . unlike the above two wds , cspn k1 - 22
has no obvious source coincident with the wd .
the data were very noisy , probably due to many foreground and background sources ( evident at 24 as well ) .
the estimated 1-@xmath16 point source sensitivity is @xmath14 mjy .
wd0005 + 511 is not detected at 70 , either ; its 1-@xmath16 sensitivity is 5.4 mjy .
irac observations of three targets , cspn k1 - 22 , wd0103@xmath19732 , and wd0127@xmath19581 were obtained at 3.6 , 4.5 , 5.8 and 8.0 in our program 50629 . these observations all used a 30 s frame time and a cyclic dither pattern with medium offsets to obtain 5 frames for a total integration time of 150
s for each target .
the basic calibrated data ( bcd ) frames from these observations were reduced using standard routines within the mopex package .
we have measured flux densities in each of the irac bands using the iraf task to perform aperture photometry .
these measurements used a 36 radius source aperture with a background estimated from the surrounding annulus of radii 36 to 84 .
an aperture correction based on the results tabulated in the irac data handbook ( ver 3.0 ) was then applied to obtain the flux densities reported in table 3 .
archival irac observations are available for wd0439 + 466 ( program 30432 ; pi : burleigh ) , wd0726 + 133 ( program 30285 ; pi : fazio ) , and the cspn ngc2438 ( program 68 ; pi : fazio ) .
these archival data were downloaded and reduced in the same manner as our own observations .
their flux densities are also reported in table 3 .
we have also obtained follow - up irs observations for cspn k1 - 22 , wd0103 + 732 , wd0127 + 581 , and wd0439 + 466 in our program 50629 .
all sources were observed using the low - resolution modules sl1 ( 7.414.5 ) , sl2 ( 5.27.7 ) , ll1 ( 19.538.0 ) , and ll2 ( 14.021.3 ) .
see @xcite for a more detailed description of the irs and its capabilities .
cspn k1 - 22 and wd0103 + 732 were observed in the irs staring mode after a peakup on a nearby source .
wd0127 + 581 , the faintest target , was observed in mapping mode where the target was sequentially stepped along the slit to facilitate an improved background subtraction .
the sl1 and sl2 observations of wd0127 + 581 used 8 pointings spaced by 3 while the ll1 and ll2 observations used 12 pointings spaced by 6 .
the mips 24 observations of wd0439 + 466 showed diffuse emission around the wd , thus the irs observations of wd0439 + 466 were made in the mapping mode to obtain spectra of the wd and the diffuse emission by taking spectra from a series of dense slit positions centered on the wd and sequentially offset in the direction perpendicular to the slit . specifically , the sl1 and sl2 maps were comprised of 9 pointings spaced by 36 while the ll1 and ll2 maps are comprised of 5 pointings spaced by 106 .
we summarize the modules , exposure times , and setups used in table [ tab_irssetup ] .
all spectra were reduced using the cubism software @xcite with the latest pipeline processing of the data and the most recent calibration set ( irs_2009_05_20-pb - pfc - trim - omeg - lhllbiasfork.cal ) .
each low - resolution module contains two subslits that are exposed at the same time , e.g. , sl1 on - source and sl2 off - source , and vice versa .
we examine the off - source frames , select the ones free of any source contamination , trim the maxima and minima , and use the average to produce a background frame .
this 2d background frame is then subtracted from all on - source frames in the corresponding module to remove astrophysical background and to alleviate the bad pixels that contaminate the irs data .
after the 2d background subtraction , we flag the global and record - level bad pixels first using the default cubism parameters for automatic bad pixel detection , and then through manual inspection of each bcd record , as well as by backtracking the pixels contributing to a given cube pixel .
the spectra are extracted with aperture sizes large enough to enclose the 24 source .
see table 5 for the aperture sizes used for the spectral extractions . for the ll orders ,
two local background spectra on either side of the source are extracted and averaged for a 1d subtraction of the local background . for the sl orders ,
a single background spectrum is used for the local 1d background subtraction .
special notes for spectral reduction of individual objects are given below : + .
the frames used for the construction of ll1 background contain a very faint point source . as the wd is bright and the point source
is practically removed in the min / max trimming and averaging process , we choose to ignore this faint source so that we can perform the 2d background subtraction to maximize the quality of the final data cube .
in addition , the frames used for the construction of the ll2 background show a small knot of h@xmath21 line emission at 17 .
this h@xmath21 emission knot is near the center of one slit position , and at the edge of the other slit position .
we therefore use only the latter slit position for background subtraction .
the pixels of the on - source bcd frames affected by over - subtraction of the emission knot are flagged as record - level bad pixels , and not used for cube construction .
furthermore , this h@xmath21 knot is far enough from the wd position that both a target spectrum and a local background spectrum can be extracted outside the position of the h@xmath21 emission knot in the slit . the spectrum of wd0103 + 732 is not detected in the sl2 observations .
this wd is faint , with @xmath22 = 0.34 mjy .
the orientation of the irs slit , determined by the roll angle of the spacecraft , was such that a very bright neighboring star was included in the slit .
consequently , the weak emission of wd0127 + 581 was overwhelmed by the elevated background from the neighboring star .
the spectrum of the wd is not unambiguously detected in the background - subtracted frame ; the signal - to - noise ratio is too low for meaningful spectral extraction .
+ . in its sl2 order ,
the bcd frames from the three final mapping positions show an abrupt jump in brightness across the center of each frame . since these slit positions do not contain the target wd , we do not use them for the sl2 cube construction or the sl1 background construction .
in our mips 24 survey of hot wds , nine of the 71 objects show 24 point sources detected at the given wd positions . to assess the physical significance of these 24 detections and the 3-@xmath16 upper limits ( see table 2 ) ,
we have computed the expected photospheric emission at 24 by extrapolating from the optical and near - ir measurements of each wd . we first searched the web version of the mccook and sion wd catalog and the simbad database for available photometric measurements in the @xmath23 bands and the sloan digital sky survey ( sdss ) @xmath24 bands . for wds that were discovered from the sdss @xcite ,
usually only @xmath24 photometric measurements were available .
we also searched the 2mass point source catalog @xcite for near - ir counterparts to the wds .
the effective temperatures of our program wds are so high that it is reasonable to approximate their photospheric emission with a blackbody in the rayleigh - jeans limit , i.e. , the flux density @xmath25 @xmath26 @xmath27 .
the amounts of extinction to most of these wds are unknown , but will have smaller impact at longer wavelengths .
we thus use the photometric measurement with the longest wavelength and without extinction correction to extrapolate the expected flux density at 23.7 , the central wavelength of the mips 24 band . the fiducial photometric measurement and the expected photospheric emission at 24 are given in columns 6 and 7 of table 2 , respectively .
the expected photospheric emission from these hot wds are all less than the nominal sensitivity of the 24 observations , 0.034 mjy .
it is not surprising that all nine detections are at least two orders of magnitude higher than the expected photospheric emission , and thus represent ir excesses . in the cases of non - detections , the 3-@xmath16 upper limits are several to several tens times higher than the expected photospheric emission .
these non - detections do not exclude the existence of low - level ir excesses , but no useful deductions can be made from these results ; thus , the non - detections will not be discussed further in this paper .
we have combined all available photometric measurements of the nine wds and pre - wds with 24 excesses to produce the seds plotted in figure 2 .
below we describe individually the nine hot wds and pre - wds with 24 excesses .
detailed modeling of the seds and irs spectra will be reported by bilikova et al .
( 2011b , in preparation ) .
multi - wavelength optical and ir images of cspn k1 - 22 are displayed in figure 3 .
the dss2 red image shows a source at 2 north of the cspn , but no counterpart of this source is seen in any of the other optical or ir images .
it is not clear whether this red source is spurious or transient .
the _ hst _
images show two sources separated by 035 at the center of k1 - 22 ; the blue northeast component is the cspn and the red southwest component is a cool companion @xcite .
the ir images show a source coincident with this close pair of stars , but can not resolve them . in the sed plot in figure 2 ,
the flux densities of cspn k1 - 22 and its close companion are individually plotted in the _ hst _ f555w and f814w bands , while the flux densities in the 2mass @xmath28 and _ spitzer _ irac and mips bands are for the two stars combined .
an extinction of @xmath29 @xcite has been corrected from the observed flux densities .
the atmospheric emission of the wd has been modeled as a blackbody ; for the companion , we adopt the kurucz atmospheric model for a k2v star , the spectral type implied by the _ hst _ photometry .
these stellar emission models are plotted in thick solid curves in figure 2 .
the irac and mips flux densities are all higher than the expected atmospheric emission from these two stars .
the ir excesses in the near - ir and in the mid - ir can not be described by a single blackbody ; however , two blackbody emitters at temperatures of 700 and 150 k appear to match the ir excesses in the irac and mips 24 bands .
the combined sed of the wd , its companion , and the two blackbody emitters is plotted as a thin solid line in figure 2 . the irs spectra extracted from cspn k1 - 22 and adjacent background regions all show [ ] 15.55 @xmath3 m , [ ] 18.71 , [ ] 25.89 , and [ ] 33.48 line emission ( see figure 4 ) . the background - subtracted irs spectrum of cspn k1 - 22 shows a weak continuum component and residual line emission , especially prominent in the [ ] line . to determine the nature of the residual line emission , we compare the spatial distributions of the continuum and line emission .
figure 5 shows the surface brightness profiles along the slit for the 24 continuum , [ ] 25.89 line , and [ ] 18.71 line , as extracted from the irs data cube .
it is evident that the continuum originates from a point source , the cspn , while the line emission from the pn is extended .
the [ ] line has a lower excitation potential ( i.e. , ionization potential of s@xmath30 , 23.3 ev ) and shows an extended , nearly flat surface brightness profile ; while the [ ] 25.89 line has a higher excitation potential ( i.e. , ionization potential of o@xmath31 , 54.9 ev ) and shows a narrower , centrally peaked surface brightness profile .
these different spatial distributions are consistent with the expectation from the ionization stratification in a pn .
the mips 24 image of k1 - 22 ( figure 1 ) shows that the cspn is superposed on diffuse emission @xcite .
the irs spectra show that this diffuse emission is dominated by the [ ] 25.89 line .
the centrally peaked morphology of the diffuse [ ] emission makes it difficult to subtract the background accurately .
as the background spectrum is the average of those extracted from the two regions at @xmath2020@xmath32 from the central source , it can be seen from the [ ] surface brightness profile in figure 5 that the excess emission at the center is @xmath135% of the background emission .
the residual [ ] emission in the background - subtracted spectrum in figure 4 is indeed about 1/3 as strong as the emission from the background regions .
therefore , the apparent [ ] emission in the background - subtracted spectrum of cspn k1 - 22 is most likely a residual from imperfect background subtraction .
figure 4 shows that the mips 24 band photometric measurement ( @xmath1 1 mjy ) of the cspn , plotted as an open diamond , appears higher than the continuum flux density of the background - subtracted spectrum ( @xmath10.75 mjy ) .
this discrepancy is likely caused by the contamination of [ ] line in the mips 24 band .
the photometric measurements in the irac 5.8 and 8.0 bands , plotted as open diamonds in figure 4 , are also higher than the continuum flux densities ; however , these discrepancies are less than 25% and are reasonable for the s /
n of the spectrum at these wavelengths . the background - subtracted irs spectrum of cspn k1 - 22 shows continuum emission well above the expected photospheric emission of the cspn and its red companion ( @xmath10.011 mjy ) , and thus represents an ir excess due to dust continuum .
nevertheless , it is not known whether the excess ir emission is associated with the cspn or its red companion .
high - resolution images in the mid - ir wavelengths are needed to resolve cspn k1 - 22 and its companion .
cspn ngc2438 shows bright 24 emission more than four orders of magnitude higher than the expected photospheric emission .
the mips observation of ngc2438 was made after the _ spitzer _ cycle 5 proposal deadline ; thus ngc2438 was not an irs target in our program 50629 .
however , @xcite had found ir excesses of cspn ngc2438 through the analysis of archival irac observations ( see irac images in figure 6 ) , and included ngc2438 in another _ spitzer _ program to study ir excesses of cspns ( program 50793 ; pi : bilikova ) .
the irs observations used only the short wavelength modules and therefore only extend to @xmath115 .
the background - subtracted irs spectrum of the cspn ngc2438 exhibits continuum emission ( bilikova et al . 2011b , in preparation ) .
the sed of cspn ngc2438 ( figure 2 ) shows that the 24 excess is much greater than the ir excesses in the irac bands .
the ir excesses in the irac bands and the mips 24 band can not originate from a single - temperature emitter ; instead , they can be roughly described by two emitters at 1200 k and 150 k , respectively .
unfortunately , 2mass did not detect the cspn ngc2438 in the @xmath33 and @xmath5 bands to allow a more precise modeling of the sed to determine the temperature of the hotter emitter and whether it originates from a low - mass stellar companion or from a dust disk .
wd0103 + 732 is the central star of the pn egb1 .
_ hst _ images do not show any companion stars @xcite .
see figure 7 for optical , 2mass @xmath34 , irac , and mips 24 @xmath3 m images of wd0103 + 732 .
the sed of wd0103 + 732 ( figure 2 ) shows optical and near - ir flux densities following a blackbody curve closely , although the wd is not detected in the 2mass @xmath33 and @xmath5 bands , where 2-@xmath16 upper limits are plotted .
the flux density in the irac 8.0 band starts to rise above the photospheric emission level , and in the mips 24 band the flux density is more than three orders of magnitude higher than expected from photospheric emission . the mips
24 image ( figure 1 ) shows that the point source of wd0103 + 732 is superposed on diffuse emission @xcite .
the irs spectrum of the background ( see figure 8) shows that the diffuse emission is dominated by line emission with a weak but appreciable continuum at wavelengths @xmath0 20 , and that the major contributor to the mips 24 emission is the [ ] 25.89 line . the background - subtracted spectrum of wd0103 + 732 ( figure 8) is dominated by continuum emission .
the apparent deficit at the [ ] and [ ] lines and excess at the [ ] line are caused by imperfect background subtraction .
figure 9 shows the surface brightness profiles of the 24 @xmath3 m continuum , [ ] and [ ] lines extracted from the irs data cube .
the nebular emission near wd0103 + 732 also shows ionization / excitation stratification , as the [ ] emission profile is narrower and more strongly peaked toward the center than [ ] , resulting in an under - subtraction of background in the [ ] line and over - subtractions in the [ ] and [ ] lines .
interestingly , the background shows an emission feature at 17 , which could be associated with transitions of h@xmath21 or polycyclic aromatic hydrocarbons ( pah ) .
as the emission feature is narrow and the pah 11.2 feature is not present , we identify this 17 feature as the h@xmath21 ( 0 , 0 ) s(1 ) line emission . this h@xmath21 emission is associated with a much more extended background and does not vary significantly over the regions of spectral extractions , no residual h@xmath21 17 line emission is present in the background - subtracted spectrum of wd0103 + 732 . despite the morphological differences between the pne egb1 and the helix nebula , the sed and irs spectrum of wd0103 + 732 are very similar to those of wd2226@xmath4210 @xcite .
both show ir excess starting in the irac 8 @xmath3 m band and peaking near the mips 70 @xmath3 m band . because of the bright nebulosity and lower angular resolution of the mips 70 @xmath3 m camera , we consider the 70 flux density of wd0103 + 732 , 55 mjy , an upper limit .
as the effective temperature of wd0103 + 732 , @xmath1150,000 k , is higher than that of wd2226@xmath4210 , @xmath1110,000 k , the physical properties of their dust disks are different ( bilikova et al . 2011b , in preparation ) .
note that the light curve of wd0103 + 732 shows sinusoidal variations in the @xmath35 bands , but the nature of these variations is uncertain ( hillwig et al .
2011 , in preparation ) .
this wd is not surrounded by any known pn .
it is not in a crowded region .
the optical and 2mass sources are coincident within 1 ; the 24 source is faint , but also coincident with the optical source within 1 - 2 ( see figure 10 ) .
mips 24 images are usually infested with faint background galaxies .
we identified @xmath120 sources within a 3@xmath36 area centered on wd0109 + 111 .
if these sources are randomly distributed within this area , the probability to have one source landing within 2 from wd0109 + 111 is @xmath37 .
thus , it is unlikely that the 24 source is a chance superposition of a background source .
the sed of wd0109 + 111 ( see figure 2 ) shows that the optical and near - ir flux densities fall nicely along a blackbody curve , but the 24 emission is more than 100 times higher than the expected photospheric emission . wd0109 +
111 is the second faintest 24 source among the nine detections .
no irs spectra are available for wd0109 + 111 .
wd0127 + 581 is surrounded by the pn sh2 - 188 .
this wd is not detected in the 2mass @xmath28 bands .
we have thus used the kitt peak national observatory 2.1 m telescope with the flamingos detector and measured @xmath34 = [email protected] and @xmath38 = [email protected] .
we have also obtained irac observations of this object .
see figure 11 for these images of wd0127 + 581 . as wd0127
+ 581 is a faint source superposed on a bright background , the errors in photometric measurements are large .
the wd is easily visible in the 3.6 and 4.5 bands , but not convincingly detected in the 5.8 and 8.0 bands .
as explained in section 2.3 , the weak emission from wd0127 + 581 is superposed on an elevated background from a bright neighboring star in the irs observations .
the high noise level prohibits the detection of the spectra of wd0127 + 581 , and hence no spectra are extracted .
the sed of wd0127 + 581 ( figure 2 ) is complex .
the near - ir excess in the @xmath34 and @xmath5 bands is indicative of a late - type companion , while the ir excesses in the irac bands and mips 24 band require two emitters at different temperatures .
two blackbody emitters at temperatures of 900 k and 150 k are plotted in figure 2 to illustrate one possibility .
wd0439 + 466 is the central star of the pn sh2 - 216 at a distance of 129@xmath206 pc @xcite .
the wd is coincident with a point source surrounded by diffuse emission in the mips 24 image @xcite .
the sed of wd0439 + 466 follows the photospheric blackbody curve throughout the optical , 2mass , and irac bands , but shows a large deviation in the mips 24 band .
see figure 12 for optical and ir images of wd0439 + 466 .
the irs spectra of wd0439 + 466 and its background are shown in figure 13 .
the spectrum of the diffuse emission adjacent to wd0439 + 466 shows the [ ] 25.89 and the h@xmath21 ( 0 , 0 ) s(1 ) 17 lines , in addition to a weak continuum that is appreciable at wavelengths greater than @xmath125 .
the spectrum extracted at the position of wd0439 + 466 also shows the [ ] and h@xmath21 lines , but these emission features are effectively removed by the subtraction of the local nebular background .
the background - subtracted spectrum of wd0439 + 466 is totally dominated by continuum emission in the 1535 range ; furthermore , its flux densities agree well with the photometric measurements in the irac 5.8 and 8.0 and mips 24 bands .
the continuum - dominated nature of this spectrum is similar to that of the helix central star .
the detailed modeling of wd0439 + 466 s sed and spectrum will be presented by bilikova et al .
( 2011b , in preparation ) .
wd0726 + 133 is the central star of the pn abell 21 , also known as ym29 .
the wd appears as a point source superposed on diffuse emission in the mips 24 image @xcite .
see figure 14 for optical and ir images of wd0726 + 133 .
the flux densities of wd0726 + 133 in the optical , 2mass @xmath28 , and irac bands are all consistent with a blackbody approximation of the stellar photospheric emission .
the flux density in the mips 24 band is almost three orders of magnitude higher than the expected photospheric emission .
several 24 sources are detected near wd0726 + 133 , and their optical counterparts are resolved into spiral galaxies in archival _ hst _ f555w and f814w images , but wd0726 + 133 remains unresolved and shows no companions @xcite .
no irs spectra are available for this wd .
wd0950 + 139 is surrounded by the pn egb6 .
this wd previously gained attention because of its strong [ ] and [ ] emission lines @xcite and near - ir excesses @xcite .
this puzzle was partially solved by an [ ] image taken with the _ hst _ faint object camera , which revealed an unresolved source 018 from the wd @xcite .
apparently , wd0950 + 139 has a late - type companion and the line emission is associated with the companion .
see figure 15 for optical and ir images of wd0950 + 139 .
our mips 24 observation of wd0950 + 139 shows a bright unresolved source , and the 24 flux density is four orders of magnitude higher than the photospheric emission expected from wd0950 + 139 ( figure 2 ) .
the sed in the irac and mips 24 bands can not be accounted for by a single - temperature emitter ; instead , it may be described by two emitters at temperatures of 500 and 150 k. a _ spitzer _
irs spectrum of wd0950 + 139 was obtained through guaranteed time observations , and the spectrum confirmed that the emission in the mips 24 band is dominated by continuum ( su et al .
2011 , in preparation ) .
this wd was discovered in the sdss .
it is below the detection limit of 2mass .
figure 16 shows optical and ir images of wd1342 + 443 .
the sed of wd1342 + 443 in figure 2 shows the sdss photometric measurements falling along the blackbody model curve with the 24 flux density more than 400 times higher than the expected photospheric emission .
this wd has the faintest 24 flux density among the nine hot wds detected .
no irs spectra are available for wd1342 + 443 .
among our sample of 71 hot wds , nine show 24 excesses , corresponding to almost 13% . figure 17
shows the distribution of the sample in @xmath39 and @xmath34 .
the number of detections in each magnitude bin is too small to provide meaningful statistics .
if the sample is divided into a brighter group that has photometric measurements and a fainter group that has no photometric measurements , it can be seen that 24 excesses are detected in 15 - 16% of the brighter hot wds , and only @xmath18% among the fainter hot wds .
as all of the wds in our sample have high temperatures and a small range of radii , their brightnesses are indicative of distances , with the fainter ones being at larger distances .
the different percentages of brighter and fainter wds exhibiting 24 excesses most likely reflect the fact that the limiting sensitivity of our mips 24 survey precludes the detection of distant objects .
the true percentage of hot wds exhibiting 24 excesses is likely greater than 15 - 16% . for the nine hot wds with 24 excesses ,
their 24 flux densities are plotted against their @xmath34-band flux densities in figure 18 .
no correlations are seen in this plot .
this is expected , as the @xmath34-band flux density is a rough indicator of distance and the 24 excess should not be dependent on distance .
another reason for the lack of correlation is the diverse physical conditions of the 24 emitters , as discussed later in section 4.2 .
seven of the nine hot wds with 24 excesses are still surrounded by pne : cspn k1 - 22 , cspn ngc2438 , wd0103 + 732 in egb1 , wd0127 + 581 in sh2 - 188 , wd0439 + 466 in sh2 - 216 , wd0726 + 133 in abell21 ( ym29 ) , and wd0950 + 139 in egb6 ; while two are not in pne : wd0109 + 111 and wd1342 + 443 .
there is a striking difference in the frequency of occurrence of 24 excesses between the wds and pre - wds in pne and those without pne : 20% and 56% , respectively .
the two wds not in known pne are also the faintest in 24 among the nine .
wds without pne are more evolved than those that are still surrounded by pne
. the smaller 24 excesses of the wds without pne appear to indicate an evolutionary trend of diminishing excess ; however , this trend is not obvious among the 24 excesses of the wds with pne , if the nebular sizes @xcite are indicative of their evolutionary status . considering the diversity in the progenitors masses and evolutionary paths of these wds , the current sample of hot wds with 24 excesses is too small to allow us to distinguish between an evolutionary effect and other effects .
the ir excesses of the cspn helix are detected in the 8 , 24 , and 70 bands , but not at shorter wavelengths @xcite .
high spatial resolution _ hst _
observation has ruled out any resolved companion earlier than m5 @xcite .
furthermore , the photometric accuracies in the irac bands ( 1-@xmath16 of 20 , 24 , 26 and 17 @xmath3jy at the 3.6 , 4.5 , 5.4 , 8.0 @xmath3 m bands , respectively ) can further constrain the mass of a possible companion . at an age of 1 gyr ,
the 2mass and irac photometry can also rule out , at the 3-@xmath16 level , any companion with mass greater than 20 jupiter masses , i.e. , early t dwarfs with @xmath40650 k. three of the nine new detections of mid - ir excesses of hot wds show similar seds : wd0103 + 732 ( cspn egb1 ) shows excess emission at 8 and 24 , while wd0439 + 466 ( cspn sh2 - 216 ) and wd0726 + 133 ( cspn abell 21 ) show excess emission at only 24 .
their lack of near - ir excesses does not support the presence of late - type companions .
four of the hot wds with 24 excesses also exhibit excess emission in the irac bands : cspn k1 - 22 , cspn ngc2438 , wd0127 + 581 ( cspn sh2 - 188 ) , and wd0950 + 139 ( cspn egb 6 ) . irs observations indicate that this ir excess is continuum in nature , and crude fits to the seds suggest blackbody emitter temperatures of 5001200 k. these temperatures are in the ranges for brown dwarfs , but the emitting areas , @xmath41 to @xmath42 cm@xmath11 , are too large for brown dwarfs .
it is most likely that these ir excesses in the irac bands originate from hot dust emission .
the two cases with cooler emitter temperatures ( 500 - 700 k ) , cspn k1 - 22 and wd0950 + 139 , are known to have late - type companions @xcite .
the relationship between these companions and the hot dust components is uncertain .
future searches for companions of cspn ngc2438 and wd0127 + 581 may help us understand the roles played by stellar companions in producing the dust component responsible for the irac excesses . finally , two hot wds with 24 excesses , wd0109 + 111 and wd1342 + 443 , have no irac observations to determine whether excess emission is present in the irac bands ; their 2mass measurements do show any excess emission . in summary , among the ten hot wds and pre - wds that show excess emission at 24 , 40% also show near - ir excesses associated with an additional warmer emission component that might be related to the presence of a companion .
whether a 24 excess is accompanied by near - ir excess or not , the shape of the sed suggests that the 24 emission originates from a component distinct from the wd s photospheric emission or another near - ir emitter .
the 24 emission must originate from a source cooler than 300 k. assuming that this source is heated solely by stellar radiation , we can determine the covering factor of the emitter from the luminosity ratio @xmath44 , where @xmath45 is the luminosity of the excess emitter and @xmath46 is the luminosity of the wd .
we adopt the stellar effective temperature , assume a blackbody model , and use the distance and extinction corrected photometry to determine the stellar luminosity @xmath46 .
( useful physical parameters are listed in table 6 . )
the luminosity of the excess emitter is calculated by assuming a 150 k blackbody model normalized at the observed 24 flux density . in the case of
wd0103 + 732 and wd2226@xmath4210 , blackbody temperatures are determined from model fits to the 8 , 24 , and 70 flux densities .
for the four wds exhibiting excess emission in the irac bands , we add another blackbody component to fit these measurements .
these ir emitter models , illustrated in figure 2 , are used to calculate their approximate luminosities .
the @xmath44 ratios of the nine new cases and the helix cspn are listed in table 6 .
the covering factors range from @xmath47 to @xmath48 .
a perfect absorbing body heated by a 100,000 k wd to 150 k would be at a distance of @xmath110 au , and 100 k at 20 au . even at a distance of 10 au
, a covering factor of @xmath49 corresponds to @xmath50 au@xmath11 , or @xmath51 @xmath52 , too large to be a star or a planet .
the most likely origin of this mid - ir emitter is a dust disk , as proposed for the cspn helix nebula @xcite .
the presence of near - ir excesses indicates the existence of a binary companion or a dust disk at higher temperatures , or both .
detailed modeling and high - resolution images are needed to decipher the true nature of the ir excesses .
we defer the modeling of the seds and irs spectra of hot wds with 24 excesses to another paper ( bilikova et al .
2011b , in preparation ) . the connection among cspns , circumstellar dust disks , and binarity has been alluded to by various observations and theoretical models . for example , keplerian circumstellar dust disks have been observed to be associated with binary post - agb stars @xcite .
furthermore , the close binary stellar evolution of cspns has been suggested to play a very important role in the formation and shaping of pne @xcite , although such binary companions are difficult to detect .
our detection of a warmer emitter ( 5001200 k ) in addition to the colder dust component ( at 100200 k ) in four cspns , two of which have known companions , appear to further the connection .
however , the dust disks responsible for 24 excesses are much larger than the circumstellar disks ejected from common - envelope binaries ( a few au at most ; * ? ? ?
* ) , and have very different geometry from the keplerian circumstellar dust disks around binary post - agb stars , especially in the covering factor ( @xmath44 ) .
the keplerian circumstellar dust disks around binary post - agb stars typically have @xmath44 @xmath1 0.20.5 @xcite , several orders of magnitude higher than those responsible for 24 excesses of hot wds and pre - wds .
this large discrepancy in covering factors suggests that these dust disks have different origins , and the small covering factors are more consistent with debris disks observed in main sequence stars @xcite .
it is thus likely that the origin of the 24 excesses of hot wds and pre - wds is dynamically rejuvenated debris disks as suggested by @xcite .
the central wd of the helix nebula has been shown to exhibit excess emission in the _ spitzer _ mips 24 and 70 bands and it is suggested that this ir excess originates from a dust disk produced by collisions among kbos @xcite . inspired by the wd in the helix ,
we have conducted a mips 24 survey of 71 hot wds and pre - wds and found excess 24 emission for nine of them .
we have further obtained _ spitzer _
irac and irs follow - up observations for a subset of these wds with 24 excesses .
the detection of 24 excesses is limited by the sensitivity of the observations . among the hot wds with optical or near - ir photometric measurements ,
15 - 16% are detected at 24 with excesses .
the true occurrence rate of 24 excesses must be still higher .
thirty - five of the 71 wds and pre - wds in our survey sample are in known pne ; among these 20% show 24 excesses , while among the 36 wds without known pne only 56% show 24 excesses .
the 24 excesses are accompanied by different levels of excesses in the @xmath28 and irac bands .
the excess emission in the @xmath28 bands for cspn k1 - 22 and wd0950 + 139 ( cspn egb6 ) originates from their known low - mass companions .
excess emission in the irac bands are present in cspn k1 - 22 , cspn ngc2438 , wd0950 + 139 , and wd0127 + 581 ( cspn sh2 - 188 ) ; irs spectra show continuum emission clearly in the three former objects and faintly in wd0127 + 581 , indicating the presence of a dust component at temperatures of @xmath11000 k. only wd0103 + 732 shows 8 , 24 , and 70 excesses similar to those in the sed of the helix central wd . in two cases ,
wd0439 + 466 and wd0726 + 133 , no ir excess is present in the irac bands ; the irs spectrum of wd0439 + 466 shows that its 24 band emission is dominated by dust continuum .
the emitters responsible for the 24 excesses have large emitting surface areas that can be provided only by dust disks .
furthermore , the excess emission in these long wavelength bands indicates dust temperatures @xmath53300 k. the @xmath44 ratios of dust disks responsible for the 24 excesses of hot wds and pre - wds are in the range of @xmath47 to @xmath48 , similar to those observed in debris disks around main sequence stars , but several orders of magnitude lower that those of keplerian circumstellar dust disks around binary post - agb stars .
it is likely that these dust disks around hot wds and pre - wds are indeed rejuvenated debris disks as suggested by @xcite for the helix central star . in two cases , cspn k1 - 22 and wd0950 + 139 in egb6 , late - type companions have been resolved by _ hst _ images , but _
images could not resolve these cspns and their companions .
it is thus uncertain whether the dust disks are associated with the cspns or their companion .
high - resolution mid - ir images are needed to establish the associations unambiguously .
this research was supported by the nasa grants jpl 1319342 and 1343946 and sao go8 - 9026 .
m.a.g . acknowledges support from the spanish ministerio de ciencia e innovacin ( micinn ) through grant aya2008 - 01934 .
we thank george rieke for useful discussion , adam myers for advice on the sdss photometry , adeline caulet and ian mcnabb for assisting in the preliminary irs data reduction , and the anonymous referee and jay farihi for suggestions to improve the paper .
the digitized sky survey images used were produced at the space telescope science institute under u.s .
government grant nag w-2166 .
the images of these surveys are based on photographic data obtained using the uk schmidt telescope .
the plates were processed into the present compressed digital form with the permission of these institutions .
images from the nasa / esa hubble space telescope were obtained from the data archive at the space telescope science institute .
stsci is operated by the association of universities for research in astronomy , inc . under nasa contract
nas 5 - 26555 .
this research has made use of the simbad database , operated at cds , strasbourg , france .
this research has also made use of the nasa / ipac infrared science archive , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
llrllll wd 0005 + 511 & 00 08 18.11 & + 51 23 16.9 & 120 & doqz.4 & & + pg 0038 + 199 & 00 41 35.31 & + 20 09 17.5 & 115 & do & & + wd 0044@xmath4121 & 00 47 03.31 & @xmath411 52 18.9 & 150 & pg1159 & ngc 246 & [email protected] + wd
0103 + 732 & 01 07 07.74 & + 73 33 25.2 & 150 & da.34 & egb 1 & 124.0 + 10.7 + wd 0108 + 100 & 01 11 06.59 & + 10 21 38.2 & 84 & doz.6 & & + wd 0109 + 111 & 01 12 23.06 & + 11 23 36.1 & 110 & doz.46 & & + wd 0121@xmath4756 & 01 22 52.97 & @xmath475 21 13.8 & 180 & pg1159 & & + wd 0123@xmath4842 & 01 21 55.0 & @xmath484 01 23.0 & & pg1159 & & + wd 0127 + 581 & 01 30 33.22 & + 58 24 50.7 & 102 & dao.49 & sh 2 - 188 & [email protected] + wd 0130@xmath4196 & 01 32 39.31 & @xmath419 21 40.4 & 100 & pg1159.5 & & + wd 0237 + 241 & 02 40 28.47 & + 24 22 10.4 & 100 & da0.5 & & + ic 289 & 03 10 19.36 & + 61 19 00.5 & 100 & & ic 289 & 138.8 + 02.8 + wd 0316 + 002 & 03 18 58.28 & + 00 23 25.8 & 100 & da0.5 & & + wd 0322 + 452 & 03 27 15.42 & + 45 24 20.2 & 125 & dao.40 & hdw 3 & [email protected] + ngc 1360 & 03 33 14.63 & @xmath425 52 17.9 & 110 & & ngc 1360 & [email protected] + wd
0439 + 466 & 04 43 21.20 & + 46 42 06.4 & 83 & da.61 & sh 2 - 216 & 158.5 + 00.7 + wd 0444 + 049 & 04 47 04.51 & + 04 58 41.7 & 100 & dqzo.5 & & + wd 0500@xmath4156 & 05 03 07.51 & @xmath415 36 22.5 & 100 & dao.51 & abell 7 & [email protected] + k 1 - 27 & 05 57 02.37 & @xmath475 40 21.1 & 105 & & k 1 - 27 & [email protected] + wd 0556 + 106 & 05 59 24.87 & + 10 41 39.9 & 141 & da.36 & wede 1 & [email protected] + wd 0615 + 556 & 06 19 33.95 & + 55 36 43.7 & 94 & dao.54 & puwe 1 & 158.9 + 17.8 + wd 0615 + 655 & 06 20 30.11 & + 65 34 21.3 & 100 & da.51 & & + abell 15 & 06 27 02.04 & @xmath425 22 49.3 & 110 & & abell 15 & [email protected] + abell 20 & 07 22 57.64 & + 01 45 33.4 & 119 & & abell 20 & 214.9 + 07.8 + wd 0726 + 133 & 07 29 02.64 & + 13 14 49.7 & 130 & pg1159 & abell 21 & 205.1 + 14.2 + ngc 2438 & 07 41 50.50 & @xmath414 44 07.7 & 114 & & ngc 2438 & 231.8 + 04.1 + wd 0753 + 535 & 07 57 51.69 & + 53 25 17.5 & 125 & dqzo.4 & jner 1 & 164.8 + 31.1 + wd 0823 + 316 & 08 27 05.57 & + 31 30 08.2 & 100 & da0.5 & & + ngc 2610 & 08 33 23.37 & @xmath416 08 57.6 & 100 & & ngc 2610 & 239.6 + 13.9 + wd 0915 + 201 & 09 18 33.11 & + 19 53 08.2 & 100 & da0.5 & & + wd 0939 + 262 & 09 42 50.65 & + 26 01 00.0 & 100 & da0.5 & & + wd 0948 + 534 & 09 51 25.98 & + 53 09 31.0 & 126 & da.46 & & + lss 1362 & 09 52 44.50 & @xmath446 16 47.1 & 100 & & hedr 1 & 273.6 + 06.1 + wd 0950 + 139 & 09 52 58.94 & + 13 44 34.9 & 110 & da.46 & egb 6 & 221.5 + 46.3 + wd 1003@xmath4441 & 10 05 45.79 & @xmath444 21 33.4 & 120 & pg1159 & lo 4 & 274.3 + 09.1 + wd 1034 + 001 & 10 37 03.81 & @xmath400 08 19.6 & 100 & doz.5 & & + wd 1111 + 552 & 11 14 47.72 & + 55 01 08.3 & 94 & dao.54 & ngc 3587 & 148.4 + 57.0 + k 1 - 22 & 11 26 43.78 & @xmath434 22 11.5 & 141 & & k 1 - 22 & 283.6 + 25.3 + wd 1144 + 004 & 11 46 35.21 & + 00 12 33.1 & 150 & do.34 & & + lotr 4 & 11 52 29.22 & @xmath442 17 38.6 & 120 & & lotr 4 & 291.4 + 19.2 + bldz 1 & 11 53 06.71 & @xmath450 50 57.1 & 128 & & bldz 1 & 293.6 + 10.9 + be uma & 11 57 44.84 & + 48 56 17.9 & 105 & & be uma & 144.8 + 65.8 + wd 1159@xmath4034 & 12 01 45.97 & @xmath403 45 41.3 & 140 & dqzo.4 & & + wd 1253 + 378 & 12 55 14.77 & + 37 32 29.7 & 100 & da.5 & & + mewe 1 - 3 & 13 28 04.90 & @xmath454 41 58.4 & 100 & & mewe 1 - 3 & 308.2 + 07.7 + wd 1342 + 443 & 13 44 26.87 & + 44 08 33.3 & 79 & da.7 & & + wd 1424 + 534 & 14 25 55.40 & + 53 15 25.2 & 110 & pg1159.46 & & + wd 1501 + 664 & 15 02 09.94 & + 66 12 19.6 & 170 & dzq.3 & & + wd 1517 + 740 & 15 16 46.23 & + 73 52 07.0 & 110 & do.5 & & + wd 1520 + 525 & 15 21 46.56 & + 52 22 03.6 & 150 & pg1159.3 & javdst 1 & 085.4 + 52.3 + wd 1522 + 662 & 15 22 56.70 & + 66 04 41.5 & 140 & do & & + wd 1532 + 033 & 15 35 09.85 & + 03 11 16.0 & 100 & da.5 & & + wd 1547 + 015 & 15 49 44.98 & + 01 25 54.6 & 100 & da.5 & & + wd 1622 + 323 & 16 24 49.05 & + 32 17 02.0 & 78 & da.65 & & + wd 1625 + 280 & 16 27 33.73 & + 27 54 33.6 & 117 & dao.43 & abell 39 & 047.0 + 42.4 + wd 1707 + 427 & 17 08 47.75 & + 42 41 00.4 & 100 & doz.5 & & + wd 1729 + 583 & 17 29 50.37 & + 58 18 09.0 & 84 & da0.6 & & + wd 1738 + 669 & 17 38 02.60 & + 66 53 47.7 & 95 & da.53 & & + wd 1749 + 717 & 17 49 04.50 & + 71 45 08.9 & 100 & dao.5 & & + wd 1751 + 106 & 17 53 32.27 & + 10 37 23.7 & 117 & pg1159 & abell 43 & 036.0 + 17.6 + hatr 7 & 17 54 09.30 & @xmath460 49 57.1 & 100 & & hatr 7 & [email protected] + wd 1827 + 778 & 18 25 08.72 & + 77 55 37.1 & 76 & da.68 & & + wd 1830 + 721 & 18 30 04.90 & + 72 11 34.5 & 100 & do.5 & & + wd 1851@xmath4088 & 18 54 37.18 & @xmath408 49 38.8 & 90 & dao.56 & ic 1295 & [email protected] + wd 1917 + 461 & 19 19 10.21 & + 46 14 51.2 & 88 & dao.57 & abell 61 & 077.6 + 14.7 + wd 1958 + 015 & 20 00 39.25 & + 01 43 40.6 & & pg1159 & ngc 6852 & [email protected] + wd 2114 + 239 & 21 16 52.34 & + 24 08 50.9 & 108 & dao.47 & abell 74 & [email protected] + wd 2115 + 339 & 21 17 08.30 & + 34 12 27.3 & 170 & pg1159.3 & mwp 1 & [email protected] + wd 2209 + 825 & 22 08 25.54 & + 82 44 55.3 & 100 & do & & + wd 2246 + 066 & 22 49 25.29 & + 06 56 45.8 & 100 & da.51 & & + wd 2324 + 397 & 23 27 15.97 & + 40 01 23.4 & 126 & do.4 & & + wd 2333 + 301 & 23 35 53.33 & + 30 28 06.2 & 170 & doz.3 & jn 1 & [email protected] + llccrrcl wd 0005 + 511 & 23366144 & 0.289 & 0.157 & @xmath53 0.761 & @xmath5 = 14.19 & 0.01164 & + pg 0038 + 199 & 23366400 & 0.014 & 0.050 & @xmath53 0.163 & @xmath33 = 15.37 & 0.00374 & + wd 0044@xmath4121 & 23366656 & -2.677 & 9.641 & @xmath53 26.247 & @xmath5 = 12.87 & 0.03925 & very bright nebular emission + wd 0103 + 732 & 23366912 & 2.760 & 0.141 & & @xmath39 = 16.35 & 0.00058 & superposed on diffuse emission + wd 0108 + 100 & 23367168 & 0.058 & 0.057 & @xmath53 0.230 & @xmath34 = 16.33 & 0.00130 & a point source at 12 away + wd 0109 + 111 & 23367424 & 0.269 & 0.055 & & @xmath33 = 16.05 & 0.00200 & + wd 0121@xmath4756 & 23367680 & 0.002 & 0.038 & @xmath53 0.116 & @xmath33 = 16.16 & 0.00181 & + wd 0127 + 581 & 23368192 & 0.338 & 0.142 & & @xmath39 = 17.44 & 0.00021 & superposed on diffuse emission + wd 0130@xmath4196 & 23368448 & 0.055 & 0.049 & @xmath53 0.201 & @xmath5 = 14.00 & 0.01386 & + wd 0237 + 241 & 23368704 & 0.073 & 0.067 & @xmath53 0.274 & @xmath34 = 16.28 & 0.00137 & a point source at 95 away + cspn ic 289 & 23368960 & 1210.9 & 18.2 & @xmath53 1265.6 & @xmath5 = 15.14 & 0.00485 & very bright diffuse emission + wd 0316 + 002 & 23369216 & -0.026 & 0.056 & @xmath53 0.142 & @xmath54 = 19.66 & 0.00007 & + wd 0322 + 452 & 23369472 & 0.059 & 0.062 & @xmath53 0.245 & @xmath39 = 17.20 & 0.00026 & + cspn ngc 1360 & 23369728 & -0.605 & 1.752 & @xmath53 4.651 & @xmath5 = 12.37 & 0.06221 & bright nebular emission + wd 0439 + 466 & 23369984 & 9.200 & 0.157 & & @xmath5 = 13.66 & 0.01896 & superposed on diffuse emission + wd 0444 + 049 & 23370240 & -0.003 & 0.064 & @xmath53 0.187 & & & two sources at 1011 away + wd 0500@xmath4156 & 23370496 & 0.013 & 0.047 & @xmath53
0.156 & @xmath33 = 16.16 & 0.00181 & + cspn k 1 - 27 & 23370752 & 11.603 & 0.372 & @xmath53 12.721 & @xmath34 = 16.40 & 0.00122 & bright nebular emission + wd 0556 + 106 & 23371008 & 0.051 & 0.071 & @xmath53 0.264 & @xmath39 = 17.40 & 0.00022 & a point source at 125 away + wd 0615 + 556 & 23371264 & 0.110 & 0.049 & @xmath53 0.256 & @xmath34 = 15.90 & 0.00194 & two sources at 34 away + wd 0615 + 655 & 23371520 & 0.028 & 0.042 & @xmath53 0.153 & @xmath39 = 15.70 & 0.00105 & + cspn abell 15 & 23371776 & 14.221 & 0.342 & @xmath53 15.248 & @xmath39 = 15.72 & 0.00103 & very bright nebular emission + cspn abell 20 & 23372032 & 0.021 & 2.645 & @xmath53 7.956 & @xmath39 = 16.56 & 0.00048 & very bright nebular emission + wd 0726 + 133 & 23372288 & 0.916 & 0.114 & & @xmath34 = 16.58 & 0.00104 & superposed on diffuse emission + cspn ngc 2438 & 23372544 & 12.410 & 13.659 & & @xmath34 = 17.02 & 0.00069 & superposed on diffuse emission + wd 0753 + 535 & 23372800 & 0.197 & 0.159 & @xmath53 0.675 & @xmath34 = 16.58 & 0.00104 & very bright nebular emission + wd 0823 + 316 & 23373056 & -0.073 & 0.055 & @xmath53 0.092 & @xmath5 = 15.73 & 0.00282 & + cspn ngc 2610 & 23373312 & 127.5 & 7.2 & @xmath53 149.0 & @xmath33 = 16.29 & 0.00160 & very bright nebular emission + wd 0915 + 201 & 23373568 & 0.039 & 0.057 & @xmath53 0.210 & @xmath39 = 16.64 & 0.00044 & + wd 0939 + 262 & 23373824 & -0.008 & 0.053 & @xmath53 0.151 & @xmath33 = 15.57 & 0.00311 & + wd 0948 + 534 & 23374080 & 0.060 & 0.043 & @xmath53 0.189 & @xmath33 = 16.15 & 0.00183 & + cspn lss 1362 & 23374336 & -2.934 & 4.032 & @xmath53 9.164 & @xmath5 = 12.81 & 0.04148 & very bright nebular emission + wd 0950 + 139 & 23374592 & 11.740 & 0.066 & & @xmath5 = 16.10 & 0.00200 & + wd 1003@xmath4441 & 23374848 & 8.821 & 3.688 & @xmath53 19.887 & @xmath39 = 16.60 & 0.00046 & very bright nebular emission + wd 1034 + 001 & 23375104 & 0.964 & 0.133 & @xmath53 1.364 & @xmath5 = 14.37 & 0.00986 & bright nebular emission + wd 1111 + 552 & 23375360 & 2.190 & 1.320 & @xmath53 6.149 & @xmath34 = 16.71 & 0.00092 & bright nebular emission + cspn k 1 - 22 & 23375616 & 1.070 & 0.143 & & @xmath5 = 14.27 & 0.01081 & superposed on diffuse emission + wd 1144 + 004 & 23375872 & -0.044 & 0.067 & @xmath53 0.158 & @xmath33 = 16.13 & 0.00186 & + cspn lotr 4 & 23376128 & 32.298 & 0.359 & @xmath53 33.375 & @xmath39 = 16.50 & 0.00050 & very bright nebular emission + cspn bldz 1 & 23376384 & 3.746 & 1.109 & @xmath53 7.073 & @xmath39 = 18.40 & 0.00009 & bright nebular emission + cspn be uma & 23376640 & 0.031 & 0.045 & @xmath53 0.165 & @xmath5 = 13.63 & 0.01949 & + wd 1159@xmath4034 & 23376896 & 0.044 & 0.058 & @xmath53 0.218 & @xmath5 = 15.73 & 0.00282 & + wd 1253 + 378 & 23377152 & 0.034 & 0.042 & @xmath53 0.159 & @xmath5 = 15.45 & 0.00365 & + cspn mewe 1 - 3 & 23377408 & 77.376 & 0.794 & @xmath53 79.759 & & & very bright nebular emission + wd 1342 + 443 & 23377664 & 0.218 & 0.041 & & @xmath54 = 17.56 & 0.00049 & + wd 1424 + 534 & 23377920 & 0.141 & 0.044 & @xmath53 0.273 & @xmath54 = 16.97 & 0.00084 & a point source 9 away + wd 1501 + 664 & 23378176 & 0.011 & 0.038 & @xmath53 0.125 & @xmath39 = 15.90 & 0.00087 & a point source 2 away + wd 1517 + 740 & 23378432 & 0.005 & 0.038 & @xmath53 0.118 & @xmath55 = 15.90 & 0.00069 & + wd 1520 + 525 & 23378688 & 0.232 & 0.042 & @xmath53 0.358 & @xmath34 = 16.38 & 0.00125 & in diffuse 24 um emission + wd 1522 + 662 & 23378944 & 0.016 & 0.036 & @xmath53 0.125 & @xmath55 = 16.40 & 0.00043 & + wd 1532 + 033 & 23379200 & 0.029 & 0.050 & @xmath53 0.179 & @xmath33 = 16.42 & 0.00142 & + wd 1547 + 015 & 23379456 & -0.034 & 0.047 & @xmath53 0.107 & @xmath34 = 16.43 & 0.00119 & + wd 1622 + 323 & 23379712 & 0.039 & 0.038 & @xmath53 0.152 & @xmath5 = 13.77 & 0.01713 & + wd 1625 + 280 & 23379968 & 0.093 & 0.153 & @xmath53 0.553 & @xmath34 = 16.13 & 0.00157 & superposed on diffuse emission + wd 1707 + 427 & 23380224 & -0.069 & 0.038 & @xmath53 0.046 & @xmath39 = 16.70 & 0.00042 & + wd 1729 + 583 & 23380480 & -0.025 & 0.038 & @xmath53 0.090 & @xmath54 = 19.76 & 0.00007 & + wd 1738 + 669 & 23380736 & 0.021 & 0.036 & @xmath53 0.129 & @xmath5 = 15.45 & 0.00365 & a point source 125 away + wd 1749 + 717 & 23380992 & 0.004 & 0.039 & @xmath53 0.123 & @xmath34 = 16.78 & 0.00086 & + wd 1751 + 106 & 23381248 & -1.840 & 2.641 & @xmath53 6.084 & @xmath5 = 15.33 & 0.00407 & bright nebular emission + cspn hatr 7 & 23381504 & 0.194 & 0.294 & @xmath53 1.075 & @xmath5 = 15.75 & 0.00277 & bright nebular emission + wd 1827 + 778 & 23381760 & 0.022 & 0.037 & @xmath53 0.133 & @xmath34 = 16.64 & 0.00098 & a point source 125 away + wd 1830 + 721 & 23382016 & -0.006 & 0.040 & @xmath53 0.113 & @xmath55 = 17.00 & 0.00025 & a point source 7 away + wd 1851@xmath4088 & 23382272 & 5.071 & 5.496 & @xmath53 21.557 & @xmath39 = 16.90 & 0.00035 & very bright nebular emission + wd 1917 + 461 & 23382528 & 0.049 & 0.090 & @xmath53 0.318 & @xmath39 = 17.39 & 0.00022 & bright nebular emission + wd 1958 + 015 & 23382784 & 136.0 & 1.5 & @xmath53 140.4 & @xmath39 = 17.90 & 0.00014 & very bright nebular emission + wd 2114 + 239 & 23383040 & -0.073 & 0.061 & @xmath53 0.109 & @xmath39 = 17.05 & 0.00030 & + wd 2115 + 339 & 23383296 & -0.120 & 0.341 & @xmath53 0.903 & @xmath5 = 14.18 & 0.01174 & bright nebular emission + wd 2209 + 825 & 23383552 & -0.010 & 0.040 & @xmath53 0.110 & @xmath34 = 16.58 & 0.00104 & + wd 2246 + 066 & 23383808 & 0.067 & 0.052 & @xmath53 0.222 & @xmath55 = 16.80 & 0.00030 & + wd 2324 + 397 & 23384064 & 0.003 & 0.075 & @xmath53 0.227 & @xmath5 = 15.40 & 0.00382 & near a patch of diffuse emission + wd 2333 + 301 & 23384320 & 0.363 & 0.154 & @xmath53 0.824 & @xmath34 = 16.70 & 0.00093 & complex emission , no point source + lcccccc cspn k1 - 22 & 829@xmath2042 & 706@xmath2036 & 681@xmath2038 & 808@xmath2042 & 1,070@xmath20143 & @xmath5312,000 + cspn ngc2438 & 103@xmath2040 & 91@xmath2057 & 82@xmath2058 & 117@xmath2095 & 12,410@xmath2013,700 & ... + wd0103 + 732 & 88@xmath2015 & 71@xmath2019 & 58@xmath2020 & 132@xmath2035 & 2,760@xmath20141 & @xmath5355,000 + wd0109 + 111 & ... & ... & ... & ... & 269@xmath2055 & ... + wd0127 + 581 & 76@xmath2047 & 56@xmath2026 & 185@xmath20161 & 92@xmath2045 & 338@xmath20142 & ... + wd0439 + 466 & 862@xmath2044 & 514@xmath2026 & 336@xmath2022 & 180@xmath2014 & 9,200@xmath20157 & 9,200 @xmath208,400 + wd0726 + 133 & [email protected] & [email protected] & @xmath5319.8 & @xmath5317.0 & 916@xmath20114 & ... + wd0950 + 139 & 977@xmath2015 & 1176@xmath2015 & 1773@xmath2036 & 3772@xmath2037 & 11,740@xmath2066 & ... + wd1342 + 443 & ... & ... & ... & ... & 218@xmath2041 & ... + llccccc cspn k1 - 22 & 2009 feb 26 & staring & 7@xmath860 s & 7@xmath860 s & 8@xmath8120 s & 8@xmath8120 s + wd0103 + 732 & 2008 sep 13 & staring & 8@xmath860 s & 8@xmath860 s & 4@xmath8120 s & 8@xmath8120 s + wd0127 + 581 & 2008 oct 09 & mapping & 2@xmath860 s & 2@xmath860 s & 2@xmath8120 s & 2@xmath8120 s + wd0439 + 466 & 2008 sep 7 & mapping & 8@xmath860 s & 16@xmath860 s & 4@xmath8120 s & 4@xmath8120 s llrlrrclll cspn k1 - 22 & cspn & 141,000 & 0.076 & 1,330 & 3.3 & 700 + 150 & 1.1@xmath56 & egb6-like & 1 , 2 + cspn ngc2438 & cspn & 114,000 & 0.25 & 1,200 & 4.1 & 1200 + 150 & 4.7@xmath56 & egb6-like & 1 , 3 + wd0103 + 732 ( cspn egb1 ) & da.34 & 150,000 & 0.58 & 650 & 3.7 & 190 & 1.4@xmath57 & helix - like & 4 , 5 + wd0109 + 111 & doz.46 & 110,000 & 0.065 & 280 & 2.3 & 150 & 4.9@xmath58 & helix - like & 6 , 7 , 8 + wd0127 + 581 ( cspn sh2 - 188 ) & dao.49 & 102,000 & 0.27 & 600 & 1.7 & 900 + 150 & 6.6@xmath57 & egb6-like & 4 , 9 + wd0439 + 466 ( cspn sh2 - 216 ) & da.61 & 83,000 & 0.065 & 129 & 2.4 & 150 & 2.4@xmath57 & helix - like & 10 , 11 + wd0726 + 133 ( cspn abell21 ) & pg1159 & 130,000 & 0.13 & 541 & 2.0 & 150 & 1.6@xmath57 & helix - like & 2 , 11 , 12 + wd0950 + 139 ( cspn egb6 ) & da.46 & 110,000 & 0.21 & 645 & 3.0 & 500 + 150 & 4.7@xmath56 & egb6-like & 5 , 13 + wd1342 + 443 & da.7 & 79,000 & & 437 & 1.4 & 150 & 5.1@xmath57 & helix - like & 13 + wd2226@xmath4210 ( cspn helix ) & dao.49 & 110,000 & 0.03 & 210 & 2.6 & 120 & 2.5@xmath56 & helix - like & 11 , 14 + | two types of dust disks around white dwarfs ( wds ) have been reported : small dust disks around cool metal - rich wds consisting of tidally disrupted asteroids , and a large dust disk around the hot central wd of the helix planetary nebula ( pn ) possibly produced by collisions among kuiper belt - like objects . to search for more dust disks of the latter type
, we have conducted a _ spitzer _
mips 24 survey of 71 hot wds or pre - wds , among which 35 are central stars of pne ( cspns ) .
nine of these evolved stars are detected and their 24 flux densities are at least two orders of magnitude higher than their expected photospheric emission . considering the bias against detection of distant objects ,
the 24 detection rate for the sample is @xmath015% .
it is striking that seven , or @xmath120% , of the wd and pre - wds in known pne exhibit 24 excesses , while two , or 56% , of the wds not in pne show 24 excesses and they have the lowest 24 flux densities .
we have obtained follow - up _ spitzer _
irs spectra for five objects .
four show clear continuum emission at 24 , and one is overwhelmed by a bright neighboring star but still show a hint of continuum emission . in the cases of wd0950 + 139 and
cspn k1 - 22 , a late - type companion is present , making it difficult to determine whether the excess 24 emission is associated with the wd or its red companion .
high - resolution images in the mid - ir are needed to establish unambiguously the stars responsible for the 24 excesses . |
in the famous 1966 paper @xcite v. arnold described the dynamics of an ideal fluid as a geodesic flow on the group of volume - preserving diffeomorphisms of a fixed domain equipped with the metric defined by the kinetic energy .
he also showed how sectional curvature of this group enters the problem of lagrangian stability of eulerian fluid motions . in this paper
we are concerned with the exterior geometry of the group of volume - preserving diffeomorphisms considered as an infinite - dimensional submanifold of the group of all diffeomorphisms .
we study its second fundamental form and introduce the associated shape operator .
the focal points of this infinite - dimensional submanifold turn out to correspond to shocks forming in potential solutions of the inviscid burgers equation on the underlying finite - dimensional domain .
this provides a common geometric framework for both eulerian hydrodynamics of ideal fluids and the phenomenon of shocks of the inviscid burgers equation .
more precisely , let @xmath0 be a compact @xmath1-dimensional riemannian manifold .
consider the group @xmath2 of sobolev class diffeomorphisms of @xmath0 along with its subgroup @xmath3 of diffeomorphisms preserving the riemannian volume form @xmath4 .
( as usual , if @xmath5 both groups can be considered as smooth hilbert manifolds . ) for a curve @xmath6 in @xmath2 defined on an interval @xmath7 $ ] its @xmath8-energy is given by @xmath9 where the norm @xmath10 is defined by the @xmath8-inner product @xmath11 on each tangent space @xmath12 .
the corresponding ( weak ) riemannian metric on @xmath2 is right - invariant when restricted to the subgroup @xmath3 of volume - preserving diffeomorphisms , although it is not right - invariant on the whole of @xmath2 .
arnold @xcite proved that the euler equation of an ideal incompressible fluid occupying the manifold @xmath0 , @xmath13 for a divergence - free field @xmath14 , corresponds to the equation of the geodesic flow of the above right - invariant metric on the group @xmath3 .
he also showed that lagrangian instability of such flows , regarded as geodesic deviation , can be estimated in terms of sectional curvatures of the group of volume - preserving diffeomorphisms @xmath3 and provided first such curvature estimates of this group .
below , by regarding the group of volume - preserving diffeomorphisms as a subgroup of the group of all diffeomorphisms ( cf .
e.g. @xcite ) , we describe its sectional curvatures by means of the second fundamental form of the embedding @xmath15 and relate its principal curvatures to the distance from its nearest focal point . recall that the distance from a submanifold @xmath16 to the first focal point along a geodesic in the direction normal to this submanifold gives a lower bound on the principal curvature radius of @xmath16 ( in the subspace containing the normal direction ) : @xmath17 @xmath18 hence a lower bound on the distance to a focal point provides an upper bound for ( principal ) curvatures of @xmath16 .
this motivates our study of the riemannian geometry of the embedding @xmath15 as the following sequence of implications , which we make precise below : @xmath19 geodesics in the full diffeomorphism group @xmath2 with respect to the above @xmath8-metric are described by solutions of the inviscid burgers equation .
the geodesics normal to the submanifold @xmath15 of volume - preserving diffeomorphisms are given by potential burgers solutions , see figure 1 .
@xmath20 the first focal point along a ( potential ) burgers solution determines the moment when the shock wave develops .
the geometry of the initial profile allows one to precisely estimate the lifespan of this solution .
@xmath21 focal points along normal geodesics are in one - to - one correspondence with conjugate points along the projection of these geodesics to the space of densities @xmath22 equipped with the wasserstein @xmath8-metric .
@xmath23 on the other hand , sectional curvatures of @xmath3 can be explicitly computed from the gauss - codazzi equations using the second fundamental form ( or , the corresponding shape operator ) of the embedding @xmath15 .
@xmath24 the location of the first focal point along a normal geodesic can be estimated in terms of the spectral radius of the shape operator of @xmath25 .
epsf in the next three sections we describe the geometry behind relations @xmath26 , respectively .
the second fundamental form , the shape operator and its spectral properties in items @xmath27 are described in section [ shape ] .
a detailed analysis of the shape operator will be postponed to a future publication . in the last section we recall the properties of asymptotic directions on the diffeomorphism groups discussed in @xcite within the framework of the differential geometry of @xmath2 .
finally , we mention that the connection between the lagrangian instability of ideal fluids and negativity of curvatures of the corresponding diffeomorphism group was precised by v.arnold in @xcite , and further explored in @xcite , @xcite .
the geometric characterization of the onset of shock waves in terms of focal points of the embedding @xmath15 is based on the following observation .
geodesics in the group @xmath2 with respect to the @xmath8-metric ( [ l2-metric ] ) correspond to solutions of the burgers equation : each particle moves with constant velocity along a geodesic in @xmath0 .
geodesics normal to the submanifold @xmath25 have potential initial conditions .
first assume for simplicity that @xmath0 is equipped with a flat metric .
then the statement that particles move with constant velocity along their own geodesics in @xmath0 follows from the `` flatness '' of the @xmath8-metric ( [ l2-metric ] ) . to see that the corresponding velocity field satisfies
the burgers equation denote the flow of fluid particles by @xmath28 and let @xmath14 be its velocity field @xmath29 the chain rule immediately gives @xmath30 and hence the burgers equation on @xmath0 @xmath31 is equivalent to @xmath32 the equation of freely flying non - interacting particles in @xmath0 . in the general case
, any riemannian metric on @xmath0 induces a unique levi - civita @xmath8-connection @xmath33 on @xmath2 , which is determined pointwise by the riemannian connection @xmath34 on the manifold @xmath0 itself ( see for example @xcite or @xcite ) .
then the same chain rule leads to the burgers equation @xmath35 equivalent to the riemannian version of freely flying particles : @xmath36 .
the latter equation has the explicit form @xmath37 , where @xmath38 are the christoffel symbols of @xmath34 in a local coordinate system @xmath39 on @xmath0 . finally , observe that the tangent space to the subgroup @xmath25 at the identity consists of divergence - free vector fields and hence the space of normals is given by gradients of @xmath40 functions on @xmath0 .
thus horizontal geodesics are the ones whose initial velocities are gradient fields @xmath41 .
the following result describes a connection between focal points and formation of shocks for the burgers equation . the first focal point to the submanifold @xmath42 in the direction @xmath43
is given by the moment of the shock wave formation for the burgers equation with this initial condition @xmath41 .
in the 1-dimensional case the shock wave solutions of the burgers equation appear from the inflection points of the initial velocity profile , see figure 2 .
such points correspond to @xmath44 , i.e. to @xmath45 for @xmath46 .
epsf before the appearence of a shock wave any solution with smooth initial data remains smooth and the shock wave is the first moment of non - smoothness .
furthermore , for a given potential @xmath47 , any such solution is given by a family of diffeomorphisms @xmath48 parameterized by time @xmath49 .
the loss of smoothness occurs when the potential @xmath50 ceases to be @xmath51-concave with respect to the distance function @xmath52 on the manifold . in this case
@xmath6 is no longer a diffeomorphism . on the other hand ,
as long as the potential is @xmath51-concave , the curve @xmath6 remains the shortest curve joining the diffeomorphisms @xmath53 and @xmath6 , see @xcite , and focal points on @xmath6 can not appear . finally , we note that the moment when a shock wave appears corresponds to a focal point ( i.e. , to a caustic ) , rather than to a cut ( or maxwell ) point .
this follows from the fact that in any shock wave the first collisions start between neighboring particles , see e.g. @xcite . to see why such collisions lead to the existence of a neighboring shorter geodesic
, one can employ the following heuristic argument .
consider a shock wave generated by a solution to the 1-dimensional burgers equation with the initial velocity @xmath54 , see figure 2 .
assume that this shock wave emerged at time @xmath55 and consider it at time @xmath56 .
consider a perturbation @xmath57 of the original solution which at the initial moment differs from the initial velocity @xmath14 on a small segment .
this segment consists of exactly those particles which are glued together in the shock wave between the times @xmath58 $ ] .
perturbations @xmath57 , whose velocity profile on the @xmath59-plane ( see figure 2 ) lies below that of @xmath14 correspond to smooth perturbations of the initial velocities of these particles , slowing them down to ensure that the shock wave develops after time @xmath55 but no later than @xmath60 .
note that such a perturbation defines another solution which coincides with the original one after time @xmath60 but represents a shorter geodesic .
indeed , since we decreased the velocity vectors of the particles , the length of the new geodesic , being the integral of the velocity squares , can only get smaller . by arbitrarily shortening the segment
@xmath58 $ ] this implies that the first moment @xmath55 of the original shock wave for the initial condition @xmath14 indeed defines a focal point , rather than a cut point , since there exists a shorter geodesic nearby . in higher dimensions
the shock waves are first generated from the special points of the initial potential @xmath61 : singularities of type @xmath62 modulo certain linear and quadratic terms in local orthogonal coordinate charts , see @xcite .
the list of initial singularities , possible bifurcations of the shock waves and other related questions for the inviscid burgers equations , can be found in @xcite .
a given shock wave for the burgers equation forms at a caustic corresponding to a deeper degeneration of this potential , see e.g. @xcite .
the most degenerate caustic points correspond to the endpoints of the spectrum for the shape operator @xmath63 , which we describe below .
the lifespan of a solution before the shock formation provides an estimate for the spectrum of this operator : a _ lower bound _ for the lifespan , and hence for the distance to the first focal point gives the _ upper bound _ for the ( principal ) curvatures of the submanifold @xmath15 .
next we would like to show that conjugate points along geodesics in the wasserstein space of densities are in 1 - 1 correspondence with focal points on the group of diffeomorphisms .
namely , consider the fibration @xmath64 , where @xmath65 is the projection of diffeomorphisms onto the space of ( normalized ) densities on @xmath0 .
thus , two diffeomorphisms @xmath66 and @xmath67 belong to the same fiber if and only if @xmath68 for some volume - preserving diffeomorphism @xmath69 .
( moreover , this projection can be defined for more general maps by tracing how they transport the given standard density form on @xmath0 . )
the map @xmath65 is a riemannian submersion @xcite and the above @xmath8-metric on @xmath2 induces the wasserstein ( or , kantorovich
rubinstein ) metric in the space of densities @xmath22 .
the following general observation holds .
consider a projection @xmath70 between two ( possibly weak ) riemannian manifolds which is a riemannian submersion and where @xmath71 an infinite - dimensional lie group .
let @xmath72 be a geodesic starting at a point @xmath73 in @xmath74 and let @xmath75 be a horizontal lift of this geodesic to @xmath71 , i.e. a horizontal geodesic in @xmath71 whose initial point lies in the fiber @xmath76 and whose projection to @xmath74 is @xmath72
. assume , in addition , that the metric restricted to @xmath76 is right invariant .
( in our case , the fiber @xmath76 is the group @xmath3 of volume - preserving diffeomorphisms for @xmath73 ) .
[ prop : gamma ] the points along the geodesic @xmath77 conjugate to the initial position @xmath73 are in one - to - one correspondence with the focal points of the fiber @xmath78 , regarded as a submanifold in @xmath71 , along a horizontal geodesic @xmath75 in @xmath71 .
moreover , the multiplicities of the conjugate points in @xmath74 coincide with the multiplicities of the corresponding focal points in @xmath71 .
recall that conjugate points and their generalizations , the focal points , arise as singularities of the riemannian exponential map . in a general infinite - dimensional space
two types of such points , called epi - conjugate ( epi - focal ) and mono - conjugate ( mono - focal ) points , can be found .
roughly speaking , geometric significance of the former has to do with covering properties of the exponential map , while the latter are responsible for the minimizing properties of geodesics .
the statement follows from the submersion property of @xmath65 .
since geodesics in the base manifold have unique lifts to horizontal geodesics in @xmath71 we can identify the geodesics emanating from @xmath73 that are close to @xmath72 and which come together near a conjugate point in @xmath79 with the horizontal geodesics emanating from @xmath76 , that are close to @xmath80 and come together near the focal point in @xmath71 .
furthermore , the argument works for both mono- and epi - conjugate ( or focal ) points , since the exponential map is nondegenerate along the fibers .
indeed , shifts along the fibers correspond to right multiplication by an element of @xmath76 , while the metric in @xmath71 is @xmath76-invariant .
geodesics in the wasserstein space @xmath22 of ( smooth ) densities are projections of the horizontal geodesics in the diffeomorphism group @xmath2 . for a given potential @xmath81
they are the diffeomorphisms @xmath82 parameterized by @xmath49 . according to the theorem on polar decomposition on manifolds @xcite every non - degenerate map @xmath83 has a unique decomposition @xmath84 into a `` gradient map '' @xmath85 for a @xmath51-concave potential @xmath81 and a volume - preserving map @xmath86 .
consequently , the projection of the family of diffeomorphisms @xmath87 remains a shortest geodesic in @xmath74 ( in the space of non - degenerate maps ) as long as the potential @xmath88 remains @xmath51-concave .
we next turn to the shape operator of the embedding @xmath15 .
the eigenvalues of this operator are traditionally called the principal curvatures and the corresponding eigenvectors of unit norm are the principal directions . in our case , at each point in @xmath25 there is a family of shape operators parametrized by vectors normal to the submanifold .
the _ shape operator _ of the submanifold @xmath3 at the identity is the operator @xmath89 on @xmath90 defined by @xmath91 where @xmath92 , @xmath93 and @xmath94 are in @xmath90 and @xmath95 is the second fundamental form of @xmath25 . by right invariance
we similarly define the shape operator at any point @xmath75 in @xmath25 .
recall that the ( weak ) riemannian metric on @xmath2 induces on both the group @xmath2 and its subgroup @xmath25 the unique levi - civita connections @xmath33 and @xmath96 , where @xmath97 is the hodge projection @xmath98 onto divergence - free vector fields on @xmath0 conjugated with the right translation by @xmath75 ( see @xcite for more details ) .
the second fundamental form @xmath99 of the submanifold @xmath100 is the difference of the two connections , which at the identity is given by @xmath101 where @xmath102 is the hodge projection onto the gradient fields on @xmath0 .
the second fundamental form , as well as the curvature tensors of the connections on @xmath2 and @xmath25 , are bounded multilinear operators satisfying the gauss - codazzi equations , see @xcite .
the above gives an explicit formula for the shape operator at the identity @xmath103 one can check that the shape operator is bounded and self - adjoint .
the next result provides an estimate of the distance to the first focal point along a normal geodesic in terms of the spectral radius of @xmath63 .
[ thmf ] suppose that @xmath0 is a compact manifold of non - negative sectional curvature .
let @xmath6 be the geodesic normal to @xmath25 at @xmath104 with @xmath105 and let @xmath106 .
then @xmath6 contains a focal point in the interval @xmath107 . in this case our strategy to show existence of focal points along geodesics normal to the fibre @xmath25 is to obtain estimates on the second variation of the @xmath8-energy ( [ energy ] ) .
the latter is given by the following formula @xmath108 for any vector field @xmath109 along @xmath75 such that @xmath110 is tangent to the fibre @xmath25 and @xmath111 .
here @xmath112 is the curvature of the connection @xmath33 on the group @xmath2 .
we discuss this theorem and its proof in detail in a future publication .
we point out that the conclusion of theorem [ thmf ] holds provided that the geodesic @xmath75 is defined on the interval @xmath113 $ ] .
furthermore , this geodesic does not minimize the @xmath8-distance between @xmath25 and any @xmath6 with @xmath114 and hence the focal point is in fact mono - focal .
another corollary of the same formula for the second variation is as follows .
[ cor : non - pos ] suppose now that @xmath0 has non - positive sectional curvature .
consider any initial vector @xmath115 with a convex function @xmath47 .
then there can be no focal points on the corresponding geodesic .
indeed , for such a manifold @xmath0 and any @xmath109 one obtains @xmath116 in particular , if @xmath47 is convex then its hessian @xmath117 is non - negative and so @xmath118 , implying the absence of focal points between @xmath53 and any other @xmath119 .
asymptotic directions for the group of volume - preserving diffeomorhisms also appear naturally in the context of the exterior geometry of this group .
recall , that a vector tangent to a riemannian submanifold is _ asymptotic _ if the ( vector - valued ) second fundamental form evaluated on it is zero .
the geodesics issued in the direction of this vector , one in the submanifold and the other in the ambient manifold , have a second order of tangency .
( note that in general two geodesics with a common tangent will have only a simple , i.e. first order , tangency . ) asymptotic vectors for the submanifold of volume - preserving diffeomorhisms among all diffeomorhisms are given at the tangent space to the identity by vector fields @xmath120 satisfying @xmath121 , see @xcite . such fields rarely exist .
for instance , for diffeomorhism groups of surfaces one has the following sufficient condition for non - existence of such directions .
consider the square length function @xmath123 on @xmath0 . at its maximum point @xmath124
we get from @xmath125 that the jacobi matrix @xmath126 is degenerate at @xmath124 .
this implies that @xmath127}=0 $ ] for a divergence - free field @xmath120 . now employing the identity @xmath128 where @xmath129 stands for the ricci curvature of the metric @xmath130 , which holds for any divergence - free vector field @xmath120 on @xmath0 , we find that @xmath131 since in two dimensions the ricci curvature @xmath129 and the gaussian curvature @xmath132 coincide .
however , for an asymptotic field @xmath120 this implies that @xmath133 contradicting the assumption @xmath134 on @xmath0 , and in particular at the point @xmath124 .
vanishing of the second fundamental form on asymptotic vectors implies vanishing of the projections of this form to any gradient direction .
in particular , such vectors must be asymptotic simultaneously for all the projections of the second fundamental form described above .
this implies the following sufficient condition for the non - existence of asymptotic directions : if at least in one of the gradient projections the second fundamental form is sign - definite , then there are no asymptotic directions .
the latter sufficient condition can also be given in terms of gradient solutions of the burgers equation : in such a gradient direction all shocks develop only as @xmath49 changes in the one direction ( say , increases ) , and do not develop as @xmath49 changes in the other direction .
we thank c.villani for fruitful discussions and r.wendt for drawing figures .
b.k . is grateful to the msri at berkeley for kind hospitality .
this research was partially supported by an nserc grant and conducted during the period the first author was employed by the clay mathematics institute as a clay book fellow .
arnold , v.i . , baryshnikov , yu.m .
, bogaevsky , i.a . , _ singularities and bifurcations of potential flows _ , in : gurbatov , s. n. , malakhov , a. n. , saichev , a. i. , _ nonlinear random waves and turbulence in nondispersive media : waves , rays , particles _
, nonlinear science : theory and applications .
manchester university press , manchester , 1991 , 290300 . | we establish a simple relation between curvatures of the group of volume - preserving diffeomorphisms and the lifespan of potential solutions to the inviscid burgers equation before the appearance of shocks .
we show that shock formation corresponds to a focal point of the group of volume - preserving diffeomorphisms regarded as a submanifold of the full diffeomorphism group and , consequently , to a conjugate point along a geodesic in the wasserstein space of densities .
this establishes an intrinsic connection between ideal euler hydrodynamics ( via arnold s approach ) , shock formation in the multidimensional burgers equation and the wasserstein geometry of the space of densities . |
a series of improper ferroelectrics - geometric , magnetic , and electronic - has been found to exhibit the multiferroic coupling as well.@xcite the extent of coupling , however , varies . among them , the magnetic ferroelectrics ( tbmno@xmath3 , dymno@xmath3 ) exhibit stronger coupling though at a lower temperature ( below the commensurate to incommensurate magnetic transition temperature @xmath10 @xmath028 k ) .
the magnetic transition point for the geometric ferroelectrics ( ymno@xmath3 , lumno@xmath3 ) too turn out to be low ( @xmath080 - 100 k ) . for both of these classes , detailed work has already been done to reveal the magnitude of the coupling as well as its underlying mechanism.@xcite except lufe@xmath6o@xmath11 [ ref . 4 ] , the electronic ferroelectrics , on the other hand , seem to have received relatively lesser attention in spite of the fact that in quite a few of them both the ferroelectric and magnetic transition points are higher and closer to room temperature .
the electronic ferroelectricity originates from noncentrosymmetric charge / orbital order@xcite - even frustrated charge order - and is expected to offer a sizable polarization ( @xmath05 - 6 @xmath12c/@xmath13).@xcite however , because of intrinsic electronic inhomogeneity and consequent large leakage current , it is difficult to measure the polarization directly by global electrical measurements.@xcite in several other compounds too , such as , pr@xmath14ca@xmath15mno@xmath3 ( x = 0.45 - 0.55)@xcite , fe@xmath3o@xmath11,@xcite pr(sr@xmath4ca@xmath5)@xmath6mn@xmath6o@xmath7 etc,@xcite broken symmetry of the charge / orbital order creates polar domains . yet the extent of multiferroic coupling in them has not been investigated in detail .
it is important to mention here that even in a well - known multiferroic compound tbmno@xmath3 , which exhibits very strong coupling , the ferroelectric polarization is small ( @xmath00.09 @xmath12c/@xmath13).@xcite therefore , small polarization in electronic ferroelectrics need not be detrimental to strong multiferroicity .
motivated by this observation and also because of relatively higher ferroelectric and magnetic transition points in electronic ferroelectric systems , we explored , in this work , the extent of magnetocapacitance in two such compounds - bilayer pr(sr@xmath4ca@xmath5)@xmath6mn@xmath6o@xmath7 and perovskite pr@xmath1ca@xmath2mno@xmath3 .
the bilayer system undergoes two consecutive charge / orbital order transitions at @xmath16 @xmath0380 k and at @xmath17 @xmath0300 k. the crystallographic structure transforms from orthorhombic amam to pbnm at @xmath16 and from pbnm to am2 m at @xmath17 with the rotation of orbital ordered chains from along b- to a - axis .
the chains stack in the form of stripes along a- and b - axis , respectively , in these phases .
this rotation is associated with a rearrangement of mn@xmath18 and mn@xmath19 ions to yield a polarization along b - axis.@xcite the three - dimensional magnetic structure , sets in at @xmath20 @xmath0150 k , is antiferromagnetic with easy axis oriented along b - axis . in the case of the perovskite system , on the other hand , mixed site- and bond - centered charge ordered structure breaks the symmetry and yields a polarization at the charge order transition point @xmath0220 k.@xcite for this system the magnetic transition point t@xmath8 is @xmath0100 k.@xcite we observed a sizable magnetocapacitance ( @xmath05 - 90% ) under 5 t field , in these systems , across 20 - 100 k below the magnetic transition point ( t@xmath8 ) .
the single crystals of pr(sr@xmath4ca@xmath5)@xmath6mn@xmath6o@xmath7 and pr@xmath1ca@xmath2mno@xmath3 have been grown by float - zone technique in an image furnace.@xcite the disc shaped samples of diameter @xmath02.0 - 3.0 mm and thickness @xmath00.5 - 1.0 mm have been extracted from the boule . the laue pattern and
edx data have been included in the supplementary document.@xcite fresh surface has been cleaved out for all the electrical measurements .
the measurements have been carried out in two - probe configuration using silver conducting paste and silver wires .
apart from dc resistivity , dielectric spectra have been recorded across 100 hz - 5 mhz with 1v amplitude over a temperature range 20 - 100 k both under zero and 5 t magnetic field .
a cryogenic instrument cryostat and a superconducting magnet have been used for covering the temperature and field range while an impedance analyzer of wayne kerr ( model 6500b ) has been used for the dielectric property measurement .
the protocol of dielectric measurements was as follows .
the sample was cooled down from room temperature under zero magnetic field to 20 k and the dielectric spectra were recorded at several temperatures in between 20 and 100 k. the sample was heated back to the room temperature and then cooled down again to 20 k under zero field . a magnetic field of 5 t
was then applied and the dielectric spectra were recorded again at those temperatures in between 20 and 100 k. we restricted all the dielectric measurements within a temperature regime of 20 - 100 k as with the rise in temperature the conductivity of the sample increases .
this , in turn , makes accurate determination of the intrinsic capacitance of the samples difficult as contribution from interface starts dominating the overall dielectric response .
since our main aim in this paper was to determine the extent of magnetocapacitance in these electronic ferroelectric manganite systems , we carried out all the measurements within a temperature range where the influence of conductivity is small .
in fig . 1 we show the dc resistivity ( @xmath21 ) versus temperature ( @xmath22 ) patterns for the pr(sr@xmath4ca@xmath5)@xmath6mn@xmath6o@xmath7 ( pscmo ) and pr@xmath1ca@xmath2mno@xmath3 ( pcmo ) samples across the temperature range of our interest .
the @xmath23 has been measured along parallel ( @xmath24 ) and perpendicular to the c - axis ( @xmath25 ) . for the bilayer sample , reasonably large anisotropy
could be observed whereas for the perovskite one , not much anisotropy has been observed . since not much anisotropy
could be observed , all the data presented here for the perovskite sample have been obtained from measurements within the ab - plane .
the onset of magnetic order is shown by a distinct anomaly around t@xmath8 in all the @xmath23 plots .
for the bilayer system , there is a hysteresis around t@xmath8 in the @xmath26 signifying first - order transition while @xmath27 does not show any hysteresis . the first order transition within ab - plane possibly results from stronger influence of magnetic order with easy axis oriented along b - axis on the charge / orbital order essentially confined within ab - plane and stacked along c - axis .
a second hysteresis could also be seen in the @xmath26 plot around 110 k which possibly reflects the low temperature structural transition ( from orthorhomic to monoclinic ) observed by others.@xcite for the perosvkite system , of course , no such hysteresis in @xmath23 pattern could be observed around t@xmath8 ( inset of fig .
1b ) . in fig .
2 , we show the real and imaginary parts of the permittivity ( @xmath28 , @xmath29 ) across 20 - 100 k measured under zero and 5 t field for both the samples . for the bilayer sample , the @xmath30 , @xmath31 and @xmath32 , @xmath33 versus
temperature plots under zero field at different frequencies are shown in figs .
2(a ) and 2(b ) whereas the corresponding plots under 5 t field are shown in figs .
2(c ) and 2(d ) , respectively . for the perovskite sample ,
the plots under zero and 5 t field are shown in figs .
2(e ) and 2(f ) , respectively . in this case
, only the data obtained from the measurement within ab - plane are shown .
qualitatively all the plots exhibit nearly similar features .
there is a maximum in the real part of the permittivity which shifts toward lower temperature under the magnetic field .
this could be associated with a crossover in the mechanism of polarization in these samples .
while at lower temperature the ferroelectric order governs the polarization , with the rise in temperature , enhanced contribution of the charge carrier hopping starts playing an important role .
signature of such a crossover could be observed in the plot of magnetocapacitance versus temperature as well ( discussed later ) .
3 shows the % change in the real part of the permittivity ( @xmath34 = @xmath35 ) at a frequency of 2 mhz under 5 t field . for the bilayer pscmo system
, @xmath35 is negative and its magnitude is @xmath020% at low temperature ( 20 - 30 k ) .
the magnitude increases up to @xmath060% around 50 k and then decreases and eventually changes the sign to become positive and reaches @xmath030% at 90 k. interestingly , the temperature dependence of the magnetodielectric data measured both within and perpendicular to ab - plane nearly resemble each other .
this is quite expected as the dielectric permitivity versus temperature plots too , do not show any significant change in the patterns ( fig .
2 ) . there is a slight frequency dependence as well whose significance will be clear in the analyses that followed .
the extent of anisotropy varies within 2 - 5 across the entire temperature range . in the case of the perovskite pcmo sample ,
on the other hand , the @xmath35-@xmath22 plot turns out to be quite different . here
, @xmath35 is positive and @xmath040% at low temperature ( 20 - 50 k ) .
it decreases and changes sign at @xmath075 k to become negative and then increases sharply up to @xmath0250% by @xmath0100 k. therefore , while the bilayer sample exhibits a switch from negative to positive , the perovskite one exhibits a switch from positive to negative as far as the magnetodielectric effect is concerned over a temperature range 20 - 100 k. the large magnetodielectric effect reflects a strong multiferroic coupling in these systems .
the origin of the multiferroic coupling in electronic ferroelectric systems could either be fundamental - the coupling between charge / orbital and spin degrees of freedom@xcite - or the striction effect as in other conventional multiferroic systems .
the dielectric constant under a magnetic field , in the present cases , has both decreased or increased with respect to that under zero field at low temperature ( @xmath020 k ) and then changed substantially as the temperature is raised in order to trigger a change in the sign of the magnetodielectric effect ( fig .
3 ) . it has been found already that in different multiferroic systems such as bifeo@xmath3 [ ref .
20 ] , lufe@xmath6o@xmath11 [ ref . 21 ] , livcuo@xmath11 [ ref .
22 ] , the polarization or dielectric constant is suppressed at the onset of magnetic order . in bifeo@xmath3 ,
the suppression results from the structural effects - enhanced antiferrodistortive rotation of the octahedra around the axis of polarization at the onset of magnetic order leads to the suppression of the noncentrosymmetry.@xcite there are , however , contradictory results as well . for instance , in lufe@xmath6o@xmath11 , while ikeda @xmath36.@xcite have observed a rise in the polarization at t@xmath8 , subramanian @xmath36.@xcite have noticed a drop in the dielectric constant under a magnetic field below t@xmath8 .
switching of the polarization axis under magnetic field@xcite - as noted in tbmno@xmath3 and dymno@xmath3 - from along one to another crystallographic axis can , of course , lead to such a contradictory result when measured in a rather polycrystalline sample or in a twinned crystal because of variation in the averaging effect .
it has been clearly shown that for strongly correlated electron systems , the charge / orbital order is intimately coupled with the spin order@xcite and , in fact , spin order helps in strengthening the charge / orbital order and even recovering the charge order lost under photoexcitation .
therefore , in the present case , where the ferroelectricity is originating from the noncentrosymmetric charge / orbital order , it could be expected that the polarization or the dielectric constant would increase at the onset of magnetic order at t@xmath8 or under a magnetic field at below t@xmath8 .
however , we do not notice any consistent pattern of increase or suppression of the dielectric constant under the magnetic field for both the samples .
we contrast this result with the result of magnetocapacitance ( shown and discussed below ) to point out that the dielectric constant , measured even under a relatively higher frequency ( 1 - 5 mhz ) , does not really represent the data corresponding to a relaxed state@xcite and , therefore , could be misleading as far as the behavior of the intrinsic polarization is concerned .
we also point out that we observe a wide dispersion of the dielectric permittivity with frequency and the relaxation dymanics deviates from the ideal debye - type relaxation by a large extent signifying a much broader range of relaxation time constant ( @xmath37 ) . as a result ,
as mentioned above , the magnetodielectric effect ( measured even at a higher frequency 1 - 5 mhz ) exhibits a slight frequency dependence .
we , therefore , explored the complete dielectric spectra across 100 hz to 5 mhz .
we describe the complex plane impedance plots in the @xmath38 frame,@xcite for the entire range of temperature both under zero and 5 t magnetic field , by the generalized davidson - cole model of relaxation@xcite @xmath39 where @xmath40 and @xmath41 are the static and high frequency resistances , respectively , @xmath42 is the frequency , @xmath43 is the relaxation time scale , and @xmath44 is the relaxation exponent which varies between 0 and 1 for the non - debye correlated relaxation .
fitting the experimental data with an equivalent circuit model of a resistance ( @xmath45 ) and a capacitance ( @xmath46 ) connected in parallel representing the bulk resistance and the capacitance , respectively , we obtain the circuit parameters @xmath45 and @xmath46 . in fig .
4 , we plot the @xmath47 and @xmath48 under zero and 5 t field for both the samples .
we also show the relaxation exponent @xmath44 which varies within 0.5 - 0.9 and signifies strongly correlated relaxation . in figs .
4(a ) , ( b ) , and ( c ) , the @xmath47 , @xmath48 , and @xmath49 obtained along and perpendicular to the c - axis for the bilayer sample are shown . and figs .
4(d ) , 4(e ) , and 4(f ) show the corresponding plots obtained from measurement within ab - plane for the perovskite sample . finally , in fig
5 , the % change in the circuit capacitance ( @xmath50 = @xmath51 ) is shown for both the samples . once again , for the bilayer sample , we show the magnetocapacitance obtained from measurements both along and perpendicular to the c - axis .
the change appears to be varying within @xmath05 - 90% for both the bilayer and the perovskite systems .
the extent of anisotropy varies rather nonmonotonically within a factor of 1 - 5 across the entire temperature range .
it is important to point out here that ( i ) the magnetic field influences the circuit resistance ( @xmath45 ) as well and ( ii ) even though , there is a qualitative similarity between the overall temperature dependence of the magnetodielectric and magnetocapacitive effects ( figs . 3 and 5 )
, the switch in the magnitude of the magnetodielectric effect from positive to negative or reverse could not be observed in the case of the magnetocapacitance .
the circuit model applied here helps in extracting the relaxed and hence intrinsic dielectric constant of the sample .
the magnetocapacitance turns out to be always positive for both the systems across 20 - 100 k signifying higher capacitance under a magnetic field even though the difference does have a strong temperature dependence .
the nonmonotonic temperature dependence of the magnetocapacitance for both the systems can possibly be understood in the following way . at low temperature ( @xmath020 k )
, the increase in capacitance under a magnetic field could result from enhanced polarization via spin order - charge / orbital order coupling . with the increase in temperature ,
then , the @xmath51 decreases sharply for the bilayer and broadly for the perovskite systems . beyond a certain temperature - varying within 40 - 75
k - it rises again .
the decrease in @xmath51 at intermediate temperature range could result from thermalization of the charge / orbital / spin structure .
further rise at higher temperature signifies a crossover in the origin of the overall capacitance of the sample - from purely ferroelectric order to a situation where ferroelectric order coexists with charge carrier transport.@xcite the hopping of the charge carriers within a charge / spin / orbital ordered background gives rise to a stronger effect in the high temperature range .
enhanced hopping under a magnetic field as a result of local spin alignment and/or enhanced charge disproportionation - from mn@xmath18 , mn@xmath19 to mn@xmath53 , mn@xmath54 within an ordered structure - could possibly lead to a rise in the dielectric constant which , in turn , would yield an enhanced magnetocapacitance at higher temperature .
the small charge transferred via disproportionation seems to be significant in increasing the dipole moment.@xcite the local spin alignment under a magnetic field also contributes in increasing the hopping and hence the dielectric constant .
it is important to mention here that the circuit resistance ( @xmath45 ) increases substantially in the case of the bilayer system and decreases only slightly in the case of the perovskite system .
therefore , it appears that the hopping in presence of the local charge transfer via disproportionation and spin alignment enhances the dipole moment substantially and does not contribute much to the charge conduction .
it is also noteworthy , that the temperature range of the crossover varies from the bilayer to the perovskite system .
this might result from variation in the extent of coupling between charge / orbital and spin degrees of freedom in different systems .
stronger coupling leads to weaker temperature dependence - as observed in the case of the perovskite system - and vice versa .
we contrast our observation to that made in a non - charge / orbital ordered , non - ferroelectric system such as la@xmath6nimno@xmath55 where the asymmetric hopping in presence of disorder promotes the dielectric constant.@xcite application of magnetic field , in that case , reduces the disorder and hence the dielectric constant yielding negative magnetodielectric effect . in our case of a charge / orbital ordered network , on the contrary , the hopping seems to be promoted by an enhanced local alignment of spin structure ( which , in turn , introduces a local disorder within an otherwise 3d antiferromagnetic background ) and/or charge disproportionation .
we , therefore , observe a positive magnetocapacitance effect .
this comparsion between the magnetodielectric and magnetocapacitive effects shows that it is not sufficient to study the influence of magnetic field on the dielectric constant alone for a system where dielectric relaxation takes place over a wider frequency range and/or interface contribution too plays a significant role.@xcite one needs to extract the contribution of the purely capacitive part from an accurate description of the dielectric relaxation dynamics to find out the magnetocapacitive effect . in systems where direct determination of the ferroelectric polarization is difficult , it is important to determine the influence of magnetic field on the capacitive component of the dielectric response to estimate the intrinsic multiferroic coupling of the systems .
in summary , we show that there is a sizable magnetocapacitance ( @xmath05 - 90% ) in electronic ferroelectric pr(sr@xmath4ca@xmath5)@xmath6mn@xmath6o@xmath7 and pr@xmath1ca@xmath2mno@xmath3 systems below the magnetic transition point even though their ferroelectric polarization could be small .
there is a remarkable difference between the magnetodielectric effect at higher frequency ( 1 - 5 mhz ) and the magnetocapacitive effect even though the qualitative patterns are similar .
the change - over from positive to negative or vice versa , observed in the magnetodielectric effect , could not be noticed in the magnetocapacitance part .
this result shows that it is important to extract the magnetocapacitance directly for the systems where the dielectric relaxation dymanics deviates from the debye - type relaxation by a large extent and the relaxation takes place over a wider frequency range . for the bilayer sample ,
both the magnetodielectric and magnetocapacitance exhibit a reasonably strong anisotropy even though the temperature dependence of them appear to follow qualitatively similar patterns .
not much anisotropy could , however , be observed for the perovskite sample .
large magnetocapacitance at higher temperature possibly originates , primarily , from strong influence of magnetic field on the hopping in presence of charge disproportionation where the role of the ferroelectric order could be smaller . notwithstanding this
, the large magnetocapacitance in these electronic ferroelectric systems , anyway , enhances the scope of their use in magnetoelectric sensor applications enormously .
@xmath56 this work has been supported by the department of science and technology ( dst ) , govt of india .
one of the authors ( uc ) acknowledges support in the form of a senior research fellowship from dst .
another author ( sg ) acknowledges support in the form of a research associateship from csir , govt of india .
99 s .- w .
cheong and m. mostovoy , nature mater .
@xmath57 , 13 ( 2007 ) .
s. picozzi and a. stroppa , eur .
j. b @xmath58 , 240 ( 2012 ) .
see , for example , d.i .
khomskii , physics @xmath59 , 20 ( 2009 ) .
n. ikeda , h. ohsumi , k. ohwada , k. ishii , t. inami , k. kakurai , y. murakami , k. yoshii , s. mori , y. horibe , and h. kito , nature ( london ) @xmath60 , 1136 ( 2005 ) .
j. van den brink and d.i .
khomskii , j. phys.:condens .
matter @xmath61 , 434217 ( 2008 ) .
see , for example , s. picozzi , k. yamauchi , b. sanyal , i.a .
sergienko , and e. dagotto , phys .
@xmath62 , 227201 ( 2007 ) .
p. barone , k. yamauchi , and s. picozzi , phys .
@xmath63 , 077201 ( 2011 ) .
see , for example , a.m.l .
lopes , j.p .
araujo , v.s .
amaral , j.g .
correia , y. tomioka , and y. tokura , phys .
@xmath64 , 155702 ( 2008 ) and all the relevant references therein .
efremov , j. van den brink , and d.i .
khomskii , nature mater .
@xmath65 , 853 ( 2004 ) .
k. yamauchi , t. fukushima , and s. picozzi , phys .
b @xmath66 , 212404 ( 2009 ) .
y. tokunaga , t. lottermoser , y.s .
lee , r. kumai , m. uchida , t. arima , and y. tokura , nature mater .
@xmath67 , 937 ( 2006 ) .
t. goto , t. kimura , g. lawes , a.p .
ramirez , and y. tokura , phys .
@xmath68 , 257201 ( 2004 ) .
see , for example , v. dediu , c. ferdeghini , f.c .
matacotta , p. noar , and g. ruani , phys .
@xmath69 , 4489 ( 2000 ) .
d. shulyatev , s. karabashev , a. aresenov , ya .
mukovskii , and s. zverkov , j. cryst .
growth @xmath70 , 810 ( 2002 ) .
the supplementary material is available on request .
beale , s.r .
bland , r.d .
johnson , p.d .
hatton , j.c .
cezar , s.s .
dhesi , m.v .
zimmermann , d. prabhakaran , and a.t .
boothroyd , phys .
b @xmath66 , 054433 ( 2009 ) .
see , for example , k.i .
kugel and d.i .
khomskii , sov .
@xmath71 , 231 ( 1982 ) .
l. paolasini , r. caciuffo , a. sollier , p. ghigna , and m. altarelli , phys .
@xmath72 , 106403 ( 2002 ) .
chuang @xmath36 . , phys .
@xmath73 , 127404 ( 2013 ) .
s. goswami , d. bhattacharya , p. choudhury , b. ouladdiaf , and t. chatterji , appl .
@xmath62 , 073106 ( 2011 ) .
m. a. subramanian , t. he , j. chen , n.s .
rogado , t.g .
calvarese , and a.w .
sleight , adv .
@xmath74 , 1737 ( 2006 ) .
y. yasui , y. naito , k. sato , t. moyoshi , m. sato , and k. kakurai , j. phys .
@xmath75 , 023712 ( 2008 ) .
see , for example , f. schrettle , p. lunkenheimer , j. hemberger , v. yu .
ivanov , a.a .
mukhin , a.m. balbashov , and a. loidl , phys .
@xmath76 , 207208 ( 2009 ) .
d. davidson and r. cole , j. chem .
@xmath77 , 1484 ( 1951 ) .
d. choudhury , p. mandal , r. mathieu , a. hazarika , s. rajan , a. sunderasan , u.v .
waghmare , r. knut , o. karis , p. nordblad , and d.d .
sarma , phys .
@xmath78 , 127201 ( 2012 ) .
we observe a reasonable ferroelectric polarization ( @xmath00.02 - 0.05 @xmath12c/@xmath13 ) in these bilayer and perovskite systems at low temperature ( 20 - 40 k ) which decreases sharply with the rise in temperature further , marking a crossover in the contribution to the overall dielectric constant of the sample from purely ferroelectric polarization to a mixed scenario of polarization and charge carrier hopping ; u. chowdhury , s. goswami , d. bhattacharya , a. midya , p. mandal , p. das , and y.m .
mukhovskii ( unpublished ) . m.h .
tang , j.w .
hou , j. zhang , g.j .
dong , and w. shu , solid state commun .
@xmath79 , 1453 ( 2010 ) .
t. kimura , s. kawamoto , i. yamada , m. azuma , m. takano , and y. tokura , phys .
b @xmath80 , 180401(r ) ( 2003 ) . | we have observed a sizable positive magnetocapacitance ( @xmath05 - 90% ) in perovskite pr@xmath1ca@xmath2mno@xmath3 and bilayer pr(sr@xmath4ca@xmath5)@xmath6mn@xmath6o@xmath7 system under 5 t magnetic field across 20 - 100 k below the magnetic transition point t@xmath8 . the magnetodielectric effect , on the other hand , exhibits a crossover : ( a ) from positive to negative for the perovskite system and ( b ) from negative to positive for the bilayer system over the same temperature range .
the bilayer pr(sr@xmath4ca@xmath5)@xmath6mn@xmath6o@xmath7 system exhibits a sizable anisotropy as well .
we have also noticed the influence of magnetic field on the dielectric relaxation characteristics of these systems .
these systems belong to a class of improper ferroelectrics and are expected to exhibit charge / orbital order driven ferroelectric polarization below the transition point t@xmath9 .
large magnetocapacitance in these systems shows typical multiferroic behavior even though the ferroelectric polarization is small in comparison to that of other ferroelectrics . |
the possible existence of extra dimensions has been a fascinating idea in physics ever since kaluza - klein theory was proposed @xcite .
consistent string theories demand the existence of extra dimensions @xcite .
however , if the string scale ( @xmath5 ) is as high as the grand unification scale ( @xmath10 gev ) or the planck scale ( @xmath11 gev ) , as is the case for a weakly coupled heterotic string , then the length scale of the compactified extra dimensions @xmath12 would be too small to be appreciable experimentally .
recent developments in string theory indicate that the string scale can be much lower than the planck scale and even close to the electroweak scale @xcite .
this possibility provides new avenues towards many theoretical issues such as alternative solutions to the gauge hierarchy problem @xcite , fermion mass and flavor mixings @xcite , and new inflationary cosmological models @xcite .
more importantly , such a scenario may lead to a rich phenomenology and is thus experimentally testable at low energies @xcite .
assume that there are @xmath13 extra dimensions in which only gravity can propagate while standard model ( sm ) fields are confined to four dimensional spacetime .
the large value of the planck scale can be understood by gauss law from the relation @xmath14 where the string scale @xmath5 is taken to be the planck mass in @xmath0 dimensions . for
, @xmath15 can range from 1 fm to 1 mm for @xmath16 to 2 @xcite .
there are no direct gravitational tests sensitive to those small scales yet @xcite .
such large extra dimensions manifest themselves only through interactions involving the kaluza - klein ( kk ) modes of the gravitons with enhanced coupling strength after summing over the many contributing light kk states . the effective theory governing the graviton couplings to matter
was described in @xcite .
phenomenological studies showed that future collider experiments can provide constraints on @xmath5 typically of order , depending on the collider center of mass energies @xcite .
astrophysical ( cosmological ) considerations have been used to impose lower bounds on @xmath5 to be about 30 ( 100 ) tev for @xmath7 and very weak bounds for higher dimensions @xcite . in this paper
we consider radiative corrections to the masses of the electroweak gauge bosons @xmath17 arising from the exchange of massive spin-2 kk gravitons .
the motivation for this study is two - fold .
firstly , a rigorously renormalizable theory of gravity does not exist
. it would be interesting to explore to what extent the formalism in ref .
@xcite is finite against radiative corrections in the sense of an effective field theory .
an early attempt in ref .
@xcite for a scalar self - energy correction showed that the radiative corrections are proportional to the scalar mass , instead of the ultraviolet cutoff .
another one - loop calculation for the muon @xmath18 also reached finite results @xcite . secondly ,
if one is able to compute radiative effects from kk gravitons , one may hope to examine constraints on new physics characterized by the string scale from precision electroweak measurements .
in fact , a recent paper appeared to estimate the @xmath19 parameter @xcite , which was found to be both ultraviolet and infrared divergent .
however , our results arrive at completely different conclusions .
the rest of the paper is organized as follows . in section
ii we compute one - loop self - energy diagrams for a gauge boson from exchange of massive spin-2 kk gravitons . to do so
, we need to derive the four point couplings of two gravitons and two gauge bosons which are beyond that given in ref .
@xcite and have been left out in the previous calculations @xcite . in section iii
we adopt the @xmath1-formalism @xcite and constrain the string scale based on current experimental values of the oblique parameters .
computing oblique parameters has an advantage over calculating @xmath20 mass corrections directly since they are free of certain technical and conceptual uncertainties .
we discuss our results and conclude in section iv . in two appendices we present the relevant feynman rules and describe the regularization of the infrared divergences .
when the typical energy scale of interest is much smaller than the string scale @xmath5 , the compactified higher dimensional theory can be described by an effective theory where only relevant light degrees of freedom are retained , namely , those related to the so called kaluza - klein states . in principle , if only gravity is of higher dimensional origin , spin-0 , 1 and 2 kk states can arise @xcite .
the spin-1 states decouple from the standard model fields and make no contribution to the self - energies .
naively , one expects the spin-0 contributions to be of the same order of magnitude as that of the spin-2 states since they have the same coupling strength .
however , the properties of the spin-0 states are model - dependent since there is no symmetry to protect their masses . in the following
we shall concentrate on the spin-2 kk states ( which we will call kk gravitons ) . the physical process that we consider in this paper is the radiative corrections to the weak gauge boson masses at the one - loop level from these kk gravitons .
we focus our attention on the transverse part of the self - energy @xmath21 between gauge bosons @xmath22 and @xmath23 , as it is the only part relevant to us .
these self - energies are written as @xmath24 note that both @xmath25 and @xmath26 are identically zero , following from the simple fact that they must be proportional to the photon mass , as required by the gravitational nature of the kk graviton - matter interactions .
there are two diagrams contributing to the self - energy of a gauge boson , as shown in fig .
[ graphs ] . the first ( seagull diagram )
involves a four point coupling of two kk gravitons and two gauge bosons , and the second ( rainbow diagram ) involves two three point couplings of a kk graviton and two gauge bosons .
these two diagrams are at the same order in the gravitational coupling @xmath27 , where @xmath28 is the usual four - dimensional newton s constant . to obtain the four - point vertex feynman rule required for the seagull diagram
, one needs to expand the graviton - matter interaction lagrangian to order @xmath29 , which is beyond the order of the expansion given in @xcite .
we provide this feynman rule in appendix a. the corresponding seagull diagram is purely transverse , proportional to @xmath30 , with @xmath31 where we have included a factor @xmath32 to avoid double - counting when we sum over the kk modes and the subscript @xmath33 indicates that the variable is in euclidean space .
the summation over the kk modes in a tower can be written as an integration because of the near - degeneracy of the kk states , @xmath34 where @xmath35 is the kk state density . by convention for the torus compactification @xcite , the relation between the four - dimensional newton s constant and the @xmath36-dimensional string scale @xmath5 is given by @xmath37 using eqs .
( [ summation ] ) , ( [ density ] ) and ( [ convention ] ) , we obtain @xmath38 the above sum is divergent in the ultraviolet for @xmath39 unless @xmath40 falls off very rapidly with @xmath41 . since the effective theory is only expected to be valid below the string scale , we introduce an explicit cutoff @xmath42 to regularize the mass sum , where @xmath43 and parameterizes the sensitivity to the cutoff .
we shall use the same cutoff for the momentum integral as for the mass summation .
we believe that this regulator will best reflect the ultraviolet behavior of the divergence .
our final result for the seagull diagram is @xmath44 where @xmath45 and @xmath46 .
the necessary feynman rules for calculating the rainbow diagram have been derived in @xcite and are summarized in appendix a. the complete expression for the general @xmath47 is lengthy and unilluminating .
we will instead present the special cases relevant to our current consideration , @xmath48 and @xmath49 .
the former can be written as @xmath50\ , \label{rainbow0}\end{aligned}\ ] ] where we have introduced a dimensionless mass ratio between the gauge boson mass and the ultraviolet cutoff @xmath51 .
similarly , @xmath52}^2}}}\ , \label{rainbowm}\end{aligned}\ ] ] where @xmath53 is a feynman parameter and @xmath54 \nonumber \\ & & + r\biggl [ 4xy ( 2 - 8z + { z^2 } ) + 4{y^2 } ( 4 + 9{z^2 } ) + { x^2 } ( 1 - 14\,z + 15{z^2 } ) \biggr ] \nonumber \\ & & - x ( { x^2 } + 4xy+23{y^2})\ .\end{aligned}\ ] ] it is important to note that the on - shell mass corrections are proportional to the gauge boson mass squared @xmath55 , similar to the results obtained in ref .
@xcite for a scalar mass correction , and not to @xmath56 as found in @xcite .
consequently , there is no hard quadratic dependence on the cutoff . on the other hand , the proportionality factor
@xmath57 appearing in the above expressions implies that the results are rather sensitive to the precise value of the cutoff in comparison to the string scale @xmath5 .
this is an intrinsic uncertainty for any process involving virtual kk graviton exchanges in an effective theory .
before we end this section , a few remarks related to the potential uncertainties of the results are in order : 1 .
( [ seagull ] ) , ( [ rainbow0 ] ) and ( [ rainbowm ] ) appear to be infrared divergent as a consequence of the pole at @xmath58 for massless gravitons .
this must be an artifact of the fixed - order perturbative calculation since gravity is known to be infrared safe .
we will regularize the infrared singularity using the principles of dimensional regularization and perform the minimal subtraction .
the procedure is described in appendix b. 2 .
it is interesting to investigate the limit for a large value of the cutoff @xmath59 , or equivalently @xmath60 . as an illustration we consider @xmath7 .
the seagull diagram is independent of @xmath61 and gives @xmath62 and the rainbow diagram gives @xmath63 therefore , the total self - energy correction @xmath64 takes the following values at @xmath65 and @xmath66 : @xmath67 one would hope that the self - energy corrections vanish as the string scale is set to infinity , as required by the decoupling theorem .
this has been explicitly shown not to be the case by the naive results in eq .
( [ tot ] ) .
the problem lies in the fact that an unknown cosmological constant can also contribute to the gauge boson self - energies via gravitational interactions .
a non - zero cosmological constant term is of the form @xmath68 and would lead to an additional contribution to the self - energy by tadpole diagrams , @xmath69 this term can drastically change the previously calculated self energies . if @xmath70 , eq . ( [ pil ] ) could be at the same order of eq .
( [ tot ] ) and could thus provide the appropriate counter - term . however , due to the lack of a consistent way of determining the cosmological constant , the precise value of the term in eq . ( [ pil ] ) is not known . as a result , the self - energies and subsequently , corrections to the @xmath71 and @xmath72 boson masses have inherent uncertainties .
3 . in our calculations for the gauge boson self - energies ,
we have adopted the momentum - cutoff scheme to regularize the divergent mass sum and the momentum integral , since we consider this scheme a most direct reflection of the ultraviolet behavior .
we anticipate that the physics results would not depend upon the specific regularization scheme .
it is important to emphasize that the above potential ambiguities are of no concern to us if we adopt the @xmath1-formalism since the oblique parameters are manifestly finite @xcite .
the regulator independence built into the definition of the oblique parameters make the infrared and ultraviolet divergences as well as the irrelevant constants drop out when taking the difference of the appropriate combination of self - energies , as we will present next .
the studies of electroweak radiative corrections have proven to be powerful in constraining new physics @xcite beyond the sm .
a convenient parameterization of new physics from a higher scale is the @xmath1-formalism @xcite .
the @xmath2 , @xmath3 and @xmath4 parameters can be obtained by evaluating the self - energy corrections at the energy scales @xmath73 and 0 .
we write the oblique parameters as @xmath74 where @xmath75 and @xmath76 are the sine and cosine of the weak mixing angle and @xmath77 is the fine structure constant , all measured at @xmath78 .
the @xmath2 ( @xmath79 ) parameter measures the difference in the contribution of new physics to neutral ( charged ) current processes at different energy scales .
@xmath4 is generally small .
the @xmath3 parameter serves as a comparison between the new contributions to the neutral and charged current processes at low energy , proportional to @xmath80 .
comparing experimental data mainly from the lep and slc to sm predictions with @xmath81 gev leads to the bounds @xcite , @xmath82 we perform an oblique correction analysis using the calculations of the previous section and the sm parameters @xcite @xmath83 we assume the central values of eq .
( [ stu ] ) to be the sm predictions and attribute the error bars to the physics contribution of our current interest .
our numerical results are presented in figs .
( [ s])-([u ] ) where we have set @xmath84 . in fig .
[ s ] , we have plotted @xmath85 , the excess contribution to the sm value of the @xmath2 parameter arising from the kk graviton exchanges of the one - loop self - energy diagrams , for @xmath7 ( positive ) and @xmath86 ( negative ) .
these values must lie within the error bars stated in eq .
( [ stu ] ) to not be in conflict with precision electroweak measurements .
we see that for @xmath7 , this leads to a lower bound @xmath6 tev and for @xmath87 , @xmath88 gev at the @xmath89 level .
there is no constraint for @xmath90 from the @xmath2 parameter and we have therefore neglected to plot the cases @xmath91 .
figure [ t ] shows that the @xmath3 parameter imposes constraints on @xmath5 for all @xmath13 .
this seems to be counter - intuitive at first sight since the interactions of kk gravitons with matter should respect the custodial @xmath92 symmetry and thus lead to a null contribution to @xmath3 .
however , the mass difference of the gauge bosons make their couplings to the gravitons different , resulting in a negative graviton contribution to @xmath3 . in fact , the constraints for @xmath93 are more stringent than that obtained from the @xmath2 parameter .
for instance , the lower limit on @xmath5 is raised to 1.25 ( 0.75 ) tev for @xmath87 ( 6 ) .
the @xmath4 parameter is generally small and it is true here as well .
on the other hand , the error bars on @xmath4 are still rather large and we would not expect improvement by it . indeed , from fig .
[ u ] , we see that the @xmath4 parameter places no constraint whatsoever .
we have obtained significant constraints on the string scale @xmath5 , by considering the @xmath2 , @xmath3 and @xmath4 parameters from precision electroweak data .
one may wonder if the @xmath1-formalism is suitable since gravitons can be lighter than gauge bosons .
we consider our treatment appropriate because as a collective contribution , the relevant effects are characterized by the string scale at about a tev .
a more subtle question is whether the non - oblique corrections would also be as important and how they can be incorporated .
in fact , leading kk graviton corrections appear even at tree level and significant effects were found in the literature @xcite .
we thus view our approach to single out the oblique corrections as the next - to - leading contribution as reasonable . as we see from eq .
( [ tot ] ) , the self - energy corrections remain finite in the limit @xmath94 . as explained in the paragraph following it
, it is possible to achieve `` decoupling behavior '' by introducing a cosmological constant as a counter - term . by no means
is this necessary as far as the oblique parameters are concerned .
the effect of the heavy states does decouple in @xmath2 , @xmath3 and @xmath4 as seen from the relation eq .
( [ tot ] ) , @xmath95 altough demonstrated specifically for the case @xmath7 , we have confirmed this to be true for all @xmath13 .
it is reassuring to obtain the `` decoupling '' relations , which imply that the radiative corrections based on this effective field theory of kk gravitons are under control .
for all of our numerical analysis , the choice @xmath84 corresponds to taking the cutoff scale as the string scale .
this is the same as the choice in most phenomenological studies @xcite .
there is the intrinsic uncertainty due to the ratio parameter @xmath96 , although it is reasonable to assume it to be of order unity .
more definitive results will depend on details of a string model @xcite near the string scale . in summary
, we obtained significant lower bounds on @xmath5 from the @xmath2 and @xmath3 parameters . for one and two standard deviations
, they are @xmath97 these results are comparable to that inferred from current lep ii experiments @xcite and are slightly weaker than those anticipated at future runs of lep and the tevatron .
we thank c. goebel , j. lykken and d. zeppenfeld for discussions .
this work was supported in part by a doe grant no .
de - fg02 - 95er40896 and in part by the wisconsin alumni research foundation .
the three - point vertex is shown in fig .
[ vertices ] and the corresponding feynman rule is @xcite @xmath101\ , \ ] ] where @xmath102 is the tensor that appears in the massless graviton propagator in the de donder gauge , and @xmath103\ .\ ] ] the feynman rule for the four - point vertex as shown in fig .
[ vertices ] is @xmath104\ ] ] where @xmath105\ , \ ] ] @xmath106\nonumber\\ & & -\biggl[(\mu,\nu ) \leftrightarrow ( \rho,\sigma)\biggr]\ , \end{aligned}\ ] ] and @xmath107 + \left(\rho \leftrightarrow \sigma\right)\biggr\ } + c_{\mu\nu,\rho\sigma}k_{1\tau}k_{2\omega}\nonumber\ .\end{aligned}\ ] ] we provide details of the treatment of the infrared singularity for the case @xmath7 . from the expression for the self - energy contribution from the seagull diagram , eq .
( [ seagull ] ) , we extract the integral @xmath108\ , \nonumber \\ & & + 18\,\left ( { \frac{x}{n-2 } } + \ , { \frac{1}{n}}{}_2f_1\left[1,\,n/2,\ , ( 2+n)/2,\,-1/x\right ] \right)\end{aligned}\ ] ] where @xmath109=\sum_{k=0}^{\infty } { \frac{(a)_k ( b)_k}{(c)_k } } { \frac{d^k}{k!}}\ ] ] is the hypergeometric function and @xmath110 is the pochhammer symbol @xmath111 this result is convergent only for @xmath112 . for @xmath113 , we analytically continue the above equation and expand it in a power series in @xmath114 , @xmath115+{\cal o}(\epsilon)\ .\ ] ] integrating the @xmath116 term of the above expression over @xmath117 , we obtain @xmath118\ = \ { \frac{14}{3}}\ .\ ] ] we can repeat the procedure for @xmath48 . the @xmath116 term of the integral in eq .
( [ rainbow0 ] ) is @xmath119 + 3{r^2}\biggl[r\ln { \frac{r}{r+1 } } - 10\ln{\frac{2(r+1)}{r-1 } } \biggr ] \nonumber \\ & & \qquad + { 3\over2}r \left ( 2r-1 + 40\ln 2 \right ) -{\frac{1}{3}}\ , \end{aligned}\ ] ] where @xmath120 is the dilogarithm function .
the same procedure can be carried out for the integral in @xmath49 , but the expressions are exceedingly cumbersome and not at all illuminating .
we do not show them here .
the point we wish to emphasize is that by performing dimensional regularization in the number of extra spacetime dimensions @xmath13 , we are able to systematically isolate the infrared divergence and calculate finite quantities with a sensible physical meaning . using the results in this appendix , we showed that the oblique corrections vanish identically in the limit @xmath94 , as one would expect .
g. shiu and s .- h .
tye , phys .
rev . * d58 * , 106007 ( 1998 ) ; z. kakushadze and phys .
rev . * d58 * , 126001 ( 1998 ) ; i. antoniadis , c. bachas and e. dudas , hep - th/9906039 ; l. e. ibanez , r. rabadan and a. m. uranga , nucl .
* b542 * , 112 ( 1999 ) ; g. aldazabal , l. e. ibanez and f. quevedo , hep - th/9909172 ; s. cullen , and m. peskin , hep - ph/0001166 . | we consider the kaluza - klein scenario in which gravity propagates in the @xmath0 dimensional bulk of spacetime and the standard model particles are confined to a 3-brane .
we calculate the gauge boson self - energy corrections arising from the exchange of virtual gravitons and present our results in the @xmath1-formalism .
we find that the new physics contributions to @xmath2 , @xmath3 and @xmath4 decouple in the limit that the string scale @xmath5 goes to infinity .
the oblique parameters constrain the lower limit on @xmath5 . taking the quantum gravity cutoff to be @xmath5
, @xmath2-parameter constraints impose @xmath6 tev for @xmath7 at the 1@xmath8 level .
@xmath3-parameter constraints impose for @xmath9 .
# 1#1 [ 1]0= 0=0 1= 1=1 0>1 # 1 / |
iron - based superconductors show rich phase diagrams , with the high - temperature superconducting dome being in the close proximity to an antiferromagnetic striped phase @xcite that sets in at a temperature @xmath3 .
in addition , a structural phase transition at @xmath4 , from the high - temperature tetragonal phase into an orthorhombic phase , has been shown to closely follow the magnetic transition @xcite , i.e. : @xmath6 .
it was proposed that spin - fluctuations , associated with the striped phase , lead to emergent electronic nematic degrees of freedom at higher temperatures .
@xcite these electronic nematic degrees of freedom then couple to the lattice and induce the structural phase transition to the ortorhombic phase .
@xcite there is a mounting evidence for an electronic nematic state : resistivity - anisotropy measurements @xcite and the measurement of the elastoresistance , @xcite the observed anisotropies in thermopower , @xcite optical conductivity , @xcite torque magnetometry , @xcite and in stm measurements .
@xcite measurements of the elastic constants showed that the shear modulus strongly softens in the high temperature tetragonal phase.@xcite a theoretical analysis @xcite based upon nematic fluctuations due to a strong magneto - elastic coupling showed that the inverse shear modulus is proportional to the susceptibility of the nematic order parameter @xmath7 , which diverges at the structural phase transition , explaining its softening .
the most direct evidence for the magnetic origin of nematicity so far is the scaling of the shear modulus and the nmr spin - lattice relaxation rate , seen in iron - pnictides .
@xcite an interesting open issue in this context is the lack of such scaling behavior in iron - chalcagonides .
@xcite a relation between nematicity and the raman response of iron based superconductors was already studied in ref . where the kramers - kronig transform of the raman response was compared with the shear modulus , as well as in ref . .
here , we demonstrate , based on an explicit microscopic theory that i ) there is no enhancement of the electronic raman response function in the @xmath8 channel upon lowering the temperature , ii ) that the raman response function develops a pronounced peak at the structural phase transition in the @xmath0 symmetry , and iii ) that there is some response in the @xmath9 channel , which weakens as the temperature is lowered towards the structural transition temperature .
we start from the spin - driven scenario for the nematic phase , in which magnetic fluctuations stabilize a nematic phase , characterized by the broken c@xmath10 symmetry .
the raman response function measures the electronic density - density correlator weighted by appropriate form factors . since electrons interact with spin fluctuations , the latter will manifest themselves in the raman response function in the form of corrections to the electron self energy and the raman vertex , formally expressed in terms of aslamazov - larkin diagrams .
@xcite we show that the leading order aslamazov - larkin ( al ) diagram supports only the @xmath9 and the @xmath0 symmetry , but not the @xmath8 symmetry , which explains the lack of enhancement of the raman response signal as one approaches the structural transition in the @xmath8 channel , as seen in experiments .
@xcite however , this leading order approach can not account for the rapid increase in the amplitude of the raman response function in the @xmath0 channel , as one approaches the structural transition , as seen in the experiments of refs . .
instead it would predict a similar increase only at the magnetic phase transition .
therefore , we go beyond the leading order approximation , and take into account collisions between spin fluctuations that become more and more important as one approaches the nematic / structural transition . our approach is based on the exact same collisions between spin - fluctuations that led to the emergence of spin - induced nematicity in the first place .
formally this is accomplished by inserting a series of quartic paramagnon couplings , mediated by electronic excitations , into the raman response function .
such quartic couplings contain a product of four fermionic green s functions and give rise to a peak of the electronic raman response function at the structural phase transition in the @xmath0 channel .
on the other hand , if we re - sum such higher order al diagrams in the @xmath9 channel , this will lead to the suppression of the raman response in the aforementioned channel . here
we demonstrate that the low frequency raman response in the @xmath0 channel is given by @xmath11 where @xmath12 is the frequency difference between incoming and outgoing photons and @xmath13 the multi - index for momentum and frequency .
@xmath14 stands for the leading order aslamazov - larkin diagram , @xmath15 is the magnetic susceptibility , and @xmath16 the nematic coupling constant of the theory . on the other hand ,
the susceptibility of the nematic order parameter of our model , in the large @xmath5 limit is given by @xmath17 where in a purely electronic theory @xmath18 . in a purely electronic theory , this would then lead to the divergence of the raman response function at the structural phase transition .
however , one needs to include the effect of the lattice dynamics @xcite in order to analyze this problem .
we do so by introducing nemato - elastic coupling and find that , in this case , @xmath19 is shifted .
@xcite here @xmath20 is the elasto - nematic coupling constant , and @xmath21 the bare value of the orthorhombic elastic constant .
we show that when magnetic and structural phase transitions are split @xcite this leads to a maximum of the amplitude of the electronic raman response function in the @xmath0 channel at the structural phase transition , in agreement with the recent experiments .
@xcite the raman response function could then be used to probe the dynamic excitation spectrum of the nematic degrees of freedom , similar to inelastic neutron scattering that probes the dynamic spin excitation spectrum .
the paper is organized as follows . in sec .
[ microscopics ] we present the microscopic model for the spin - driven nematic phase .
we calculate the effective action and analyze it in the large-@xmath5 limit , where @xmath22 is the number of components of the collective paramagnon field .
following ref .
, we derive the condition for the susceptibility of the nematic order parameter to diverge . in sec .
[ raman ] we then show how to calculate the raman response function using a diagrammatic approach .
we first calculate the leading order aslamazov larkin diagram , and show that there is no response in the @xmath8 channel , and a finite response in the @xmath0 and the @xmath9 channels .
we then calculate higher order diagrams that take into account collisions between spin - fluctuations . finally , after summing an infinite number of these higher - order diagrams within a controlled @xmath23 expansion , we show i ) that the maximum of the raman response function in the @xmath0 channel occurs when the nematic susceptibility diverges , i.e. at the structural phase transition , and ii ) that the amplitude of the raman response function in the @xmath9 response gets suppressed .
we present our conclusions in sec .
[ conclusion ] .
two different approaches have been proposed in order to explain the origin of nematic phase in pnictides and its relation to the magnetic phase the orbital scenario @xcite and the spin - driven nematic scenario .
@xcite for a discussion of these approaches see for example ref . .
here we follow the approach of a spin - driven nematic state . in this scenario , the nematic phase is stabilized by magnetic fluctuations that are associated with the stripe density wave ( sdw ) phase .
the order parameter of the sdw state@xcite can be characterized by an @xmath24 manifold @xcite @xmath25 is the spin - rotational symmetry and @xmath26 a discrete symmetry associated with the choice of the ordering wave - vector , @xmath27 or @xmath28 .
let the two order parameters associated with these two ordering wave vectors be @xmath29 and @xmath30 respectively .
the sdw state is characterized by broken @xmath25 and @xmath26 symmetries . on the mean - field level
the breaking of @xmath26 and @xmath25 symmetry occurs simultaneously .
however , when one includes fluctuations , these transitions can be split . in case of joint transitions
, they are usually both first order transitions.@xcite the criterion for breaking the discrete @xmath26 symmetry via a second order transition is a threshold value of the magnetic correlation length @xmath31 . decreasing the temperature leads to an increase of @xmath31 . before the correlation length diverges at the magnetic phase transition temperature , the threshold value will be reached and spin - driven nematicity sets in .
this naturally explains why the magnetic and structural phase boundaries are correlated and leads to an intermediate phase with @xmath26 symmetry breaking without @xmath25 symmetry breaking .
this intermediate state is the nematic phase in the pnictides .
it is characterized by unequal strength of the magnetic fluctuations associated with the ordering wave vectors @xmath32 and @xmath33 : @xmath34 , but no long range magnetic order , @xmath35 . band , and the electron - like @xmath36 and @xmath37 bands , shifted by @xmath27 and @xmath28 , respectively . ] in what follows we will summarize the steps of ref . and outline the mathematical model the for spin - driven nematic phase .
we start from a simplified itinerant model where we include the bands near the @xmath38point and the @xmath39 and @xmath40 points in the brillouin zone . for our main result
no explicit knowledge of the detailed parametrization of the band structure is necessary , except for the fact that the band - structure is not perfectly nested . however , in order to obtain explicit numerical results we use the simplified model of ref . .
we consider parabolic dispersions with @xmath41 where @xmath42 are the band masses , @xmath43 is the offset energy , and @xmath44 denotes the chemical potential .
the corresponding fermi surfaces are shown in fig.[bandstructure ] . in order to study the established stripe magnetic phase
, we consider the hamiltonian that contains the interactions in the spin channel with momenta near @xmath45 and @xmath46 : @xmath47 here , @xmath48 is the creation operator of an electron with momentum @xmath49 , spin @xmath50 and in the band @xmath51 .
the spin operator is given by @xmath52 where @xmath53 denotes the @xmath22 component vector of the generators of the @xmath54 algebra .
in the case @xmath55 it holds @xmath56 with vector of the pauli matrixes @xmath57 .
@xmath58 is the coupling in the spin channel , which can be expressed in terms of density - density and pair - hopping interactions between hole and electron pockets .
@xcite the partition function is given by @xmath59 where @xmath60 is the inverse temperature .
since , the interaction hamiltonian is quadratic in the fermionic spin @xmath61 , we can decouple it using a hubbard - stratonovich decoupling in the spin - channel .
this way , we eliminate the quartic interaction between fermions at the expense of a functional integral over two additional bosonic fields @xmath62 and @xmath63 , with @xmath22 components .
the bosonic fields couple linearly to the corresponding fermionic spin densities . after introducing the spinor @xmath64 where @xmath50 denotes every possible value of the @xmath54 spin index , we can write the partition function as : @xmath65 } , \label{z}\ ] ] with the action : @xmath66=-\int_{k}\psi_{k}^{\dagger}\mathcal{g}_{\delta , k}^{-1}\psi_{k}^{}+\frac{2}{u_{\mathrm{s}}}\int_{x}\left(\boldsymbol{\delta}_{x}^{2}+\boldsymbol{\delta}_{y}^{2}\right ) .
\label{s}\ ] ] here , the matrix of the inverse green s function @xmath67 is given by : @xmath68 with the bare term : @xmath69 and the interacting term : @xmath70 @xmath71 with @xmath72 and @xmath73 unit matrix @xmath74 .
we invert the matrix equation ( [ g ] ) by expanding the geometric series and obtain the following expression for @xmath75 that we will use later - on : @xmath76 in this section , we first show how to obtain the ginzburg - landau expansion of the effective action in powers of the spin fluctuation fields @xmath77 in the limit of large @xmath5 , in the spirit similar to that of ref . where only @xmath55 was considered .
next , we re - formulate this effective action in terms of the collective nematic ising variable @xmath78 , and analyze the equation of state for @xmath78 .
we deduce the condition for the onset of the nematic phase by examining the susceptibility of the nematic order parameter .
we begin by integrating out the fermionic degrees of freedom from eq .
( [ z ] ) .
it follows : @xmath79}\ ] ] with action : @xmath80 & = & -\mathrm{tr}\ln\left(1-\mathcal{g}_{0}\mathcal{v}_{\delta}\right ) \nonumber \\ & & + \frac{2}{u_{\mathrm{s}}}\int_{x}\left(\boldsymbol{\delta}_{x}+\boldsymbol{\delta}_{y}^{2}\right)\label{effective_s1}.\end{aligned}\ ] ] here , @xmath81 refers to sum over momentum , frequency , spin , and band indices .
we further expand in powers of @xmath77 to obtain : @xmath80 & = & \frac{1}{2}\mathrm{tr}\left(\mathcal{g}_{0,k}\mathcal{v}_{\delta}\right)^{2}+\frac{1}{4}\mathrm{tr}\left(\mathcal{g}_{0,k}\mathcal{v}_{\delta}\right)^{4}\nonumber \\ & & + \frac{2}{u_{\mathrm{s}}}\int_{x}\left(\boldsymbol{\delta}_{x}^{2}+\boldsymbol{\delta}_{y}^{2}\right ) .
\label{effective_s2 } \end{aligned}\ ] ] after using a series of identities for the generators of the @xmath54 algebra , needed to evaluate the above traces ( for details see [ eff_action_derivation ] ) , we arrive at the following effective action in the large @xmath5 limit : @xmath82=\sum_{i}r_{0,i}\delta_{i}^{2}+\sum_{i , j}u_{ij}\delta_{i}^{2}\delta_{j}^{2 } , \label{effective_s3}\ ] ] with the coefficients : @xmath83 we used the notation @xmath84 .
the index @xmath85 combines the momentum @xmath49 and the matsubara frequency @xmath86 . using the identities @xmath87 valid because the underlying hamiltonian obeys the full c@xmath10 symmetry , we can write the action in the more convenient form : @xmath88 & = & r_{0}({\mathbf{\delta}}_{x}^{2}+{\mathbf{\delta}}_{y}^{2})+\frac{u}{2}({\mathbf{\delta}}_{x}^{2}+{\mathbf{\delta}}_{y}^{2})^{2 } \nonumber \\ & & -\frac{g}{2}({\mathbf{\delta}}_{x}^{2}-{\mathbf{\delta}}_{y}^{2})^{2 } , \label{action_ug1}\end{aligned}\ ] ] with coefficients @xmath89 @xmath90 , @xmath91 and @xmath92
have been calculated as a function of temperature and band parameters in ref . .
it was found that @xmath93 and @xmath94 in general .
the coupling @xmath92 vanishes for circular electron pockets , but is positive for a non - zero ellipticity . in order to investigate the possibility of the nematic transition occuring before the magnetic transition
, we follow the steps of ref . , and introduce two auxiliarly hubbard - stratonovich scalar fields @xmath78 and @xmath95 to decouple the quartic terms in the action ( [ action_ug1 ] ) ; @xmath96 and @xmath97 .
the resulting effective action is given by @xmath98 and we have added a field @xmath99 conjugate to the nematic order parameter @xmath100 .
this term is needed in order to calculate the susceptibility of the nematic order parameter .
a finite value of @xmath78 implies non - zero expectation value of @xmath101 and the system develops nematic order .
the large-@xmath5 mean field value of @xmath95 is always non - zero and describes the strength of magnetic fluctuations . in case of split magnetic and structural phase transitions ,
there is no magnetic order right below the structural transition temperature , i.e @xmath102 .
next we integrate out the @xmath22 component fields @xmath103 .
if we further rescale the coupling constants to @xmath104 and @xmath105 , required to reach a sensible large-@xmath5 limit , the effective action can be written as @xmath106&=&n^2\int_{q}\left\ { \frac{\phi^{2}}{2\tilde g}-\frac{\psi^{2}}{2 \tilde u } \right \ } \nonumber \\ & & + \frac{n^2}{2 } \int_{q}\left\{\log\left[\left(\chi_{q}^{-1}+ \psi \right)^{2}-\left ( \phi+h_{\mathrm{n}}\right ) ^{2}\right ] \right\}. \nonumber \\ \label{s_phi_n}\end{aligned}\ ] ] we note that the effective action ( [ s_phi_n ] ) has an overall pre - factor @xmath107 .
for @xmath108 the integral over the fields @xmath78 and @xmath95 can be performed via the saddle - point method , i.e. by analyzing the extremum of the action . after solving for @xmath109/\partial\phi=\partial
s_{\mathrm{eff}}\left[\phi,\psi\right]/\partial\psi=0 $ ] , we obtain the equations of state for @xmath78 and @xmath95 : @xmath110 by differentiating the second equation in ( [ phi_psi ] ) with respect to the conjugate field , we find that , for small @xmath78 @xmath111 where , from now on , we have shifted @xmath112 , which simply corresponds to the re - normalisation of the mass term due to fluctuations .
similarly to the result of ref . , we find that the electronic contribution to the susceptibility of the nematic order parameter @xmath113 is given by @xmath114 where @xmath115 is the inverse magnetic susceptibility , and @xmath116 is the nematic coupling constant of the theory . in ref .
it was found that for the classical phase transition in @xmath117 and @xmath118 the nematic transition pre - empts the magnetic transition , i.e. the transition lines are split .
also , the nematic transition was found to be of second order .
this is the regime we are interested in .
what we have calculated so far is the purely electronic contribution to the nematic susceptibility .
one , however needs to include the effect of the lattice , as was pointed out in ref . .
the nemato - elastic coupling is given by the following hamiltonian @xmath119 where @xmath120 is the nemato - elastic coupling constant and @xmath121 the phonon displacement field .
the phonons renormalize the nematic coupling constant to a frequency and momentum dependent coupling @xmath122 where @xmath21 is the elastic constant and @xmath13 the momentum along the soft directions . in particular ,
if one wants to determine the location of the nematic phase transition , which is dictated by the condition of divergent nematic susceptibility , one needs to look at the static limit of the coupling constant , i.e. the limit where @xmath12 is set to zero .
this leads to @xmath123 .
the full nematic susceptibility , including the effect of the coupling to the lattice , in the large @xmath5 expansion is therefore given by @xmath124 where @xmath125
raman scattering is a valuable tool to study strongly correlated electronic systems @xcite , since it probes lattice , spin and electronic degrees of freedom .
it has been used to extract information about the momentum structure and symmetry of the excitations in the cuprates @xcite and pnictides .
the differential photon scattering cross section in raman spectroscopy is directly proportional to the structure factor @xmath126 : @xmath127{\rm i m } r_q , \label{sf}\end{aligned}\ ] ] which is related to the imaginary part to the raman response function @xmath128 through the fluctuation - dissipation theorem .
@xcite here , @xmath129 is the bose - einstein distribution function , and @xmath130 .
since the momentum of light is much smaller than the typical lattice momentum , one normally replaces @xmath131 in eq .
( [ sf ] ) .
the raman response functions measures correlations between `` effective charge density '' fluctuations @xmath132 , @xmath133 the effective density , weighted by the form factors that can be changed via the geometry of the photon polarization , is defined as @xmath134 @xmath135 is the spin index , @xmath51 the band index , and the operator @xmath136 creates an electron with spin @xmath135 and momentum @xmath137 in band @xmath51 , where @xmath138 .
the function @xmath139 is related to the incident and scattered photon polarization vectors and depends on the curvature of the bands .
@xcite here , we will consider the intra - orbital contributions to @xmath140 only , as this is the dominant process .
the multi - orbital nature of different bands has been pointed out in ref . .
in order to determine the raman response function , we couple an external source field to the weighted densities and introduce the generating functional @xmath141 according to : @xmath142-\psi^\dagger \mathcal{v}_{h } \psi } , \nonumber \\ z & = & \int d\delta_{i}d\psi\mathrm{e}^{-s\left[\psi,\delta_{i}\right ] } , \label{wh}\end{aligned}\ ] ] where @xmath143 $ ] is given in eq .
( [ s ] ) .
the elements of the matrix @xmath144 in momentum / frequency , spin and band space are @xmath145 with @xmath146 being the field conjugate to the effective density .
the raman response function ( [ r_densities ] ) is obtained by differentiating the generating functional @xmath141 ( [ wh ] ) with respect to the conjugate field @xmath146 : @xmath147 due to the single particle character of the source term , the generating functional eq .
( [ wh ] ) can be written in the form @xmath148 since @xmath141 contains the action that is quadratic in fermions , we integrate out the fermions and obtain : @xmath149 } , \nonumber \\ s_{h } \left [ \delta_{i}\right ] & = & \frac{2}{u_{\mathrm{s } } } \int_{x}\left(\boldsymbol{\delta}_{x}^{2}+\boldsymbol{\delta}_{y}^{2}\right)-\mathrm{tr}\ln \left ( \mathcal{g}^{-1}_{\delta , h}\right ) .
\label{g_delta_h}\end{aligned}\ ] ] we further expand : @xmath150 then , using ( [ g_delta_h ] ) and ( [ rh ] ) , @xmath151 } \nonumber \\ & & \frac{\delta^2}{\delta h_{q } \delta h_{-q } } \left .
\exp { \left [ -\mathrm{tr}\left ( \mathcal{g}_{\delta}\mathcal{v}_{h}\right ) -\frac{1}{2}\mathrm{tr}\left ( \mathcal{g}_{\delta}\mathcal{v}_{h}\right ) ^2\right ] } \right|_{h=0}. \nonumber \\ \label{rgamma}\end{aligned}\ ] ] here @xmath152 = s_{h}[\delta_i]\vert_{h=0}$ ] is the effective action given by ( [ effective_s2 ] ) .
we define the matrix @xmath153 next , we analyse the leading order contributions to the raman response function . these arise from the self - energy and vertex correction diagrams depicted in fig .
[ selfenergy ] . both of these diagrams arise from differentiating the second term in the exponential ( [ rgamma ] ) twice with respect to @xmath146 @xmath154 } \mathrm{tr}\left [ \left ( \mathcal{g}_{\delta } \gamma \right ) ^2 \right ] , \label{vs}\end{aligned}\ ] ] and we replace @xmath155 , where @xmath156 is the quadratic action given by @xmath157=\frac{2}{u_{\mathrm{s}}}\int \left(\delta_{x}^{2}+\delta_{y}^{2}\right ) + \frac{1}{2 } \mathrm{tr}\left ( \mathcal{g}_{0}\mathcal{v}_{\delta}\right ) ^2 . \label{quadraticaction}\end{aligned}\ ] ] in order to get the vertex correction , we replace both @xmath75 in ( [ vs ] ) by @xmath158 , which comes from the perturbative expansion of eq .
( [ invgdelta ] ) : @xmath159 } \mathrm{tr}\left [ \left ( \mathcal{g}_{0 } \mathcal{v}_{\delta } \mathcal{g}_{0 } \gamma
\right ) ^2 \right ] . \label{v}\end{aligned}\ ] ] in order to get the self - energy correction , we replace one of @xmath75 in ( [ vs ] ) by @xmath160 , and the other one by @xmath161 to get @xmath162 } \mathrm{tr}\left [ \mathcal{g}_{0 } \gamma \left ( \mathcal{g}_{0 } \mathcal{v}_{\delta } \right ) ^2 \mathcal{g}_{0 } \gamma \right ] .
\nonumber \\ \label{v}\end{aligned}\ ] ] due to the integral over the square of the @xmath139 factor , the self - energy and vertex corrections occur in all symmetry channels .
if one evaluates the sum @xmath163 explicitly , in the hot - spot approximation , one finds that there are partial cancellation in the @xmath9 and in the @xmath0 channels , and no cancellations in the @xmath8 channel .
one can easily show that in @xmath117 @xmath164 where we have used @xmath165 , where @xmath31 is the magnetic correlation length .
, in @xmath117 . ]
the aslamazov - larkin contribution to the raman response function , analyzed in ref . arises from differentiating the first term inside the exponential in ( [ rgamma ] ) twice , and from replacing @xmath161 , which comes from the perturbative expansion of eq .
( [ invgdelta ] ) : @xmath166 } \left [ \mathrm{tr}\left ( \left ( \mathcal{g}_{0 } \mathcal{v}_{\delta } \right ) ^2 \mathcal{g}_{0}\gamma
\right ) \right ] ^{2}. \label{diagramatics_al } \nonumber \\\end{aligned}\ ] ] here @xmath152 = s_{h}[\delta_i]\vert_{h=0}$ ] is the effective action given by ( [ effective_s2 ] ) . as we will see below ,
the key assumption of a description based on the aslamasov - larkin diagrams is that one neglects the interactions between spin fluctuations . in other words ,
one approximates the effective action in ( [ diagramatics_al ] ) by quadratic action eq .
( [ quadraticaction ] ) .
while this assumption is frequently justified , it is not allowed in the theory of spin - driven nematicity , as we will show later .
the leading order aslamazov - larkin diagram , depicted in fig .
[ leading ] can be calculated as @xmath167 with @xmath168 similar to what was found in ref . . in the concept of the pairing symmetry in high - temperature superconductors successful theoretical models supported by experiments
have been developed in order to explain the symmetry sensitivity of the raman response function .
@xcite similarly , here , before we explicitly evaluate the leading order aslamazov - larkin diagram , we analyze the contribution to it in the various symmetry channels .
higher order corrections that will be discussed later do not alter this symmetry - based analysis .
we show that the aslamazov - larkin diagram , given by eq .
( [ leading_al ] ) and ( [ lambda ] ) , only supports the @xmath0 and the @xmath9 symmetry channels .
let us consider the structure of the terms in ( [ leading_al ] ) which arise from @xmath169 ^ 2 \nonumber \\ & & \times \chi ( { \mathbf{q}},\omega_{n } ) \chi ( { \mathbf{q } } , \omega_{n}-\omega ) .
\label{r11}\end{aligned}\ ] ] the term ( [ r11 ] ) can be re - written in the following form @xmath170 where we have classified the appropriate combinations of green s functions according to their symmetry into @xmath171 and we have defined @xmath172 . from eq .
( [ e ] ) , we see that the response will be non - zero only for @xmath173 factors in the @xmath9 or the @xmath0 symmetry . similarly , using the same line of arguments , one can show that all other terms in ( [ leading_al ] ) support the @xmath9 or the @xmath0 symmetry only .
we have thus ruled out the response in the @xmath8 channel .
the leading order aslamazov - larkin diagram has been evaluated in ref . , assuming that the main contribution comes from the hot - spot regions and that the momenta of the fluctuations are peaked around @xmath174 . after the analytic continuation to the real frequencies
, we found that the imaginary part of the raman response function , which is a quantity of experimental interest , is given by @xmath175 \nonumber \\ & & \times \int_{{\mathbf{q}}}\mathrm{im}\left [ \chi^{r}(\epsilon,{\mathbf{q}})\right ] \mathrm{im}\left [ \chi^{r}(\epsilon+\omega,{\mathbf{q } } ) \right ] , \nonumber \\ \label{r0_integral}\end{aligned}\ ] ] with the spin propagator in the tetragonal phase given by : @xmath176 where @xmath177 tunes the distance from the magnetic transition , see eq . ( [ action_ug ] )
. in @xmath117 the @xmath178 integral in eq .
( [ r0_integral ] ) in can be performed exactly , which leads to the following expression @xmath179_{d=2 } = & & \int_{0}^{\infty}{\textrm{d}}\epsilon \left [ n(\epsilon_{+})-n(\epsilon_{- } ) \right ] \frac{\epsilon_{+}\epsilon_{-}}{\epsilon_{+}^{2}-\epsilon_{-}^{2 } } \nonumber \\ & & \times \left [ f(\epsilon_{+})-f(\epsilon_{- } ) \right ] , \label{2draman}\end{aligned}\ ] ] with @xmath180 we defined @xmath181 .
the plot of the function ( [ 2draman ] ) is shown in fig . [ leading ] .
in particular one can deduce that , in the regime where temperature @xmath182 is the biggest scale , @xmath183 , @xmath184 for small frequencies @xmath12 , while the amplitude of the raman response function scales as @xmath185 in this regime . in summary
, we have shown that the leading order aslamazov - larkin diagram gives a non - zero response in the @xmath0 and @xmath9 symmetries only .
it predicts the divergence of the raman response at the magnetic transition , and does not carry any signatures of the structural transition .
we therefore need to go beyond the leading order aslamazov - larkin diagram .
next , we go beyond the quadratic action approximation for @xmath186 in ( [ diagramatics_al ] ) , and include the full quartic action to evaluate the raman response function .
as we will show , diagrammatically this corresponds to inserting a series of fermionic boxes that resemble the structure of the nematic coupling constant @xmath16 into the leading order aslamazov - larkin diagram in the @xmath0 symmetry .
these diagrams take into account the collisions between spin fluctuations which were not accounted for in the leading order aslamazov - larkin diagram .
first we show how these terms arise from the diagrammatic expansion .
we start from eq .
( [ diagramatics_al ] ) , but this time we go beyond the quadratic approximation for the effective action , and include quartic terms @xmath187 } \left [ \mathrm{tr}\left ( \left ( \mathcal{g}_{0 } \mathcal{v}_{\delta } \right ) ^2
\mathcal{g}_{0}\gamma \right ) \right ] ^{2 } \label{s0}\end{aligned}\ ] ] where @xmath188 & = & s_{0}\left[\boldsymbol{\delta}_{i } \right ] + \frac{1}{4}\mathrm{tr } \left ( \mathcal{g}_{0}\mathcal{v}_{\delta } \right ) ^{4},\end{aligned}\ ] ] with the bare action @xmath189 & = & \frac{2}{u_{\mathrm{s}}}\int_{x}\left(\boldsymbol{\delta}_{x}^{2}+\boldsymbol{\delta}_{y}^{2}\right)+\frac{1}{2}\mathrm{tr } \left ( \mathcal{g}_{0}\mathcal{v}_{\delta } \right ) ^{2}.\end{aligned}\ ] ] we further expand the exponential @xmath190^{m}\ ] ] to obtain @xmath191 where we averaged the following terms with respect to the gaussian collective spin action : @xmath192^m \left [ \mathrm{tr}\left ( \left ( \mathcal{g}_{0 } \mathcal{v}_{\delta } \right ) ^2 \mathcal{g}_{0}\gamma \right ) \right ] ^{2 } \right \rangle_{s_0}. \label{r_renormalized } \nonumber \\\ ] ] in order to evaluate the expectation values one performs contractions of the @xmath193 fields .
we obtain a series of diagrams that look like the leading order aslamazov - larkin diagram with an arbitrary number of inserted fermionic boxes , depicted in fig .
[ resum ] .
the higher order diagrams effectively take collisions between spin fluctuations into account , which have been neglected in the leading order aslamazov - larkin diagram . as one approaches the transition line ,
collisions between spin fluctuations become more and more important and one would anticipate significant changes in the raman response function due to these processes .
as we will show , the re - summation of boxed aslamazov - larkin diagrams will lead to the maximum of the raman response function at the structural phase transition in the @xmath0 channel , and the suppression of the response in the @xmath9 channel .
the next task is to re - sum an infinite number of such diagrams .
every box can be characterized by two indices : the first one denotes the type of incoming spin fluctuations , this can be either @xmath36 or @xmath37 and the second one denotes the type of exiting spin fluctuation .
let us denote this box @xmath194 .
is shaded grey
. the first index of the matrix @xmath195 denotes the type @xmath196 of entering spin fluctuations , and the second index the type of exiting spin fluctuations .
we insert the grey shaded box into the raman reponse , and make some further simplifications to evaluate the raman response function ( see the main text ) . ]
summing all boxed diagrams can be most efficiently expressed as : @xmath197 for our analysis it is sufficient to calculate the box @xmath194 at momenta @xmath198 and zero frequencies , which is justified for small incoming raman frequency @xmath12 , and in the classical regime relevant near a finite temperature phase transition .
we write the raman reponse function in the tetragonal phase : @xmath199 where @xmath200 is the leading order diagram .
the symmetry of the fermionic triangle is such that @xmath201 this relation can be obtained by simply performing a coordinate system rotation by @xmath202 inside the momenta integrals in ( [ lambda ] ) .
this allows us to explicitly perform the matrix multiplication , which yields : @xmath203 next we need to determine an expression for the full box @xmath204 , i.e. perform a sum over the leading box - diagrams within the @xmath23 expansion .
this is illustrated in fig .
[ resum ] and can be written as : @xmath205 the matrix @xmath195 was deduced from eq .
( [ r_renormalized ] ) and eq .
( [ tm ] ) . for details about explicit evaluation of the @xmath54 trace pre - factor ( which arises from contractions of products of @xmath206 matrices in ( [ v_delta ] ) ) for boxed diagram containing arbitrary number of boxes @xmath207 , please see appendix [ trace_prefactor ] .
the matrix @xmath195 of irreducible boxes is then given as @xmath208 where we used the abbreviation @xmath209 and used that by symmetry : @xmath210 .
the @xmath207th power of the matrix @xmath195 is given by @xmath211 where @xmath212 . from this analysis
follows that @xmath213 where @xmath214 is precisely the nematic coupling constant of eq .
( [ action_ug ] ) for the effective action , and @xmath215 is the other quartic term in eq .
( [ action_ug ] ) , with @xmath216 , as found in ref . from ( [ raman_final ] )
, we see that the raman response in the @xmath9 channel gets suppressed , due to the term in the denominator , which grows as one approaches the transition . on the other hand , in the @xmath0 channel , after performing the analytic continuation to real frequencies and taking the imaginary part , we get that @xmath217 \left ( 1 + \tilde g { \chi}_{\mathrm{nem}}^{\mathrm{el } } \right ) , \label{semifinal}\end{aligned}\ ] ] where @xmath218 is the electronic contribution to the nematic susceptibility calculated in the large @xmath5 limit @xcite for the model described in sec .
[ eff action ] . as was pointed out in ref .
, the enhancement of the static nematic coupling constant ( [ gstat ] ) does not enter the raman response , due to the fact that the raman response operates in the dynamical limit ( @xmath219 and finite @xmath12 ) , and the static and dynamic limits do not commute @xcite . at the nematic / structural phase transition the nematic susceptibility ( [ nematic_susc ] ) diverges , and @xmath220 consequently the raman response function in the @xmath0 channel , given by ( [ semifinal ] ) , has a maximum rather than a divergence at the structural phase transition .
in summary , we have shown that the raman scattering can be used as a tool to probe the nematic phase in pnictides .
we have presented a calculation that demonstrates that , in the low - frequency limit , and large @xmath5 limit , the raman response function shows a clear maximum at the structural transition temperature in the @xmath0 channel . in our model , the electronic nematic phase in pnictides is stabilized by spin - fluctuations associated with the striped phase , and occurs as a thin sliver above the magnetic transition temperature . in order to calculate the raman response function , we have gone beyond the leading order aslamazov - larkin diagram , and included higher order diagrams that contain a series of quartic paramagnon couplings , mediated by electronic excitations .
such quartic couplings contain a product of four fermionic green s functions and include the effect of collisions between spin fluctuations .
when re - summed these diagrams lead to the maximum of the electronic raman response function at the structural transition in the @xmath0 channel , and the suppression of the response in the @xmath9 channel .
the method that we developed analysed the raman response function only in the regime of small frequencies .
it would be desirable to extend it to the entire frequency range , such that one can analyse the entire shape of the raman response function as a function of temperature , and possibly be able to extract some information about the dynamical nematic susceptibility .
further , one might expect a charge driven nematic phase to have similar signatures in the raman response function
. this could be relevant to the peculiar case of fese , where the nematic phase has been detected , but no magnetic phase has been seen .
@xcite in order to do so , we would need to develop a theoretical method that goes beyond the large @xmath5 expansion .
we acknowledge useful discussions with a. chubukov and r. fernandes . u.k . acknowledges the support from the helmholtz association , through helmholtz post - doctoral grant pd-075 `` unconventional order and superconductivity in pnictides '' .
, t .. b . and r.h .
acknowledge the support from deutsche forschungsgemeinschaft ( dfg ) through the priority program spp 1458 `` hochtemperatur - supraleitung in eisenpniktiden '' ( project - nos .
schm 1031/5 - 1 and ha2071/7 - 2 ) .
y.g . acknowledges financial support from anr grant pnictides
. + note : in the final stages of the preparations of the manuscript we became aware of arxiv:1504.05054 , where the behaviour of the raman response function in the vicinity of the structural transition has been analyzed . where there is overlap with this work , our results agree .
here , we present some useful identities for the structure constants of @xmath54 . they have been used to determine the scaling of the boxed aslamazov - larkin diagrams with @xmath5 , and to develop the ginzburg - landau expansion of the effective action in powers of spin fluctuation fields @xmath193 ( see sec . [ microscopics ] ) .
we begin by listing some standard @xmath54 identities for the matrices @xmath221 , where @xmath222 .
all repeated indices are summed over .
@xmath223 & = & i{\mathbf{f}}_{jkl}\boldsymbol{\lambda}_{l}\;\;\ ; { \mathbf{f}}_{jkl}=-{\mathbf{f}}_{kjl } , \label{com } \\ \boldsymbol{\lambda}_{j}\boldsymbol{\lambda}_{k } & = & \frac{1}{2n}\delta_{jk}+\frac{1}{2}{\mathbf{r}}_{jkl}\boldsymbol{\lambda}^{l } , \label{a } \\ { \mathbf{r}}_{jkl } & \coloneqq & { \mathbf{d}}_{jkl}+ i{\mathbf{f}}_{jkl}. \label{rdef}\end{aligned}\ ] ] here @xmath224 is symmetric under the exchange of its indices , while @xmath225 is antisymmetric under the exchange of neighbouring indices .
further , some useful relations for the summations of structure constants can be derived , @xcite which read @xmath226 useful identities that involve the traces of the @xmath54 matrices are @xmath227 in order to analyse the trace of the product of four @xmath54 generators we evaluate @xmath228\nonumber \\ & = & \frac{1}{4n}\delta_{ij}\delta_{kl}+\frac{1}{8}{\mathbf{r}}_{ijp}{\mathbf{r}}_{klp } , \label{trace4}\end{aligned}\ ] ] where we used the identity ( [ a ] ) in the first line , as well as eq .
( [ b ] ) and eq .
( [ trace2 ] ) in the second line .
these results will be of importance for the subsequent analysis of higher order diagrams .
first we calculate the quadratic terms in the free energy expansion .
this is given by @xmath229 where @xmath196 and we used the identity ( [ trace2 ] ) . next we calculate the quartic term in the free energy expansion @xmath230 with @xmath231 and we used the notation @xmath232 for not @xmath50 , i.e. if @xmath233 then @xmath234 and vice versa .
we further substitute the identity ( [ trace4 ] ) in ( [ quartic_0 ] ) , to write @xmath235 where @xmath236 since @xmath237 , while @xmath238 , the term @xmath239 can be omitted in the large @xmath5 limit . combining ( [ k1k2 ] ) and ( [ quadratic_action ] ) ,
the effective action in the large @xmath5 limit can be written as @xmath82=\sum_{i}r_{0,i}\delta_{i}^{2}+\sum_{i , j}u_{ij}\delta_{i}^{2}\delta_{j}^{2 } , \label{effective_s3}\ ] ] with the coefficients : @xmath83 we note that in the large @xmath5 approximation there are no @xmath240 terms in the action ; however if one considers corrections to large @xmath5 these terms might appear in the effective action .
in this appendix we derive further identities for the traces of the @xmath54 generators , which have been used to deduce the dependence of the aslamazov - larkin boxed diagrams on @xmath5 .
in particular , we would like to calculate @xmath241 we begin by considering @xmath242 .
written out explicitly , it follows : @xmath243 where we have used ( [ trace2 ] ) and ( [ trace4 ] ) to get to the second line , and the fact that @xmath244 in the penultimate line , which is a consequence of ( [ d ] ) and the antisymmetry of @xmath245 . using the same set of identities , we find that @xmath246 similarly , one can deduce that @xmath247 | we establish a relation between the raman response function in the @xmath0 channel and the electronic contribution to the nematic susceptibility within the spin - driven approach to electron nematicity of the iron based superconductors .
the spin - driven nematic phase , characterized by the broken @xmath1 symmetry , but unbroken @xmath2 ) spin - rotational symmetry , is generated by the presence of magnetic fluctuations associated with the striped phase .
it occurs as a separate phase between @xmath3 and @xmath4 in systems where the structural and magnetic phase transitions are separated .
detecting the presence of nematic degrees of freedom in iron - based superconductors is a difficult task , since it involves measuring higher order spin correlation functions .
we show that the nematic degrees of freedom manifest themselves in the experimentally measurable raman response function .
we calculate the raman response function in the tetragonal phase in the large @xmath5 limit by considering higher - order aslamazov - larkin type of diagrams .
they are characterized by a series of inserted quartic paramagnon couplings mediated by electronic excitations that resemble the nematic coupling constant of the theory
. these diagrams effectively account for collisions between spin fluctuations . by summing an infinite number of such higher order diagrams
, we demonstrate that the electronic raman response function shows a clear maximum at the structural phase transition in the @xmath0 channel .
hence , the raman response function can be used to probe nematic degrees of freedom . |
carbon - based nanostructured composites ( n@xmath0-c ) @xcite are believed to intermingle properties of carbon nanostructures with those of single - phase amorphous carbon ( @xmath0-c)@xcite .
these nanocomposites are hybrid forms of carbon in which nanocrystallites are embedded in the @xmath0-c matrix . in this sense ,
their properties can be tailored by controlling the type and size of the embedded structures , as well as the matrix material .
such composites , also investigated in this work , are relevant to both bulk and thin film materials .
a variety of carbon nanostructures can compose n@xmath0-c , ranging from diamond crystallites , to entirely three - dimensional sp@xmath1 covalent conformations with no dispersion / van der waals ( vdw ) bonding , i.e. open graphene structures with negative curvature ( schwarzites ) @xcite , to carbon nanotubes ( cnt ) @xcite bound by vdw forces .
n@xmath0-c films have been synthesized before by using different methods @xcite .
the synthesis of nanodiamond composites ( nd/@xmath0-c ) and its growth mechanisms have been explored in both hydrogenated @xcite and pure @xcite tetrahedral amorphous carbon ( t@xmath0-c ) matrices . here , we review our recent theoretical and simulational studies of nanostructured carbon films , which shed light on their mechanical and optoelectronic properties .
this review is organized as follows : section [ sec : meth ] outlines the methodology followed here .
section [ sec : ene ] discusses the energetics and stability of all modeled nanocomposites .
the elastic and mechanical properties of diamond and nanotube composites are investigated in section [ sec : elast ] . the ideal fracture and optoelectronic properties of diamond nanocomposites
are discussed in sections [ sec : fract ] , [ sec : optoel ] .
we summarize in section [ sec : concl ] .
we have recently followed a theoretical approach to investigate n@xmath0-c and unravel the fundamental principles governing the interaction of nanostructures with the @xmath0-c matrix ( see for example @xcite ) .
two different methods were used , a monte carlo ( mc ) method within the tersoff empirical potential approach @xcite , and a tight - binding molecular dynamics ( tbmd ) scheme which utilizes separately two hamiltonians , the mehl & papaconstantopoulos @xcite ( tbnrl ) , and the environment - dependent ( edtb ) wang & ho @xcite model .
the former employs non - orthogonal atomic orbitals , while the latter goes beyond the traditional two - center approximation and allows the tight - binding parameters to change according to the bonding environment . using both the mc and tbmd ,
one is able to combine higher statistical precision with higher accuracy and bridge the gap between classical and first - principles calculations .
both the tersoff potential , as well as the two hamiltonians have been used to successfully model the interatomic interactions in various crystalline and amorphous carbon forms @xcite the tbmd simulations are carried out in the canonical ( @xmath2 ) ensemble . the volume @xmath3 and the number of particles ( atoms ) @xmath4 are constant
. dispersion / van der waals ( vdw ) forces within the tersoff potential approach , will be included by adding a simple lennard - jones potential , which successfully describes properties of carbon nanocrystals @xcite .
the parameters used for this potential are @xmath5 and @xmath6 . for the calculation of the optoelectronic properties
the edtb scheme was used , which has proved to be quite efficient in similar cases @xcite .
the electronic structure can be easily extracted and is directly correlated to the imaginary part @xmath7 of the complex dielectric function .
this is in turn yields the optical gaps and absorption coefficient , which further link to the disorder in the structures through the urbach energy .
additional details can be found elsewhere @xcite .
we have generated the nanocomposites studied here by melting and subsequent quenching at constant volume various carbon crystalline cells , while keeping a certain number of atoms in the central portion of the cells frozen in their ideal crystal positions .
consequently , the cells are thoroughly relaxed with respect to atomic positions and density . for the nanotube composites ,
a core is removed from a diamond crystal leaving a void with a shape similar to that of the nanotube , in which the nanotube is then placed .
details can be found in ref .
@xcite .
the total number of atoms in the structures ranges from @xmath8 2500 to 4536 , and the number of atoms in the nanocrystals ranges from @xmath8 300 to 900 .
the size of the computational cells ranges from about 1.5 to 4.0 nm . for the tbmd relaxed structures we use smaller cells up to 512 atoms .
the diameter ( d ) of all crystalline inclusions used are in the range d=0.6 - 2.8 nm .
periodic boundary conditions ( pbc ) are applied to the cells .
the exact details for each composite studied with the tersoff potential are summarized in table [ tab : struc ] .
diamond inclusions are almost spherical , while cnts have open ends and extend through the entire length of the cube .
due to the pbc , this corresponds to a dense array of cnts packed in parallel , which is an idealized model of a cnt nanocomposite .
however , this model captures all essential features of the cnt - matrix interaction resembles cnt bundles with @xmath0-c material in between the tubes .
representative structures of a specific group of conformations are chosen and embedded in amorphous matrices : the @xmath9 conformations ( a ) diamond ( d ) , and the ( b ) high - pressure phase bc8 structure for carbon , the negatively - curved @xmath10 conformations with a @xmath11 periodic minimal surface , ( c ) polybenzene ( 6.8@xmath12 ) @xcite , ( d ) pccm ( porous conducting carbon modification ) , an sp@xmath1-bonded structure with eightfold rings @xcite , ( e )
_ c@xmath13 _ , a low - density , negative - curvature schwarzite structure @xcite , ( f ) the positively curved @xmath10 carbon nanotubes ( cnt ) @xcite with single ( swnt ) and multiple ( mwnt ) graphitic walls .
structures ( c)-(e ) do not involve vdw bonding . in fig.[fig : strucs ] , representative nanocomposites of the ones studied here are shown : a diamond of 1.70 nm diameter , a @xmath14 , and a ( 7,10 ) swnt of 1.17 nm diameter , embedded in a highly tetrahedral ( @xmath15 3.8 , @xmath16 3.3 g @xmath17 ) and low density ( @xmath15 2.8 , @xmath16 1.5 g @xmath17 , and @xmath15 3.2 , @xmath16 2.2 g @xmath17 , respectively ) matrices .
the mean coordination of the amorphous matrix and the total density of the composites are denoted z@xmath18 and @xmath19 , respectively .
fig.[fig : strucs](b ) shows clearly , that no covalent bonding between cnt atoms and the matrix exists , as the vdw forces drive a reconstruction of the surrounding medium , producing a graphitic wall at a distance of 3.4 from the cnt .
we begin with the investigation of the stability of the nanocomposites studied here .
the interaction of the embedded configuration with the host is taken care of by defining the formation energy of a nanocrystal , given by @xmath20 in this equation , @xmath21 , @xmath22 are the total cohesive energies per atom of the composite system and the crystalline phase , respectively .
@xmath23 , @xmath24 are the number of atoms in the nanocrystal , and the amorphous matrix , respectively , and @xmath25 is the cohesive energy per atom of the pure amorphous phase ( without the nanocrystal ) with coordination @xmath26 @xcite .
a negative value of @xmath27 denotes stability of the nanostructure , a positive value indicates metastability .
the formation energies , hence stabilities , of all nanocomposites in this study are depicted in fig.[fig : eform ] . of the nanostructures
considered , only diamond and the c@xmath13 schwarzite are stable ( @xmath28 ) , the former at high @xmath26 ( @xmath29 ) , the latter at low @xmath26 ( @xmath30 ) .
both crystallites are stable in a matrix with structural characteristics similar to their own , which relates to a small density gradient between the matrix and the inclusion : diamonds(schwarzites ) are dense @xmath9(low density @xmath31 ) networks , thus are embedded in a dense(dilute ) matrix . a small density gradient does not give rise to discontinuities and thereby energetically favors the embedding of the crystalline inclusion .
overall , diamonds are more stable than schwarzites .
their enhanced stability in highly tetrahedral networks indicates that once nucleation centers are formed large diamond regions can be further grown , as shown in experiments @xcite . through annealing of the diamond composites , an unstable structure extensively shrinks , and only a small core remains intact , while the stable structure retains its shape and should expand .
metastable diamonds are on the average larger , but more deformed , than stable diamonds in t@xmath0-c . for a better nanostructured material it is preferable to nucleate diamond cores in a t@xmath0-c matrix .
all other nanostructures studied in this work are metastable ( @xmath32 ) through the whole coordination range .
their curves exhibit a well defined local minimum at which @xmath27 is quite small , suggesting that they can be maintained in the amorphous matrix under moderate conditions and their synthesis is possible .
the fact that @xmath27 for 6.8@xmath12 and pccm composites are only slightly lower than that for a diamond inclusion suggests the possibility of a phase transformation of @xmath31-bonded structures to diamond under the proper conditions of pressure and temperature .
this would subsequently aid the experiments towards synthesis of such theoretical carbon crystals .
more details are discussed in @xcite .
cnt composites are also metastable with minima in their @xmath33 at intermediate coordinations @xmath26 .
the elastic and mechanical properties of carbon composites are important for potential applications as ultra - hard protective or biomedical coatings .
we focus on diamond nanocomposites ( @xmath34d/@xmath0-c ) , which are the most stable among the ones we dealt with in this work , and the nanotube nanocomposites ( cnt/@xmath0-c ) , as a representative sp@xmath1 structure with positive curvature which is of wide technological interest .
we begin with the calculations of the bulk modulus ( @xmath35 ) , which was calculated by fitting the energy vs volume curve after a hydrostatic deformation to the birch - murnacham equation @xcite
[ fig : ndmodul](a ) , [ fig : swntmodul](a ) show the calculated moduli plotted as a function of the nanocomposite density for various diamond and nanotube diameters , respectively . the bulk modulus is significantly enhanced with increasing density and inclusion size .
replacement of some amorphous material by crystalline increases noticeably the bulk modulus , with respect to @xmath35 of @xmath0-c , in both diamond @xcite and nanotube composites ( fig.[fig : swntmodul](a ) ) .
@xmath35 for some samples is even higher than that of the _ amorphous diamond _ network ( www ) @xcite and close to that of diamond @xcite .
a similar enhancement of the hardness with the grain size was found in ultra - nanocrystalline diamond @xcite and copper films @xcite .
a comparison of @xmath35 for swnt and mwnt inclusions of the same diameters embedded in @xmath0-c of similar densities reveals , that the number of walls that consist a mwnt is crucial for the whole composite s rigidity ( see fig.[fig : swntmodul](b ) ) . by increasing the number of graphitic walls within a cnt
the composite becomes more rigid .
a check whether the rule of mixtures @xcite applies to diamond and cnt composites revealed that it indeed holds for the former , but not for the latter structures .
apparently , the cnt composites are significantly influenced by the presence of the matrix . in these structures , the microstructural details of the matrix
have been significantly altered with respect to pure @xmath0-c .
we note again , that the effect of the inclusion is to drive the surrounding matrix atoms further away from the nanotube and stabilize them at a distance of about 3.4 . on the other hand , no such a significant effect was evident in @xmath34d/@xmath0-c , as there is a smooth structural transition from the inclusion to the matrix .
in other words , there is a direct connection of the inclusion - matrix bonding and the validity of the rule of mixtures : in the nanotube composites the bonding between the matrix and the inclusion is of a van der waals type and the rule does not hold , while in the diamond composites the two phases are covalently bonded and this rule still holds .
the trends for the young s moduli ( @xmath36 ) , which is easily extracted from the elastic constants discussed next , of both composites are the same as in the case of @xmath35 . for @xmath34d/@xmath0-c , @xmath36
is shown in fig.[fig : ndmodul ] .
a considerable enhancement of @xmath36 was found , especially in high density matrices , with values approaching that for diamond @xcite . in the case of cnt/@xmath0-c
, a marked characteristic is the high elastic anisotropy which is monitored by the ratio @xmath37 where @xmath38 , @xmath39 are the components of @xmath36 in the direction of the tube ( axial ) and in the transverse directions . a composite with a swnt with diameter of 1.65 nm and a density of 1.99 g @xmath17 ( 2.14 g @xmath17 ) has @xmath38 = 478(564 ) gpa and @xmath39 = 231(306 ) gpa , yielding @xmath40 = 0.48(0.54 ) . the anisotropy increases with increasing tube diameter , as the tube contribution overwhelms the isotropic matrix part .
to supplement the calculation of @xmath35 , the remaining elastic moduli are computed .
all simulation cells of the diamond composites are cubic systems , thus only the knowledge of the @xmath41 and @xmath42 is needed .
this is not exactly the case for cnt composites , as these are modeled through tetragonal computational cells . to calculate the elastic constants
, we apply the appropriate deformation to the system and compute its total energy as a function of the imposed strain . the curvature of this function at its minimum yields the desired modulus .
the results are plotted in fig.[fig : ndmodul](b ) as a function of the density for diamond composites .
all elastic constants increase with increasing density .
this is also the case in cnt/@xmath0-c , for which representative values are summarized in table [ tab : modcnt ] .
for an isotropic material , @xmath43 which holds within 4% or less , for @xmath34d/@xmath0-c . for a typical cnt composite , @xmath44 , while for all pure @xmath0-c networks @xmath45 .
we have also looked at the elastic recovery during a compression - decompression cycle from 0 to 100 gpa for diamond and nanotube composites ( with inclusions of sizes 1.7 nm and 1.2 nm , respectively ) , as well as for amorphous structures @xcite .
the elastic recovery of @xmath34d/@xmath0-c and single - phase @xmath0-c is almost perfect in the highly tetrahedral regime ( @xmath46-c ) , but is not perfect as the @xmath9 fraction declines .
an almost fully @xmath31-bonded @xmath0-c network regains its initial volume at about 95 % .
a cnt composite , having a high sp@xmath1 fraction does not fully recover and shows a hysterisis in the compression curves .
our results indicate that the compressibility from cnt/@xmath0-c towards @xmath0-c and nd/@xmath0-c is decreasing , as long as the sp@xmath1 fraction remains high .
a study beyond the elastic regime gives an insight to the ideal fracture of @xmath0-c and @xmath34d/@xmath0-c .
tensile ( along diamond s easy slip \{111 } plane ) and shear ( on the \{111 } plane in a @xmath47112@xmath48 direction ) loads were applied and the stress versus strain curves were extracted .
the nrl hamiltonian at low temperature was used in this study .
this method has already been tested in the case of diamond and the findings compared well with previous _ ab initio _
results @xcite .
three @xmath0-c samples were considered ( a typical non - hydrogenated @xmath0-c with @xmath26=3.47 , a t@xmath0-c sample with @xmath26=3.78 and a www sample ) and a @xmath34d/@xmath0-c with an inclusion of about 1.2 nm embedded in t@xmath0-c with an sp@xmath1 fraction of about 30% .
the stress vs. strain curves in the case of the tensile load are summarized in fig.[fig : fract ] , left panel .
the maximum stresses are roughly 60 , 40 , 30 , and 124 gpa for @xmath26=4.00 , 3.78 , 3.47 , and diamond , respectively .
thus , the ideal strength of @xmath0-c is proportional to its sp@xmath49 ratio . the stress vs. strain curves for the low - density @xmath0-c ( @xmath50 fraction @xmath8 50 % ) sample suggest a ductile behavior .
the fracture of @xmath0-c , which has an ideal strength half that of t@xmath0-c , takes place at a higher load than t@xmath0-c , i.e. larger strain needs to be applied to break the @xmath0-c network .
similar are the results when a shear load is applied . for the diamond composite , the easy slip plane of the embedded diamond core was not found to play a crucial role in the total strength and no weak directions were present .
the crystalline phase remains unaffected by the external load , which is almost completely taken by the surrounding amorphous matrix .
the response , thus , of @xmath34d/@xmath0-c to external load beyond the elastic regime is identical to the response of the embedding matrix .
the onset of fracture takes place at the weakly bonded @xmath9 sites in the amorphous matrix @xcite .
bonds in the crystal are at this point actually elongated .
fracture in diamond nanocomposites is inter - grain and their ideal strength is similar to the pure amorphous phase .
a snapshot of a @xmath34d/@xmath0-c at the onset of fracture ( i.e. at the maximum stress ) is shown in fig.[fig : fract ] , right panel .
finally , focus is given on the electronic and optical properties of diamond nanocomposites .
we were able to probe at a local atomic level the optoelectronic response of the composite and link it to the associated disorder , which exists at the interface between the diamond inclusions and the embedding @xmath0-c matrix .
this can be quantitatively probed by extracting the urbach energies from the optical parameters .
disorder in the nanocrystals appears in their outer shell near the interface and is manifested as bond length and angle distortions . as representative quantities of an analysis done elsewhere @xcite , the electronic density of states ( edos ) and the imaginary part ( @xmath7 ) of the dielectric function for nd/@xmath0-c
are shown in fig.[fig : opt ] .
the same approach for the study of the dielectric response for diamond - like amorphous carbon has been used before @xcite .
the results compare qualitatively well with recent experiments @xcite , taking into account that in the latter case the diamond - like film was hydrogenated
. a first look at the edos in fig.[fig : opt ] reveals that the main contribution to the formation of the gap , i.e. the absence of electronic states , and of any states located in the gap region comes from the amorphous matrix part , as this has a lower energy gap than diamond due to the presence of sp@xmath1 hybrids . as the sp@xmath1 ratio , which controls the gap through the separation of @xmath51-@xmath52 bonding - antibonding states increases , the gap region shrinks . a smaller contribution to the gap states originates from distorted sp@xmath49 atoms of the diamond inclusion at the interface .
decomposition of the total @xmath7 into contributions from the atoms in the crystalline and amorphous regions showed , that the optical response of the nanocomposite system is progressively varied .
this variation occurs as the sp@xmath1 fraction in the amorphous part is increased from having nearly diamond - like nature to having nearly graphite - like nature .
it was also clear that the energy gaps , i.e. absence of electronic transitions at the edge of the spectra , are reduced as the sp@xmath1 fraction increases .
near the gap the main contribution comes from transitions between states localized in atoms of the amorphous part .
calculation of the urbach energy ( e@xmath53 ) in @xmath34d/@xmath0-c , which is linked directly to the optical response provides information about the disorder in the structure .
we have recently shown @xcite , that e@xmath53 is a good measure of disorder in @xmath0-c , as it reflects not only bond - length and bond - angle distortions ( structural disorder ) , but also topological disorder related to the size distribution of chains and clusters of sp@xmath1 and sp@xmath49 atoms in the mixed phase @xcite .
thus , it is a measure of the inhomogeneous disorder in the two - phase network .
for example , in highly tetrahedral t@xmath0-c , the larger the sp@xmath1 cluster sizes the higher the e@xmath53 values .
this is because the embedding of minority configurations into a host phase produces disorder , which is reflected in the @xmath51-@xmath52 subgap transitions .
we also showed @xcite that transitions between states localized on unpaired sp@xmath1 sites ( dangling bonds ) , or between such states and neighboring @xmath51 or @xmath54 states , are negligible .
this is in contrast to a - si : h , where dangling bond defects are the main contributors to the urbach edge .
e@xmath53 for a typical composite is shown in fig.[fig : urbach ] and underlines a nonmonotonic behavior for both the composite and pure @xmath0-c .
the variation of e@xmath53 in the case of @xmath34d/@xmath0-c comes from the atoms in the matrix , but with higher values at the maximum of the variation and at the tetrahedral region , indicating the excess disorder produced on these atoms .
we have also broken e@xmath53 down into atomic components by calculating separately the absorption coefficient ( @xmath55 ) contributed by the atoms close to the diamond nucleus and close to the interface .
results shown in fig.[fig : urbach ] indicate that the variation in @xmath55 nearly coincides with the contribution from the atoms at the interface , while the absorption from the nucleus approaches the absorption of crystalline diamond .
the atoms at the interface determine the e@xmath53 for the whole composite .
defects at the interface ( unpaired sp@xmath1 sites and dangling bonds ) contribute negligibly to the optical transitions and the urbach edge , as in pure @xmath0-c @xcite . on the other hand , @xmath56 for
the diamond nanoinclusions takes quite high values ( @xmath8 0.6 ) and it is nearly constant as a function of sp@xmath49 fraction .
this shows the significant structural disorder at the outer shell of the nanocrystals .
in this review , we have briefly presented some of our work on the energetics , stability , mechanical and optoelectronic properties of carbon - based nanostructured composites .
these are materials consisting of a carbon crystalline nanoinclusion embedded in an amorphous carbon matrix .
the inclusion ranges from a diamond crystallite to sp@xmath49 and sp@xmath1 crystalline conformations with negative and positive curvature .
the ability to control the mechanical and optoelectronic properties of these materials by varying the density of the embedding medium and the type and size of the crystalline inclusion opens up enormous possibilities for potential technological applications of carbon nanocomposites .
we found diamonds and schwarzites to be highly stable in dense amorphous carbon matrices , while the other sp@xmath49 and sp@xmath1 conformations are metastable .
a considerable enhancement of the elastic moduli for @xmath34d/@xmath0-c and cnt/@xmath0-c was found as compared to pure @xmath0-c values with increasing density and inclusion size . @xmath0-c and @xmath34d/@xmath0-c of high density showed perfect elastic recovery , which was not the case for cnt/@xmath0-c . beyond the elastic regime , @xmath34d/@xmath0-c fracture inter - grain and in a manner similar to pure @xmath0-c . regarding the optoelectronic properties of @xmath34d/@xmath0-c ,
the matrix was shown to have a dominant role in forming both the edos and the dielectric function , with increasing sp@xmath49 fraction and density .
analysis of the optical response and the urbach energy , showed that much of the disorder in the composites originates from the interface region . + * acknowledgments * + the authors wish to thank g. kopidakis for useful discussions .
this work was supported by the ministry of national education and religious affairs of greece through the action `` @xmath57 '' ( program `` @xmath58 '' . )
mf acknowledges support from the gender issue incentive funds of the cluster of excellence in munich , germany .
m. o keeffe , g.b .
adams , and o.f .
sankey , phys .
lett . * 68 * , 1992 2325 .
d. vanderbilt and j. tersoff , phys .
* 68 * , 1992 511 .
s. j. townsend , t. j. lenosky , d. a. muller , c. s. nichols , and v. elser , phys .
69 * , 1992 921 .
s. iijima , nature * 354 * , 1991 56 .
g. a. j. amaratunga , m. chhowalla , c. j. kiely , i. alexandrou , r. aharonov , r. m. devenish , nature * 383 * , 1996 321 .
x. t. zhou , q. li , f. y. meng , i. bello , c. s. lee , s. t. lee , and y. lifshitz , appl .
. lett . * 80 * , 2002 3307 .
y. yao , m. y. liao , th .
khler , th .
frauenheim , r. q. zhang , z. g. wang , y. lifshitz , and s. t. lee , phys .
b * 72 * , 2005 035402
. w. kulisch , c. popov , h. rauscher , l. sirghi , t. sasaki , s. bliznakov , f. rossi , diam .
relat . mater .
* 17 * , 2008 1116 .
cohen , m. mehl , and d. papaconstantopoulos , phys .
b * 50 * , 1994 14694 .
tang , c.z .
wang , c.t .
chan , and k.m .
ho , phys . rev .
b * 53 * , 2996 979 . p.c .
kelires , phys .
* 73 * , 1994 2460 . c. mathioudakis , p.c kelires , y. panagiotatos , p. patsalas , c. charitidis , and s. logothetidis , phys .
b , * 65 * , 2002 205203 . j
raty , g. galli , c. bostedt , t. w. van buuren , and l. j. terminello , phys .
* 90 * , 2003 037401 . c. mathioudakis , g. kopidakis , p. c. kelires , c. z. wang , k. m. ho , phys .
b * 70 * , 2004 125202 . m.g .
fyta , i.n .
remediakis , p.c .
kelires , d.a .
papaconstantopoulos , phys .
lett , * 96 * , 2006 185503 .
p. lu , x.p .
li , and r.m .
martin , phys .
lett . * 68 * , 1992 1551 ; p. lu and w. yang , phys .
b * 49 * , 1994 11421 . | in this review , we present our recent computational work on carbon - based nanostructured composites .
these materials consist of carbon crystallites embedded in an amorphous carbon matrix and are modeled here through classical and semi - empirical quantum - mechanical simulations .
we investigate the energetics , mechano - elastic , and optoelectronic properties of these materials .
once the stability of the composites is discussed , we move on to the calculation of their elastic moduli and constants , their anisotropy and elastic recovery . at a next step ,
we focus on diamond composites , which were found to be the most stable among the composites studied , and went beyond the elastic regime to investigate their ideal fracture .
finally , for these materials , the electronic density of states , dielectric function , and optical response were calculated and linked to the disorder in the structures .
our findings unveil the high potential of these materials in nanotechnological applications , especially as ultlra - hard coatings .
carbon nanocomposites , amorphous carbon , nanotubes , mechanical properties , optoelectronic properties |
one of the promises of gravitational lenses is to measure the hubble constant @xmath0 ( refsdal 1964 ) from the observed time delays among the images of a variable background quasar source lensed by a foreground galaxy . given a model for the spatial distributions of the stars and dark matter in the lens galaxy , the time delay @xmath3 multiplied by the speed of light @xmath4 is simply proportional to the absolute distances to the lens and the source , hence @xmath5 scales with the size of the universe @xmath6 .
while the time delays are now routinely measured for many systems ( see schechter 2000 ) , a reliable determination of @xmath0 has been hampered to some extent by the intrinsic degeneracy in models of the dark matter potential of the lens ( williams & saha 2000 ; saha 2000 ; zhao & pronk 2001 ) .
the general trend is that a model with a dense dark matter halo gives a small @xmath0 with @xmath7,\ ] ] where @xmath8 is the critical density for the lens at redshift @xmath9 and the source at @xmath10 , and @xmath11 or @xmath12 is the typical surface density of luminous or dark matter at within the einstein ring @xmath13 ( falco , gorenstein , & shapiro 1985 ; kochanek 2002 ) .
given that the value of @xmath0 is now well constrained by other independent methods , such as the hst key project ( freedman et al .
2001 ) , it is interesting to reverse the angle of the question , and use equation@xmath14([trend ] ) to put more stringent constraint on the dark matter potential of the lens . more specifically in this _ letter _
, we would like to ask the question : how narrow is the allowed parameter space for the lens dark halo which is consistent with a given set of images , time delays , cosmology and @xmath0 ?
in the interest of clarity , we will consider only analytical lens models with a simplified geometry for a hypothetical image and lens system .
we believe our arguments should apply qualitatively to real galaxy lenses as well , and a more detailed application to the quadruple system pg1115 + 080 is given in a follow - up paper ( zhao & qin 2002 ) .
for simplicity we assume the four images form a perfectly symmetric einstein cross with the time delay minima on the y - axis at @xmath15 ( radian ) from the lens center , and the saddle point images on the x - axis at @xmath16 ( radian ) .
for generality we rescale the images , taking the radius @xmath13 as unity in a new @xmath17 coordinate system , so the four images are given by @xmath18_{\rm minima } = [ 0,\pm 1],~ [ x , y]_{\rm saddle } = [ \pm q,0].\ ] ] this image configuration implies that a source @xmath19 , exactly behind a spherical lens plus a linear external shear symmetric to the x and y axes .
all lensing properties can be derived from the time delay surfaces . for our model with spherical stellar lens @xmath20 plus halo @xmath21 and a linear shear of amplitude @xmath22 ,
the time delay contours are determined by a dimensionless time delay @xmath23 given by @xmath24 where @xmath25 is the cylindrical radius , and @xmath26 is a constant containing all the dependence on the cosmological density parameter @xmath27 and the lens and source redshifts @xmath28 ; crudely speaking @xmath29 for typical lens and source redshifts in the @xmath2cdm cosmology . to illustrate the non - uniqueness , we need to find reasonable lens models with similar time delay surface
let s consider the following example of lensing potential for the stars , @xmath30 where @xmath31 is the total stellar mass enclosed , @xmath32 is the half mass radius , and @xmath33 specifies the cuspiness of the stellar distribution . for the halo we use a nearly isothermal potential @xmath34\ ] ] where @xmath35 is proportional to the terminal velocity of the halo rotation curve .
note that the function @xmath36 is smoothly connected at the two sides of the radius @xmath37 , which is chosen such that it is just outside the images .
the parameter @xmath38 is a dimensionless tunable parameter to adjust the halo contribution to the surface density at @xmath37 .
the surface density can be computed as @xmath39 so the densities for the stars and the halo are given by @xmath40_{\rm max}}.\ ] ] note that the stars have an inner cusp @xmath33 , and the halo density @xmath41 is continuous across the break @xmath37 , and positive everywhere if @xmath42 . by increasing @xmath38
we can lower the mean densities at the images , hence creating the effect of a negative mass sheet .
this will not affect the positions of the images , but can increase the time delay @xmath43 between the images . as we will see
, this can result in a larger @xmath0 to be consistent with the currently favored high value of @xmath44 .
the deflection strength @xmath45 is effectively the rotation curve squared .
a flat rotation curve corresponds to a constant deflection strength . for our model ,
the stellar and halo masses enclosed inside radius @xmath46 are given by @xmath47.\ ] ] as we can see the stars converge to a finite mass @xmath31 at infinity and the halo dominates stars and approaches to a finite deflection ( terminal velocity @xmath48 ) .
let s consider a typical lensing system in a standard @xmath2cdm cosmology with @xmath49 with the lens and source redshifts @xmath50 and a time delay @xmath3 @xmath51,~q = 0.5\,\ ] ] between the saddle point image at @xmath52 and the minima image at @xmath53 .
the above cosmology specifies the distances to the lens and the source , hence the time delay normalization constant @xmath54 the observed time delay then set the following constraint on the lens model , @xmath55 we solve for the parameters of the lens and the shear to reproduce the four images at @xmath56 , @xmath57 exactly .
the images form at the extreme points of the time delay surface , hence we have the additional constraints @xmath58 it turns out that the allowed models follow a three - parameter sequence , say @xmath59 .
the parameters of a few representative models are listed in table 1 .
the images appear rigourously at the same locations for all four models , which are the minima and saddle points of the arrival time surfaces ( cf .
figure [ ctt]a ) , and the relative time delay is identical as well for all four lens models ( cf . figure [ ctt]b ) .
but the mass distributions in the four models are far from similar ( cf . figure [ vk]a and figure [ vk]b ) : lensiii and lensiv correspond to systems with a finite stellar core with a very typical half - mass radius @xmath60 about half of the einstein radius @xmath13 ; lensiv is purely in stars , and has no halo .
both lensi and lensii correspond to systems with a strong stellar cusp dominating the halo at small radii ; lensii has a bigger stellar component with a half - mass radius at one @xmath13 .
the above degeneracy is a variation of the well - known mass - sheet degeneracy .
the latter implies that we can increase @xmath0 by reducing the surface density @xmath61 at the einstein radius @xmath13 ( cf .
equation@xmath14([trend ] ) ) , e.g. , by either scaling down the stellar mass @xmath31 or scaling down the isothermal dark halo mass @xmath62 .
but if we reduce the stellar mass @xmath31 and increase the halo mass @xmath62 _ simultaneously _ such that we keep @xmath63 constant ( between 0.10 to 0.13 , cf .
table 1 ) , then @xmath64 will not change , only the terminal velocity of the rotation curve @xmath35 is raised .
the result is a rigorous degeneracy of the stellar vs. halo mass distribution , insensitive to @xmath0 and the observed time delay @xmath3 .
the flux ratio of the saddle image and the minima image can in princinple differentiate some of the models , as shown by fig . [ mag ] . but assuming a reasonable @xmath65 mag error with the magnitude , and a 10% error with the effective radius , most of the models are in fact degenerate
. in particular , it will be difficult to constrain the mass of the stellar component @xmath31 .
observations of a well - resolved stellar lens can fix the effective radius @xmath66 and perhaps the cuspiness @xmath33 .
this reduces the available lens models drasticly .
nonetheless , the degeneracy between lensiii and lensiv implies that it is still problematic to differentiate between models with dark halo and models without . to break the degeneracy one needs at least an accurate measurement of the external shear @xmath22 to 10% level ( cf .
the parameters of lensiii and lensiv in table 1 ) , perhaps by a combination of strong lensing and weak lensing data .
it is possible to construct many very different models with positive , smooth and monotonic surface densities to fit the image positions .
there are also no extra images .
these models fit the same images , time delay , @xmath0 and cosmology .
some fit the same lens light profile and image flux ratio as well .
hence the models are virtually _ indistinguishable _ from lensing data .
there are severe degeneracies in inverting the data of a perfect einstein cross to the lens models , even if given the hubble constant and cosmology .
these rigourous findings with analytical models are also consistent with earlier numerical models of saha & williams ( 2001 ) and semi - analytical models of zhao & pronk ( 2001 ) . among the acceptable models the rotation curve can be keplerian or flat ( fig .
[ vk]a ) , so lensing data plus @xmath0 can not uniquely specify the lens mass profile . among models in the literature , isothermal models and other simple smooth models of dark matter halos of gravitational lenses often predict a dimensionless time delay @xmath64 much too small ( e.g. , schechter et al . 1997 ) to be comfortable with the observed time delays @xmath3 and the widely accepted value of @xmath67 . naively speaking
the high @xmath0 suggests a strangely small halo as compact as the stellar light distribution ( kochanek 2002 ) .
but our analytical models suggest that there are still many other options .
the high @xmath0 implies that @xmath68 is small @xmath69 at the images , but this does not necessarily imply a rapidly falling density .
a high @xmath0 does not necessarily mean no dark halo , and models with a flat rotation curve does not always yield a small @xmath0 ( e.g. , compare lensiii and lensiv in figure [ vk]b , both satisfy @xmath70 ) .
we also comment that it will be difficult to determine the cosmology from strong lensing data alone because the non - uniqueness in the lens models implies that the combined parameter @xmath26 is poorly constrained by the lensing data , even if @xmath0 , @xmath9 and @xmath10 are given .
we thank the referee p. saha for insightful comments on the cause of the degeneracy .
this work was supported by the national science foundation of china under grant no .
10003002 and a pparc rolling grant to cambridge .
hsz and bq thank the chinese academy of sciences and the royal society respectively for a visiting fellowship , and the host institutes for local hospitalities during their visits .
falco , e.e . ,
gorenstein , m.v . , & shapiro , i.i . , 1985 ,
apj , 289 , l1 freedman , w.l .
, madore , b.f . ,
gibson , b.k .
, et al . , 2001 , apj , 553 , 47 kochanek , c. , 2002 , apj , submitted ( astro - ph/0205319 ) refsdal , s. , 1964 , mnras , 128 , 307 saha , p. , 2000 ,
aj , 120 , 1654 schechter , p.l .
, et al . , 1997 , apj , 475 , l85 schechter , p.l . , 2000 , iau 201 , lasenby , a.n . , & wilkinson , a. , eds .
( astro - ph/0009048 ) williams , l.l.r . , & saha , p. , 2000 , aj , 119 , 439 saha , p. , & williams l.l.r .
, 2001 , aj , 122 , 585 zhao , h.s . , & pronk , d. , 2001 , mnras , 320 , 401 zhao , h.s . , &
qin , b. , 2002 , apj , submitted ( paper ii ) lcccccccccc lensi & .2002 & .5 & 1 . &
.6220 & .2 & -.4445 & .0222 & .1110 & .1334 & .4220 + lensii & .7110 & 1 . & 1 . & .2988 & .1 & -.4458 & .0888 & .0494 & .3555 & .1988 + lensiii & .4901 & .5 & 0 . &
.2759 & .1 & -.4317 & .0784 & .0379 & .3921 & .1759 + lensiv & .7129 & .5 & 0 . &
.0013 & 0 . & -.4291 & .1140 & .0006 & .5703 & .0013 + | a degeneracy in strong lens model is shown analytically . the observed time delays and quasar image positions
might _ not _ uniquely determine the concentration and the extent of the lens galaxy halo mass distribution . simply hardwiring the hubble constant ( @xmath0 ) and the cosmology ( @xmath1 ) to the standard @xmath2cdm cosmology values might _
not _ fully lift this degeneracy , which exists rigourously even with very accurate data .
equally good fits to the images could be found in lens mass models with either a mostly keplerian or a flat rotation curve .
this degeneracy in mass models makes the task of getting reliable @xmath0 and @xmath2 from strong lenses even more daunting . |
bose einstein condensates ( becs ) of dilute atomic gases are an invaluable source of insight into the quantum behavior of matter at macroscopic scales @xcite .
topics like vortex structures or nonlinear matter waves are being elucidated through bec investigations @xcite . in particular ,
considerable effort has been invested in recent years in the study of solitary matter waves ( solitons ) @xcite due , in part , to their similarity with well - known equivalent structures in the field of nonlinear optics @xcite . a great variety of solitons
have been experimentally observed on the background of condensed atomic clouds .
dark solitons have been found @xcite as excitations in systems with repulsive interatomic interactions ( the analogue of self - defocusing nonlinear media ) , while bright solitons have been generated @xcite as ground states when the interaction is attractive ( self - focusing nonlinear media ) .
a particularly important special case of bright solitons is that of gap solitons @xcite .
these matter - wave nonlinear structures are localized states that can occur in becs with repulsive interatomic interactions if a periodic potential ( such as that produced by an optical lattice ) is present .
although gap solitons have been observed and studied in nonlinear optics since some time ago , only more recently have they been found in bec experiments @xcite .
this has triggered a renewed theoretical interest in the study of these nonlinear structures .
strictly speaking , solitons in a one - dimensional ( 1d ) setting are exact solutions of the one - dimensional nonlinear schrdinger equation ( 1d nlse ) . accordingly , in the framework of the mean - field approximation , matter - wave solitonic structures in 1d optical lattices have usually been investigated theoretically by means of the 1d gross pitaevskii equation ( gpe ) @xcite , which is nothing but a nonlinear schrdinger equation describing the evolution of the condensate wavefunction in the presence of the confining potentials
. however , soliton - like states realized in becs always have an intrinsic three - dimensional ( 3d ) character , even when the trapping potential imposes a tight radial confinement . in fact , typical experimental parameters for the generation of gap solitons in becs lie in a region of parameter space where the quasi-1d approximation is not strictly valid @xcite and , as a consequence , very often the 1d gpe can not quantitatively reproduce the dynamics of realistic matter - wave gap solitons .
various approaches have been put forward with the aim of incorporating in the description of condensed gases in elongated geometries the consequences of their intrinsic 3d character .
effective models for studying the stationary properties of highly anisotropic becs beyond the analytically solvable perturbative and thomas fermi regimes were proposed in refs .
an analytic expression for the local chemical potential @xmath0 of elongated condensates as a function of the axial linear density was independently proposed in refs . @xcite and @xcite .
such expression remains valid in the 3d1d crossover regime and , in the case of @xcite , is also valid in the presence of an axially symmetric vortex .
effective equations have also been introduced to study topics such as solitons @xcite , the influence of the transverse confining on the dynamics of becs in 1d optical lattices @xcite , or the emergence of faraday waves @xcite .
of particular relevance because of their simplicity and usefulness are the effective 1d equations derived in refs . @xcite and @xcite , which govern the condensate dynamics accounting properly for the contributions from excited transversal modes through a nonpolynomial nonlinear term .
several variants of the above equations aimed at improving their accuracy to the price of incorporating adjustable @xcite or a larger number of variational parameters @xcite have also been proposed .
however , the intended increase in accuracy hardly compensates for the complexity of the resulting equations or the arbitrariness in the choice of free parameters . on the other hand ,
a number of generalizations of the above equations have been derived to study issues such as the expansion of elongated becs @xcite , spin-1 atomic condensates @xcite , matter - wave vortices @xcite , becs in anisotropic transverse traps @xcite , binary condensates in mixed dimensions @xcite , dark bright solitons in two - component becs @xcite , dipolar becs @xcite , or localized modes in condensates with spin - orbit and rabi couplings @xcite .
the equations of refs . @xcite and
@xcite have also been used to study becs in nonlinear lattices @xcite , bright solitons in condensates with inhomogeneous defocusing nonlinearities @xcite , or the bogoliubov spectrum of elongated condensates @xcite .
since these effective equations are extensions of the standard 1d gpe ( to which they reduce in the proper limit ) and can be numerically solved with similar computational effort , they may offer a clear advantage over the latter . in this work
, we will focus on the effective equation proposed in refs .
this equation has been used in the description of matter - wave dark solitons in elongated becs and has proven to give results in good agreement with the experiments @xcite . as was demonstrated in ref .
@xcite , it can also be applied to the study of stationary gap solitons in the different physically relevant mean - field regimes .
this is a nontrivial issue because this kind of systems are characterized by a small spatial scale ( the lattice period @xmath1 ) that can represent a rather restrictive condition .
however , to establish the utility of the model in realistic situations and determine whether it may be a real alternative to a fully 3d treatment one has to analyze its ability to reproduce a number of physical properties of fundamental interest in the theoretical description of stationary matter - wave gap solitons that were not studied in ref .
one of these properties is the dependence of the soliton chemical potential @xmath2 on the number @xmath3 of constituent atoms , a quantity that can be used to define trajectories in the @xmath4 plane which are essential for the classification of gap solitons into different families .
another one is the frequency spectrum of the elementary excitations determining the dynamical stability of the gap soliton family .
to provide an answer to the previous question , in this work we analyze to what extent the 1d gpe and the effective equation of refs .
@xcite can reproduce the above - mentioned fundamental physical properties in realistic situations . to this end
we consider condensed atomic clouds confined by transverse harmonic potentials in a range of parameters typical of current experiments and compare with exact numerical results obtained from the full 3d gpe .
since in this work we are interested in the mean - field regime , we shall restrict our analysis to physical configurations with a large number of atoms per lattice well , which are amenable to a treatment in terms of a mean - field macroscopic wavefunction satisfying the gpe .
the paper is organized as follows .
section ii briefly reviews the theoretical models considered in this work . in secs .
iii and iv we analyze the numerical results : in sec .
iii we focus on the wavefunctions and the distinctive trajectories of the different gap soliton families , and in sec .
iv we study the frequency spectrum of the elementary excitations and the stability properties of these nonlinear structures .
finally , in sec .
v we summarize our results and present our conclusions .
when a periodic potential is imposed on a bec , the chemical potential of the system develops a band - gap structure . in these circumstances , it is possible to find either extended states with chemical potential inside the bands ( nonlinear bloch waves ) or localized gap solitons between bands . at zero temperature and in the mean - field regime ,
these coherent states are accurately described by the 3d gpe @xcite : @xmath5 where @xmath6 is the interaction parameter , with @xmath7 being the _ s_-wave scattering length , @xmath3 is the number of atoms , and @xmath8 is the potential of the trap . in this work
we will consider an external potential of the form @xmath9 , which besides the axial term @xmath10 accounting for a 1d optical lattice with period @xmath1 , includes a transverse harmonic trapping @xmath11 , where @xmath12 .
a description in terms of the gpe corresponds to a classical mean - field description that neglects quantum correlations .
it represents the classical limit of the underlying quantum field theory and becomes an excellent approximation in the limit @xmath13 with the nonlinear interaction parameter @xmath14 held constant . in practice , quantum fluctuations decay
rapidly with @xmath3 and one expects the gpe to be a good approximation for configurations with a few hundreds of atoms per lattice well .
although the 3d gpe can be solved numerically with no approximations @xcite , it demands a considerable computational effort when either the 3d character of the system or the nonlinearity increases . in order to simplify the computational problem , simpler models with reduced dimensionality
are commonly used in the study of gap solitons in 1d optical lattices . whenever it is possible to make the assumption that the condensate remains in the ground state of the transverse harmonic trap ( so that higher transverse modes are not excited ) , a good approximation to eq .
( [ 3dgpe ] ) is given by the 1d gpe with rescaled nonlinearity : @xmath15 where @xmath16 is the 1d interaction parameter , obtained after averaging over the ` frozen ' transverse degrees of freedom , and @xmath17 is the axial wavefunction of the condensate .
it is important to bear in mind that this equation is only valid when the axial energies are small enough in comparison with the radial quantum , @xmath18 , a condition that , in general , is not satisfied in realistic situations .
a simple dimensional analysis shows that the physical problem is characterized by three dimensionless parameters which can be conveniently chosen as@xmath19 where @xmath20 is the recoil energy of the lattice , with @xmath21 being the lattice wavevector , @xmath22 is the lattice depth , and @xmath23 is the nonlinear coupling constant determining the mean - field interaction energy . the derivation of eq .
( [ 1dgpe ] ) relies on the assumption that the 3d wavefunction @xmath24 can be factorized in independent radial and axial components , with the radial component corresponding to the gaussian wavefunction of the ground state .
such assumption entails an implicit adiabatic approximation according to which the transverse degrees of freedom can instantaneously follow the axial dynamics . in order for this to be true
, the characteristic time scale of the axial motion must be much longer than that of the radial motion ( @xmath25 ) .
moreover , since the validity of the 1d gpe ( [ 1dgpe ] ) requires the radial configuration to remain in the ground state , the axial energies involved must be sufficiently small in comparison with the characteristic energy of the transverse harmonic oscillator , @xmath18 .
taking into account that the axial energy of the underlying linear problem is of the order of the recoil energy @xmath26 and that the nonlinear interaction energy is of the order of @xmath27 ( with @xmath28 being the linear density of the gap soliton along the axial direction ) , the above requirements can be stated as@xmath29 these conditions , which ensure the permanence of the system in the transverse ground state , also guarantee automatically the validity of the adiabatic approximation .
however , in order for the mean - field approximation implicit in eqs .
( [ 3dgpe ] ) and ( [ 1dgpe ] ) to be applicable , it is also necessary that @xmath30 . incorporating this latter requirement ,
the conditions for the validity of the 1d gpe can be rewritten as@xmath31 where @xmath32 is the radial harmonic - oscillator length .
these conditions are very restrictive and usually are not met in real condensates .
for instance , for @xmath33rb atoms @xmath34 @xmath35 and @xmath36 @xmath37 ( with @xmath38 given in @xmath39 ) while , typically , the lattice parameter @xmath1 varies between @xmath40 and @xmath41 @xmath37 @xcite , the lattice depth @xmath42 between @xmath43 and @xmath44 @xcite and , for elongated condensates , @xmath38 varies between @xmath45 and @xmath46 @xmath39 .
thus , typically , @xmath47 @xmath37 , @xmath48 @xmath37 , and @xmath49 which makes the conditions ( [ con3 ] ) hard to satisfy . in particular , in the experiment of ref .
@xcite , where gap solitons were observed in a @xmath33rb condensate , @xmath50 @xmath37 , @xmath51 @xmath37 , @xmath52 , and @xmath53 atoms . in that of ref .
@xcite , @xmath54 @xmath37 , @xmath55 @xmath37 , @xmath56 , and @xmath57 @xmath33rb atoms .
it is important to note , however , that even though most experiments have been carried out in the above parameter regimes , larger values for @xmath1 and @xmath58 ( which can realize the 1d mean - field regime ) are well within experimental reach .
next , we briefly recall the 1d effective model proposed in refs .
@xcite , which is an extension of the standard 1d gpe .
as before , the starting point is the adiabatic approximation , which allows the factorization of the 3d wavefunction in the form@xmath59 where the radial wavefunction @xmath60 is assumed to be normalized to unity and @xmath61 is the local linear density@xmath62 after substituting eq .
( [ ii-4 ] ) in eq .
( [ 3dgpe ] ) and averaging over the radial degrees of freedom one obtains the following 1d effective equation@xmath63 where @xmath0 is the transverse local chemical potential which follows from the stationary 2d gpe@xmath64 when the dimensionless linear density is small enough ( @xmath65 ) , this latter equation can be solved perturbatively and , to the lowest order ( _ i.e. _ , for @xmath66 given by the gaussian ground state of the corresponding harmonic oscillator ) , one obtains @xmath67 . substituting back in eq .
( [ ii-6 ] ) and taking into account eq .
( [ ii-5 ] ) it is thus clear that one arrives at the 1d gpe ( [ 1dgpe ] ) .
however , a more general solution to eq .
( [ ii-7 ] ) can be found . using two independent approaches
it was demonstrated in refs .
@xcite that , for condensates with no vorticity ( the case we are interested in here ) , a very accurate solution , valid for arbitrary @xmath68 , is given by@xmath69 substituting this expression in eq .
( [ ii-6 ] ) , one finally finds@xmath70 this effective 1d equation governs the axial dynamics of the condensate , having a much wider range of applicability than the standard 1d gpe ( [ 1dgpe ] ) .
in particular , now the linear density @xmath68 can take arbitrary values and in the limit @xmath65 one recovers eq .
( [ 1dgpe ] ) .
yet , the applicability of the above equation still requires the fulfillment of the adiabatic approximation and , as far as the dynamical problem is concerned , this condition can be hard to satisfy in the presence of an optical lattice .
things are different , however , for the stationary problem , which is , in fact , the most relevant in the characterization of matter - wave gap solitons . in this case , the time scale of the axial dynamics tends to infinity and the adiabatic approximation always holds . as a result , unlike the usual 1d gpe ( [ 1dgpe ] ) , the effective model ( [ e1de ] ) will be able to correctly predict the trajectories in @xmath4 plane as well as the wavefunctions and stability properties of the different gap soliton families . represents the number of @xmath33rb atoms , and @xmath71 is the dimensionless chemical potential . the solid red curve is the prediction from the 1d effective equation ( [ e1de ] ) , the dashed blue curve is that from the 1d gpe ( [ 1dgpe ] ) , and the open symbols are exact results obtained by solving numerically the full 3d gpe ( [ 3dgpe ] ) .
the inset shows a representative example of a gap soliton in this family ( point a ) in terms of a density isosurface ( corresponding to @xmath72 of its maximum value ) . the local phase ( which is uniform in this case ) is also represented as a color map.,width=275 ]
next , we will apply both eqs .
( [ 1dgpe ] ) and ( [ e1de ] ) to the study of stationary gap solitons in a regime corresponding to typical experimental parameters . to this end , by using a newton continuation method we look for numerical solutions of the form @xmath73 , with @xmath2 being the condensate chemical potential .
these solutions define distinctive trajectories in @xmath4 plane which , as already said , are essential for the classification of gap solitons into different families . to investigate
how accurately the above 1d equations can predict such trajectories and , thus , the physical properties of the various families , we compare the predictions from both equations with those obtained from the numerical solution of the full 3d gpe . in what follows we consider a @xmath33rb condensate subject to a rather tight transverse potential of frequency @xmath74 @xmath39 and in the presence of a 1d optical lattice with period @xmath75 and depth @xmath76 @xmath77 . for these parameters ,
the recoil energy of the lattice takes the value @xmath78 . .
solid red lines are the predictions of the 1d effective equation ( [ e1de ] ) , dashed blue lines are those of the 1d gpe ( [ 1dgpe ] ) , and open symbols are exact results obtained by solving numerically the full 3d gpe ( [ 3dgpe]).,width=321 ] figure [ fig1 ] shows the different theoretical predictions the above equations make for the family of fundamental gap solitons .
this figure depicts the dimensionless chemical potential @xmath79 of the solitons versus the number @xmath3 of constituent atoms .
the 3d band - gap structure of the underlying linear problem is shown in the background , with shaded stripes corresponding to the allowed energy bands .
the two narrow stripes located at @xmath80 and @xmath81 are purely 3d bloch bands associated with excited transversal states and , thus , can not be accounted for by any 1d model .
they are replicas of the lowest energy band ( located at @xmath82 ) shifted up in energy , respectively , by two and four quanta of the radial harmonic oscillator , @xmath83 @xcite .
open symbols in this figure are exact results obtained by solving numerically the full 3d gpe ( [ 3dgpe ] ) , the solid red curve is the prediction from the 1d effective equation ( [ e1de ] ) , and the dashed blue curve is that from the standard 1d gpe ( [ 1dgpe ] ) . as is evident , this latter equation clearly fails .
in fact , even in the first ( 1d ) gap , it makes errors larger than @xmath84 and thus can hardly be considered valid . on the contrary , the effective model of eq .
( [ e1de ] ) is in good quantitative agreement with the 3d results over the entire curve .
gap solitons in this family are nonlinear stationary states exhibiting a single peak well localized at a lattice site . a representative example is displayed in the inset , which shows the 3d atom density of a gap soliton of this family containing @xmath85 particles ( point a marked by an open triangle in the figure ) .
its chemical potential , @xmath86 @xmath87 , reflects the fact that the corresponding transversal state is composed of multiple harmonic - oscillator modes .
the prediction the effective model ( [ e1de ] ) makes for this soliton has an error in the number of particles of only @xmath88 .
figure [ fig1 ] also shows that while the results from the 1d gpe get worse as the number of atoms increase , those from the effective equation ( [ e1de ] ) remain in good quantitative agreement with the 3d results . , but for the symmetric composite family of gap solitons consisting of two in - phase peaks.,width=283 ] this behavior is corroborated by fig .
panels a and b in this figure display the dimensionless axial densities @xmath68 ( left panels ) and wavefunctions @xmath89 ( right panels ) of the fundamental gap solitons corresponding , respectively , to points a and b in fig . [ fig1 ] . as
before , solid red curves have been obtained from the effective equation ( [ e1de ] ) , dashed blue curves from the standard 1d gpe , and open symbols are numerical results obtained from the full 3d gpe . for reference ,
the lattice potential is also shown in arbitrary units ( dotted lines ) . in order to deduce the axial wavefunction from the corresponding 3d results
we have used that @xmath90 , where the linear density can be readily obtained by performing the integral in eq .
( [ ii-5 ] ) . as for the axial phase @xmath91 ,
we have defined it as the average local phase at every @xmath92-plane . again
the agreement between the predictions of eq .
( [ e1de ] ) and the 3d results is very good , while the 1d gpe clearly fails .
similar conclusions can be drawn in the case of multiple gap solitons .
figure [ fig3 ] displays the trajectories in @xmath93 plane corresponding to the composite family consisting of two in - phase peaks ( see also fig .
[ fig4 ] ) .
solitons in this family are bound states formed by a symmetric superposition of two fundamental gap solitons of the family @xmath94 , _
i.e. _ those displayed in fig . [ fig1 ] .
here we are using the notation introduced in ref .
@xcite , which enables us to classify the different fundamental gap solitons into families characterized by the quantum numbers @xmath95 corresponding to the 3d bloch band from which they bifurcate , with @xmath96 being the band index of the corresponding 1d axial problem and @xmath97 and @xmath98 being , respectively , the angular - momentum and radial quantum numbers characterizing the transversal state . in this work , as is usually the case , we restrict ourselves to gap solitons consisting of fundamental constituents of the type @xmath99 , which are amenable to an effective treatment in terms of the 1d equations ( [ 1dgpe ] ) and ( [ e1de ] ) .
gap solitons with @xmath100 have been rarely considered in the literature @xcite .
they feature a nontrivial radial topology and require a more elaborate treatment @xcite .
the inset in fig .
[ fig3 ] shows the atomic density ( as an isosurface taken at @xmath72 of the maximum density ) of the soliton corresponding to point d ( marked by an open triangle ) .
this soliton contains @xmath101 atoms and has a chemical potential @xmath102 .
as can be seen , the two fundamental constituents occupy adjacent lattice sites and have the same phase ( shown in the inset in terms of a color map ) .
this is also apparent from fig .
[ fig4 ] , which depicts the axial linear densities @xmath68 and wavefunctions @xmath103 of the gap solitons marked by c and d in fig .
a simple comparison also shows that the wavefunctions of the composite solitons are indeed symmetric superpositions of those in fig .
[ fig2 ] . ,
but for the gap solitons corresponding to points c and d in fig .
[ fig3].,width=321 ] gap solitons consisting of two out - of - phase peaks are analyzed in fig .
the upper trajectories in this figure ( those bifurcating from the first bloch band ) correspond to composite solitons formed by an antisymmetric superposition of two fundamental @xmath94 gap solitons .
an example of a soliton of this family is displayed in the upper inset , which shows a representation of the 3d wavefunction of the soliton corresponding to point e in the figure .
this wavefunction is given in terms of an isosurface of the atomic density ( taken at @xmath72 of its maximum ) and a color map indicating the local phase at each point .
as in the previous case , the two fundamental constituents of this compound soliton occupy adjacent lattice sites , but now they exhibit a @xmath104 jump in their relative phase . a detailed comparison of the axial density and wavefunction of this soliton ( given in the upper panels of fig .
[ fig6 ] ) with those of the corresponding fundamental solitons ( given in the lower panels of fig .
[ fig2 ] ) confirms that this soliton is a bound state consisting of the antisymmetric superposition of two of the fundamental gap solitons previously studied . , but for the antisymmetric composite ( upper curves ) and fundamental ( lower curves ) gap soliton families consisting of two out - of - phase peaks.,width=283 ] the lower trajectories in fig .
[ fig5 ] , which bifurcate from the second 1d bloch band , constitute a family that has been referred to as the _ subfundamental _ family @xcite .
gap solitons in this family are associated with the first excited ( @xmath105 ) energy band of the corresponding 1d axial problem , which reflects in the fact that they exhibit two major peaks in the axial direction ( _ i.e. _ , one axial node ) localized in a single lattice site ( see the lower inset in fig .
[ fig5 ] as well as panels f in fig .
[ fig6 ] ) . unlike the previous case , these solitons ( which can be regarded as excited states featuring an embedded dark soliton ) are not bound states of two fundamental @xmath94 gap solitons .
they are fundamental solitons of the @xmath106 family which can be used as elementary constituents of more complex compound structures . , but for the gap solitons corresponding to points e and f in fig .
[ fig5].,width=321 ] as is apparent from the above results , the predictions from the 1d effective model ( [ e1de ] ) are always in good quantitative agreement with the 3d numerical results .
stability is of primary importance for the experimental relevance of gap solitons . in this section
we are interested in determining whether the effective equation ( [ e1de ] ) is able to correctly predict the stability properties of the gap solitons previously considered . to this end
we perform a stability analysis based on the above effective equation and compare with the respective 3d results .
we begin by deriving the bogoliubov de gennes ( bdg ) equations corresponding to eq .
( [ e1de ] ) and then obtain numerically the frequency spectrum of the elementary excitations . to derive the bdg equations , we perturb a given gap soliton stationary wavefunction @xmath107 in the form @xcite @xmath108 e^{-i\mu t/\hbar } \label{bdg1}\ ] ] and substitute in eq .
( [ e1de ] ) retaining only linear terms in the amplitudes @xmath109 and @xmath110 of the normal modes of the system .
the numerical solution of the resulting linear equations provides the frequencies of the elementary excitations determining the dynamical stability of the soliton .
a similar procedure has been followed to derive the bdg equations corresponding to the 3d gpe ( [ 3dgpe ] ) .
results for gap solitons b , d , e , and f are given in the panels labeled with the same letters in fig .
[ fig7 ] . )
( left panels ) and to the 3d gpe ( right panels ) .
the horizontal axis in each figure represents the real part of the excitation frequencies @xmath111 shifted by the chemical potential @xmath112 of the corresponding stationary soliton , @xmath113 , while the vertical axes show the respective imaginary parts , @xmath114 .
note that in all cases considered @xmath115.,width=321 ] left panels show the corresponding excitation spectra as obtained from the 1d effective model ( [ e1de ] ) , while right panels display the eigenfrequencies of the bdg equations associated to the 3d gpe .
the horizontal axis in each figure represents the real part of the excitation frequencies @xmath111 shifted by the chemical potential @xmath112 of the corresponding stationary soliton , @xmath116 .
vertical axes show the respective imaginary parts , @xmath114 .
as is well known , the absence of complex eigenfrequencies in the spectrum implies the linear stability of the system .
of course , the excitation spectrum of the full 3d problem necessarily contains frequencies that can not exist in the effective problem with reduced dimensionality ( for instance , those associated to radial normal modes ) .
this is an expected result .
the point , however , is that , except in those cases where the solitons exhibit a nontrivial ( unstable ) transverse configuration , one might also reasonably expect that the stability properties of the different gap solitons will be dictated by the axial degrees of freedom , which ultimately are the ones responsible for the existence of self - trapping .
if this expectation proves true , the effective 1d model ( [ e1de ] ) could be a reliable tool for the investigation of the stability of these nonlinear structures . as is apparent from a comparison between left and right panels in fig .
[ fig7 ] , this is indeed the case : except for the gap soliton f ( lower panels ) , which is a bright soliton containing a dark soliton that extends considerably in the radial direction , the effective 1d model predicts accurately the complex eigenfrequencies determining the stability of the various gap solitons .
according to the excitation spectra shown in the figure , the fundamental soliton b and the symmetric composite soliton d are linearly stable while soliton e ( the antisymmetric counterpart of the latter ) is unstable . since in all cases considered @xmath115 ,
it is clear from the figure that the spectrum of this soliton exhibits a pair of purely imaginary eigenfrequencies , indicating an exponential instability . as for soliton
f , even though the effective model ( [ e1de ] ) still predicts correctly the ( oscillatory ) unstable nature of this soliton , it can not give a quantitative account of its 3d excitation spectrum which now contains a quartet of complex frequencies not appearing in the corresponding 1d spectrum ( which exhibits another quartet of complex frequencies , although only a pair is visible in the field of view of the figure ) .
one expects these complex frequencies not appearing in the 1d spectrum to be associated with unstable transverse excitations of the embedded dark soliton ( snake - like instability ) . to verify this and
the previous conclusions we have introduced an additive gaussian white noise to produce a random small perturbation in the stationary configuration of the different gap solitons and have followed the ensuing time evolution by integrating numerically both the effective 1d equation ( [ e1de ] ) and the 3d gpe ( [ 3dgpe ] ) .
the results obtained are collected in fig .
the upper left panel shows the dynamical evolution of soliton d in terms of a density map where brighter regions correspond to higher densities .
this image , which has been obtained from the 1d effective equation ( [ e1de ] ) , is indistinguishable from the corresponding 3d result and confirms the stability of the symmetric composite soliton d as predicted by the linear stability analysis of fig .
[ fig7 ] . the same holds true for the fundamental soliton b ( not shown in the figure ) . for later convenience
, the upper right panel displays the time evolution of the real part of the wavefunction of soliton d ( in arbitrary units ) obtained from both the 1d ( top ) and the 3d models ( bottom ) .
since this soliton is stable , its dynamical evolution , apart from small local fluctuations induced by the gaussian perturbation , is given essentially by a periodic amplitude modulation @xmath117 with period @xmath118 ms , as corresponds to a ( quasi ) stationary state .
thus stability implies , in particular , that the relative phase between any pair of points in the global amplitude envelope must remain ( quasi ) constant over time .
accordingly , even though an analysis of the density evolution of the antisymmetric composite soliton e ( middle left panel in fig .
[ fig8 ] ) seems to indicate that this soliton is stable ( the same conclusions follow from both 1d and 3d results ) , a more detailed study including also phase information reveals that this is not true .
indeed , as can be seen from the middle right panel , in this case the time evolution of the real part of the wavefunction no longer is a periodic amplitude modulation . while the two major peaks are initially out of phase , both the 1d effective model ( top ) and the 3d model ( bottom ) predict that they become essentially in phase at @xmath119 ms , so that , this soliton turns out to be unstable , in agreement with the linear stability analysis of fig .
this example clearly reflects the importance of the spectrum of elementary excitations in the analysis of the dynamical stability of these nonlinear structures . in the two lower panels we compare the evolution in time of the density of soliton f obtained from eq .
( [ e1de ] ) ( left panel ) with that obtained from the 3d gpe ( right panel ) . as is apparent , in this case the effective 1d model predicts a longer soliton lifetime than the 3d model , in good agreement with the linear stability analysis of fig .
the inset in the right panel shows that the soliton decay occurs when the nodal plane of the embedded dark soliton begins to undergo a considerable deformation , thus , indicating that the decay is a consequence of a transverse modulational instability which clearly can not be accounted for with a 1d model .
one may wonder why the effective model ( [ e1de ] ) is able to properly predict the stability properties of the gap solitons even though , in the presence of the optical lattice ( and because of the high frequency axial motion it might induce ) , in general , it is unable to accurately reproduce the detailed evolution in time . to address this question , we begin by noting that the adiabatic ( born oppenheimer ) approximation essentially consists in assuming that the axial dynamics is so slow in comparison to the typical time scale of the radial degrees of freedom that , at every instant @xmath120 , the latter can adjust to their equilibrium configuration compatible with the axial configuration @xmath61 occurring at that instant
this is exactly what the stationary radial equation ( [ ii-7 ] ) reflects . in the presence of an optical lattice
, one thus reasonably expects this approximation to work better in those situations where the time variable does not play a prominent role , such as occurs for the stationary solutions of the time - dependent gpe , which are completely determined from a ( time - independent ) eigenvalue problem .
something similar can be expected for the linear stability properties of the system .
indeed , such properties are completely characterized by the frequency spectrum of the elementary excitations , which follow from the solution of an eigenvalue problem ( bdg equations ) where time does not play any role . even though an instability manifests itself in the evolution in time ,
the very existence of a certain unstable mode is a characteristic feature of the system which depends only on the interplay among the various energies and interactions contributing to the hamiltonian .
this explains why the effective model ( [ e1de ] ) can predict the stability properties of the gap solitons even though it is unable to properly account for their dynamical evolution .
most theoretical studies of gap solitons realized in condensates loaded in 1d optical lattices have been carried out in terms of the 1d gpe .
this is motivated , in part , by the analogy of the latter equation with the 1d nlse governing the evolution of similar structures in nonlinear optics and , in part , by the fact that , as occurs with dark solitons , one would reasonably expect these structures to be more stable in a quasi-1d regime , where their potential decay would be inhibited by the strong radial confinement .
however , it has been demonstrated that robust , long - lived gap solitons exist in a 3d regime where many higher - order radial modes are excited @xcite . moreover , matter - wave gap solitons are intrinsically 3d and in realistic situations can hardly satisfy conditions ( [ con3 ] ) , which are necessary for the validity of the 1d gpe . in these circumstances , effective equations like eq .
( [ e1de ] ) , which can take the 3d character of stationary becs into account by incorporating contributions from higher - order radial modes , offer a clear advantage over the latter .
however , the question remains open whether the effective 1d equation ( [ e1de ] ) represents a reliable alternative to a fully 3d treatment . in this work
we have addressed this question by analyzing the ability of the above effective equation to reproduce the fundamental physical properties of stationary matter - wave gap solitons in realistic conditions .
our results demonstrate that the predictions from the effective model ( [ e1de ] ) are in good quantitative agreement with the exact numerical results from the full 3d gpe .
in particular , unlike the standard 1d gpe ( which fails in most cases of practical interest ) , the 1d effective model ( [ e1de ] ) correctly predicts the distinctive trajectories characterizing the different gap soliton families as well as the corresponding axial wavefunctions along the entire band gaps .
our results also show that this effective model can predict correctly the stability properties of the different gap soliton families .
this follows from both a linear stability analysis in terms of the corresponding bdg equations and a representative set of numerical computations monitoring the long time response of the system to a sudden small random perturbation . in particular , by numerically solving the bdg equations we have analyzed the prediction the effective model makes for the spectrum of elementary excitations , which proves to be essential for unambiguously determining the stability properties of certain matter - wave gap solitons .
our results show that , except in those cases where the soliton features a nontrivial ( unstable ) transverse configuration , the model is able to correctly predict the spectrum of complex eigenfrequencies responsible for the dynamical stability of the system .
these results thus indicate that the above effective 1d model can be a useful tool for the physical description of realistic matter - wave gap solitons in 1d optical lattices . as a natural extension of the present work , one might consider to quantitatively investigate the effects of quantum fluctuations on the above stationary matter - wave gap solitons . in the tight - binding regime ( @xmath121 )
, the tunneling of atoms between adjacent lattice sites can be strongly inhibited . under these conditions ,
intersite atom - number fluctuations become significantly suppressed while the corresponding phase fluctuations become enhanced , which can lead to a reduced relative phase - coherence . while for fundamental gap solitons ( those located at a single lattice well ) with a few hundreds atoms these quantum effects
are not expected to be very significant , they can be more important in the case of composite in - phase or out - of - phase solitons .
the problem of quantifying the contribution of these quantum fluctuations can be addressed in terms of a stochastic phase - space description based on the truncated wigner approximation ( twa ) @xcite . in this respect ,
we note that since eq .
( [ e1de ] ) properly captures the bogoliubov excitation spectrum of the system , it represents a particularly convenient starting point for such an analysis .
this will be the subject of a future publication .
j. denschlag , j. e. simsarian , d. l. feder , charles w. clark , l. a. collins , j. cubizolles , l. deng , e. w. hagley , k. helmerson , w. p. reinhardt , s. l. rolston , b. i. schneider , and w. d. phillips , science * 287 * , 97 ( 2000 ) .
p. engels and c. atherton , phys .
* 99 * , 160405 ( 2007 ) ; c. becker , s. stellmer , p. soltan - panahi , s. drscher , m. baumert , e .- m .
richter , j. kronjger , k. bongs , and k. sengstock , nature physics * 4 * , 496 ( 2008 ) ; s. stellmer , c. becker , p. soltan - panahi , e .- m .
richter , s. drscher , m. baumert , j. kronjger , k. bongs , and k. sengstock , phys .
* 101 * , 120406 ( 2008 ) ; a. weller , j. p. ronzheimer , c. gross , j. esteve , m. k. oberthaler , d. j. frantzeskakis , g. theocharis , and p. g. kevrekidis , _
ibid_. * 101 * , 130401 ( 2008 ) ; i. shomroni , e. lahoud , s. levy , and j. steinhauer , nature physics * 5 * , 193 ( 2009 ) .
j. ruostekoski and a. d. martin , truncated wigner method for bose gases . in , quantum gases
: finite temperature and non - equilibrium dynamics ( vol . 1 cold atoms series ) , pag . 203 .
n. proukakis , s. gardiner , m. davies , m. szymanska , and n. nygaard ( eds . ) ( imperial college press , london , 2013 ) . | we consider stationary matter - wave gap solitons realized in bose
einstein condensates loaded in one - dimensional ( 1d ) optical lattices and investigate whether the effective 1d equation proposed in [ phys .
rev .
a * 77 * , 013617 ( 2008 ) ] can be a reliable alternative to the three - dimensional treatment of this kind of system in terms of the gross pitaevskii equation ( gpe ) .
our results demonstrate that , unlike the standard 1d gpe ( which is not applicable in most realistic situations ) , the above effective model is able to correctly predict the distinctive trajectories characterizing the different gap soliton families as well as the corresponding axial wavefunctions along the entire band gaps .
it can also predict the stability properties of the different gap soliton families as follows from both a linear stability analysis and a representative set of numerical computations .
in particular , by numerically solving the corresponding bogoliubov
de gennes equations we show that the effective 1d model gives the correct spectrum of complex eigenfrequencies responsible for the dynamical stability of the system , thus providing us with a useful tool for the physical description of stationary matter - wave gap solitons in 1d optical lattices . |
in the early universe , astrochemistry is still almost unexplored . observations of dwarf galaxies with low metallicities and spectral scans of ultraluminous infrared galaxies ( ulirgs ) in the local universe are a great help , not only in getting a better understanding of these interesting classes of targets , but also to improve our knowledge on molecular cloud properties and star formation under exotic conditions resembling to a certain extent those in the distant past .
however , in the end in spite of all difficulties related to their large distances nothing can replace direct detailed studies of the high-@xmath14 sources themselves . while there have been some attempts to search for molecules other than co at @xmath15 ( e.g. hcn , hnc , cn ; @xcite ) , the number of detected species is still fairly small .
right now we only see the tip of the iceberg and a broader set of observational data is mandatory .
one of the most promising molecules for studies of the early universe is heh@xmath0 @xcite .
hydrogen and helium are the main components of matter in the universe .
available chemical models predict heh@xmath0 to be one of the most abundant molecules in the post - recombination era ( e.g. * ? ? ?
several authors analysed heh@xmath0 formation and excitation in astrophysical plasmas and in various nebulae ( e.g. * ? ? ?
* ; * ? ? ?
* and references therein ) .
they found that the heh@xmath0 abundance can be particularly high in vicinity of strong extreme uv and x - ray emitting sources , sufficient to produce an observable line emission . despite of the expectations
this molecule has not been detected in space yet . with the iso lws instrument
@xcite detected an emission feature at the frequency of the heh@xmath0 @xmath1 transition at 149.18 @xmath16 m in the planetary nebula ngc7027 .
however , their analysis shows that this feature is most probably attributed to the ch @xmath3 fundamental pure rotational lines at nearby frequencies ( the iso lws spectral resolution was insufficient to separate these lines ) .
nevertheless , we may expect a rather high abundance of heh@xmath0 in dense clouds near distant quasars where its cosmological abundance can be drastically enhanced by local energetic processes . some other simple molecules ( e.g. ch ) may also be sufficiently abundant .
we report here results of a search for heh@xmath0 and ch in the spectrum of one of the highest - redshift quasars , sdss j114816.64 + 525150.3 ( @xmath5 ) . a rather strong emission in the redshifted @xmath17 fine - structure line of c@xmath0 at 157.74 @xmath16 m was detected there using the iram 30-m telescope @xcite , in addition to co emission detected earlier @xcite .
here we report on a search for the heh@xmath0(10 ) line ( @xmath2 = 2010.183873ghz , @xcite ) and simultaneously for the ch @xmath3 lines ( @xmath4 2006.8 and 2010.8ghz , @xcite ) in this object . due to the redshift these lines fall into the frequency interval from 270.5 ghz to 271 ghz which can be readily observed at the 30-m iram telescope .
the observations were performed in january 2011 in a pool observing session at the iram 30 m radio telescope in the dual beam switching mode using wobbler switching with a beam throw of @xmath18 towards the nominal source position of @xmath19 , @xmath20 .
the receiver at both polarizations was tuned to the red - shifted heh@xmath0 frequency , equal to 270.954437 ghz .
the system temperature varied from @xmath21 k to @xmath22 k. the total integration time was about 20 hours .
the antenna hpbw at this frequency is about 9 arcsec .
pointing was checked approximately every 1.5 hours and was found to be stable within @xmath23 arcsec .
the results are presented on an antenna temperature scale , @xmath24 , corrected for aperture blocking and for atmospheric attenuation .
we used two spectral backends in parallel : the wilma autocorrelator with a 2 mhz channel width and the filter bank with a 4 mhz channel width .
both backends had 4 ghz total bandwidth .
the data reduction was performed with the gildas software package .
the data set consists of about 250 individual spectral scans . each scan was checked visually ; several apparently corrupted scans were omitted . with the wilma backend
many scans show spurious spikes in a few individual channels .
however these channels are located well outside the expected frequency range for our lines and no special measures were made to exclude these spikes .
on the other hand some spectra obtained with the filter bank suffer from a strong baseline curvature on rather large scales which can affect final results .
an average spectrum obtained with the wilma autocorrelator smoothed to a 16 mhz ( 18 kms@xmath8 ) resolution after subtraction of a zero - order baseline is shown in fig .
[ fig : w - aver ] .
the expected frequencies of the heh@xmath0 and ch lines are indicated .
the shaded area shows the frequency range approximately corresponding to the width of the co and c@xmath0 lines at the half intensity level ( 300 kms@xmath8 ) . + according to the common detection criterion ( @xmath25 )
there is no detected line in this spectrum .
the rms noise in this 16 mhz resolution spectrum is @xmath6 mk ( @xmath7 mjy using a conversion factor of 8.4 jy / k ) . at the same time
one can see a possible emission feature in our spectrum close to the expected heh@xmath0 frequency .
a formal gaussian fit gives for the integrated intensity @xmath26 mk@xmath11mhz or @xmath10 jy@xmath11kms@xmath8 , i.e. the signal to noise ratio is about 2.9 .
the line width is @xmath27 mhz or @xmath28 kms@xmath8 .
the feature is shifted by @xmath29 kms@xmath8 from the expected frequency of the heh@xmath0 line . as mentioned above
this shift is well within the width of the line profiles for co and c@xmath0 . in order to check further the significance of this result
we split the data set into two approximately equal parts and analysed them separately .
the average spectra for these parts are shown in fig .
[ fig:2halves ] .
both spectra show hints for the emission features in the same frequency range .
the same feature is seen in the spectrum obtained with the filter bank ( fig .
[ fig:4 m ] ) although the baseline here is worse than in the wilma spectrum .
we conclude that we have obtained a very tentative detection of an emission line which can be attributed to heh@xmath0 , while there is no sign for ch emission or absorption . + +
what could be the reason for the differences in velocity and width between this tentative heh@xmath0 line and the detected lines of co and c@xmath0 ? as shown by @xcite the heh@xmath0 emission should arise mostly from edges of dense ionized regions . according to these works
heh@xmath0 can be effectively excited by electron impacts .
the required electron density is @xmath30 @xmath31 .
collisional excitation of heh@xmath0 by neutrals requires very high densities and seems to be unlikely .
therefore , our tentatively detected line could be produced in dense ionized regions in this qso .
it would then not be surprising that its central velocity and velocity dispersion are different from those of the neutral molecular gas traced by co and c@xmath0 .
for example , high ionization uv lines in this object have a velocity offset of @xmath32 kms@xmath8 @xcite .
it is worth mentioning also that presumably strong uv and x - ray emissions of this very luminous qso , which probably harbours a @xmath33 m@xmath13 black hole in the center @xcite , create favourable conditions for heh@xmath0 formation .
the total luminosity in the line , if real , can be estimated as @xcite @xmath34 where @xmath35 is the line luminosity in l@xmath13 , @xmath36 is the line rest frequency in ghz , @xmath37 is the luminosity distance in mpc and @xmath38 is the velocity integrated flux density in jykms@xmath8 . using the cosmology calculator @xcite with the `` standard '' cosmological parameters ( @xmath39 kms@xmath8mpc@xmath8 , @xmath40 , @xmath41 ) we obtain @xmath42 mpc .
then , @xmath12 l@xmath13 . until a reliable confirmation the estimates of the line intensity and luminosity
should be considered rather as upper limits .
our data provide no clear detection of either heh@xmath0 or ch in a high - redshift ( @xmath5 ) qso sdss j114816.64 + 525150.3 where a rather strong emission in several atomic and molecular lines has been detected earlier .
however , there is a possible ( at a level of @xmath9 ) emission feature which can be attributed to the heh@xmath0 @xmath43 line .
this feature is shifted by about 100 kms@xmath8 from the expected frequency of the heh@xmath0 line derived from the co and c@xmath0 line redshifts and is about 4 times narrower than these lines .
this shift is well within the width of the line profiles for co and c@xmath0 .
the difference in velocities as well as in the line widths can be explained by quite different conditions required for formation and excitation of heh@xmath0 on the one hand and co as well as c@xmath0 on the other hand .
the heh@xmath0 emission could probably arise in dense ionized gas in this qso .
the velocity integrated flux in this tentative line is @xmath10 jy@xmath11kms@xmath8 which corresponds to the total luminosity in the line of @xmath12 l@xmath13 .
prior to a reliable confirmation , these values should be considered rather as upper limits .
we are very grateful to the iram director , pierre cox for providing director s discretionary time for this project and to the iram staff , especially to manuel gonzalez and clemens thum for their help and advices with the observations .
we thank sergey levshakov for attracting our attention to this object .
bertoldi f. , cox p. , neri r. , carilli c. l. , walter f. , omont a. , beelen a. , henkel c. , fan x. , strauss m. a. , menten k. m. , 2003 , a&a , 409 , l47 bovino s. , tacconi m. , gianturco f. a. , galli d. , 2011 , a&a , 529 , a140 cecchi - pestellini c. , dalgarno a. , 1993 , apj , 413 , 611 dubrovich v.k .
, 1997 , a&a , 324 , 27 dubrovich v.k .
, lipovka a.a .
1995 , a&a , 296 , 307 gulin m. , salom p. , neri r. , garca - burillo , s. , graci - carpio j. , cernicharo j. , cox p. , planesas p. , solomon p. m. , tacconi l. j. , vanden bout p. , 2007 ,
a&a , 462 , l45 liu x .- w . , barlow m. j. , dalgarno a. , tennyson j. , lim t. , swinyard b. m. , cernicharo j. , cox p. , baluteau j .-
, pequignot d. , nguyen - q - rieu , emery r. j. , clegg p. e. , 1997 , mnras , 290 , l71 maiolino r. , cox p. , caselli p. , beelen a. , bertoldi f. , carilli c. l. , kaufman m. j. , menten k. m. , nagao t. , omont a. , wei a. , walmsley c. m. , walter f. , 2005 , a&a , 440 , l51 matsushima , f. , oka t. , takagi k. , 1997 , phys .
lett . , 78 , 1664 pickett h. m. , poynter r. l. , cohen e. a. , delitsky m. l. , pearson j. c. mller h. s. p. , 1998 .
rad . transfer , 60 , 883 riechers d. a. , walter f. , cox p. , carilli c. l. , weiss a. , bertoldi f. , neri r. , 2007 , apj , 666 , 778 roberge w. , dalgarno a. , 1982 , apj , 255 , 489 solomon p. m. , downes d. , radford s. j. e. , barrett j. w. , 1997 , apj , 478 , 144 wagg j. , wilner d. j. , neri r. , downes d. , wiklind t. , 2005 , apj , 634 , l13 walter f. , bertoldi f. , carilli c. , cox p. , lo k. y. , neri r. , fan x. , omont a. , strauss m. a. , menten k. m. , 2003 , nature , 424 , 406 walter f. , carilli c. , bertoldi f. , menten k. , cox p. , lo k. y. , fan x. , strauss m. a. , 2004 , apj , 615 , l17 wei a. , downes d. , neri r. , walter f. , henkel c. , wilner d. j. , wagg j. , wiklind t. , 2007 , a&a , 467 , 955 willott c. j. , mclure r. j. , jarvis m. j. , 2003 , apj , 587 , l15 wright e. l. , 2006 , pasp , 118,1711 | we performed a search for the heh@xmath0 @xmath1 line ( @xmath2 = 2010.183873ghz ) and simultaneously for the ch @xmath3 lines ( @xmath4 2006.8 and 2010.8ghz ) toward one of the highest - redshift quasars known , sdss j114816.64 + 525150.3 ( @xmath5 ) .
no clearly visible line was detected after obtaining an rms noise level of @xmath6 mk ( @xmath7 mjy ) in 16 mhz ( 18 kms@xmath8 ) channels . at a level of @xmath9
, however , there is a tentative emission feature shifted by about 100 kms@xmath8 from the expected frequency of the heh@xmath0 line .
this shift is well within the width of the line profiles for co and c@xmath0 .
the putative feature is about four times narrower than the previously detected co and c@xmath0 lines .
the difference in velocities as well as in the line widths could be explained by quite different conditions required for formation and excitation of heh@xmath0 with respect to co and c@xmath0 . the heh@xmath0 emission , if real , could probably arise in the dense ionized gas of this qso .
the velocity integrated flux in this tentative feature is @xmath10 jy@xmath11kms@xmath8 which corresponds to a total luminosity of @xmath12 l@xmath13 .
as long as there is no independent confirmation , these values should be considered rather as upper limits .
[ firstpage ] astrochemistry radio lines : galaxies galaxies : high - redshift galaxies : ism quasars : emission lines
ism : molecules |
grb s temporal variability played a major role in the understanding how grbs operate .
standard external shocks , in which the external medium slows the relativistic ejecta , can not produce efficiently variable light curves @xcite .
while internal shocks resolve the variability and agree with other properties of grb light curves ( e.g. @xcite ) they suffers from several drawbacks .
first and foremost is their low efficiency ( @xcite see however , @xcite ) .
this is particularly troublesome in view of the high efficiency implied from comparison of the prompt @xmath0-rays luminosity and the kinetic energy that remains in the outflow .
detailed models for the emission mechanisms of the prompt @xmath0-rays pose other problems @xcite .
external shocks can produce highly variable light curves if the outflow is slowed down by small external clumps .
each clump producing a short pulse .
however , this process will inevitably be inefficient @xcite as the overall covering factor of the emitting regions is @xmath1 ( @xmath2 and @xmath3 are the pulses and the burst s durations ) .
observed values of @xmath4 are typically @xmath5 and in can be as low as @xmath6 @xcite .
lyutikov & blandford ( 2002 , 2003 ) ( see also @xcite [ l05 ] , @xcite , @xcite ) proposed that variability can be recovered while maintaining high efficiently it the shell that moves with a bulk lorentz factor @xmath7 contains emitting clumps ( see fig .
[ fig : kinemticmodel ] ) that move with random macroscopic relativistic velocities ( with a lorentz factor @xmath8 ) .
a clump is observed only when its radiation cone ( with an opening angle of the order of @xmath9 in the lab frame ) points towards the observer .
the filling factor of the clumps may be unity , recovering high efficiency .
however , as only a small fraction of the clumps are observed at any given time , the light curve can show rapid variability .
the overall duration of the burst , is the larger between the angular time and the shell s light crossing time ( @xmath10 ) , where @xmath11 and @xmath12 are the shell s radius and width respectively and @xmath13 is the light speed , allowing for emission radii much larger than @xmath14 , required in the internal shocks model .
the temporal variability is then dictated by the random lorentz factor , @xmath8 , reflecting the activity of the emitting region and not those of the inner engine .
while it is unclear how macroscopic random relativistic motion can be generated , we assume that it does and examine , using two simplified toy models that includes the essential ingredients , the conditions under which the temporal features of the observed light curve can be produced ( see l05 ) .
we describe our first toy model , which we call here relativistic turbulence , and derive analytic constraints and numerical light curves in [ sec : the - relativistic - turbulence ] . in [ sec : fundamental - emitters ] we consider a second toy model proposed by lyutikov ( 2006 ) that is based on sub - jets and compare it with the first one .
we summarize the results and compare both models with observations in [ sec : conclusions ] .
move relativistically ; right : sub - jets - energy from regions of size @xmath15 is extracted into relativistic jets .
, scaledwidth=50.0% ] our ( l05 ) kinematic toy model for relativistic turbulence considers a shell which is divided into discrete randomly distributed emitters that have randomly oriented relativistic velocities .
the emitters change their direction of motion continuously , as expected in a turbulent medium . since the emitters exhibit a coherent macroscopic motion
, we require that each emitter is causally connected , and that it changes its direction on times longer than the causal time scale but shorter than the shell crossing time .
the length scale of the emitter in its own rest frame , @xmath16 , is assumed to be similar in all three dimensions , and it emits isotropically in this frame .
we define a dimensionless scale @xmath17 , which is the angular scale of an emitter that points towards the observer ( see inset of fig [ fig : kinemticmodel ] ) .
the emitters radiate as the shell moves from @xmath18 to @xmath19 . due to the turbulent motion the positions and directions of the emitters change with time .
we model this by a set of successive shells between @xmath18 and @xmath19 .
each new shell is constructed with randomly distributed emitters , representing the random changes in direction of the turbulent motion .
the time difference between two shells is , @xmath20 , the time it takes for the emitters to turn an angle of @xmath21 ( in the shell frame ) .
note that there are three frames : the lab frame ; the shell s frame , denoted by a prime , which is boosted radially with a lorentz factor @xmath7 relative to the lab ; and the frame of each emitter , denoted by two primes , which is boosted by ( randomly oriented ) @xmath8 relative to the shell frame .
the observer is , of course , at rest relative to the lab frame .
however , the observer time , namely the arrival time of photons ( denoted @xmath22 ) differs from the lab time by the usual time of flight arguments @xcite .
the doppler shift from an emitter is : @xmath23^{-1 } , \label{eq : doppler shift}\ ] ] where @xmath0 , @xmath24 and @xmath25 are the lorentz factor , velocity and the angle between the velocity and the line to the observer ( both in the lab frame ) .
the flux that reaches the observer from this emitter is : @xmath26 where @xmath27 is the specific intensity and the second relation holds for a small enough emitter ( @xmath28 is the distance to the observer ) .
an implicit k correction arises from the difference between the @xmath29 and @xmath30 .
define @xmath31 as the angle between the line that connects the origin and the emitter and the line that connects the origin and the observer ( see inset of fig [ fig : kinemticmodel ] ) .
the maximal doppler boost , @xmath32 , is obtained for emitter at @xmath33 that moves along the line of sight .
the flux decreases like @xmath34 plus a k correction .
therefore , we consider emitters only if @xmath35 .
what is then the probability that an emitter at angle @xmath31 is observed , @xmath36 ?
at @xmath33 it is @xmath37 while @xmath38 .
this suggests that @xmath39 scales as @xmath40 ( l05 ) , implying that the average probability that an emitter will be visible from an arbitrary position on the shell is : @xmath41 the factor 0.3 was evaluated numerically ( l05 ) and can be ignored at the accuracy level of our discussion .
the arrival time from an emitter at @xmath42 is : @xmath43 where @xmath44 is the distance of the emitter from the front of the shell .
as the last photons will arrive from @xmath19 , an angle of @xmath45 and @xmath46 , the overall duration of the burst will be a function only of @xmath12 and @xmath7 ( and not @xmath8 ) : @xmath47 where we define @xmath48 .
as the shell is expected to expand relativistically in its own frame @xcite but this might not be relevant here . ]
for @xmath50 the shell s width , as well as @xmath3 , are determined by the engine activity time while for @xmath51 they do nt .
the duration of a pulse arriving from a single emitter is the longest of the three following time scales :
\(i ) the duration over which the emitter points towards the observer , namely the duration over which the direction of motion varies by an angle @xmath9 in the lab frame ( @xmath52 in the shell s frame ) .
as the emitter is confined to the shell it should make at least a @xmath53 turn during @xmath54 , implying that the time to turn by @xmath52(shell s frame ) , @xmath20 , is shorter than @xmath55 .
causality puts a lower limit on @xmath20 of @xmath56 .
therefore : @xmath57 in the observer s frame , this translates to : @xmath58 \(ii ) the emitter s light crossing time in the lab frame ( in the direction along the line of sight ) . for an emitter moving towards the observer
this time is @xmath59 .
( iii ) the angular time scale at the largest possible angle , where the emitter is still visible by the observer , @xmath60 , the time difference between the first and the last photon would be @xmath61 .
overall ( ii ) and \(iii ) are of the same order and much larger than ( i ) .
thus : @xmath62 using eqs .
[ eq : burst - duration ] and [ eq : peak - duration ] we express , @xmath63 , the number of ( possibly overlapping ) pulses expected in a burst : @xmath64 where @xmath65 is the occupation number of pulses at any given observer time ( i.e. , @xmath66 implies many overlapping pulses while @xmath67 implies long quiescent periods between isolated pulses ) .
the number of emitters is @xmath68 .
the emitters obtain new random directions ( which differ by more than @xmath52 , in the shell s frame , than the previous ones ) after a time @xmath20 .
thus , the total number of independent emitters , @xmath69 , is larger by the factor @xmath70 , the ratio of the total duration over which the radius doubles and @xmath20 .
finally we introduce a filling factor @xmath71 allowing for the possibility that not all emitters are active all the time or that space is not fully covered by emitters ( @xmath72 is strongly disfavored as the efficiency is always smaller than @xmath73 ) .
overall we find : @xmath74 the condition @xmath75 yields : @xmath76 and using [ eq : tau ] : @xmath77 we demand @xmath78 since many overlapping pulses reduce the observed variability , whereas very frequent long quiescent times between pulses are not observed .
if the shell is in the freely expanding phase ( i.e. , @xmath79 ) @xmath65 will be of order unity if : @xmath80 where @xmath81 is between 2 and 3 .
@xcite have pointed out that @xmath82 if one requires that the emitters are of the maximal causally allowed size .
note that @xmath65 depends quite sensitively on @xmath83 and it increases rapidly if @xmath84 is smaller than @xmath85 .
this implies , for example , that a significant number of small eddies , which may arise in a turbulent cascade , may be problematic . using the relations [ eq : emitter - size ] and [ eq : num - of - pulses ] , and assuming the causal limit for @xmath20 : @xmath86 leading to @xmath87 for typical values of @xmath88 .
note that while the model determines @xmath8 it does not constrain @xmath7 and @xmath11 .
[ cap : lightcurvesf ] depicts simulated light curves ( l05 ) for four choices of parameters .
the two upper panels have @xmath89 with different emitter sizes .
both light curves are highly variable and densely filled with non - overlapping pulses .
however , as @xmath51 , the underlying overall envelope of the pulses is seen .
as the emitters are smaller on the right panel it has more pulses than the left one .
the envelope is observed since only a small fraction of the volume and hence fewer pulses are seen early on .
similarly at @xmath90 pulses from small @xmath31 values are not seen , implying that only lower amplitude pulses ( on average ) are observed during the last @xmath91 of the burst .
the envelope is stretched on bottom panels where @xmath92 .
the lower left panel depicts a very low @xmath65 with a rather sparse light curve .
the lower right panel depicts a light curve of a wide shell and @xmath93 , which is rather similar to observed bursts . for @xmath94 ( not shown )
the pulses are overlapping and all variability is erased , leaving only the envelope .
numerical monte carlo simulations of light curves for different combinations of parameters ( shown on each frame ; the scalings used eliminate the dependence on @xmath11 and @xmath7 ) .
the flux of each pulse is calculated assuming that the radiation efficiency is constant per unit mass for all emitters in their rest frame , namely that @xmath95 , therefore @xmath96 .
we approximate each pulse as a gaussian with the above parameters and we sum over all contributions to construct a light curve . in both upper frames @xmath97 and @xmath51 . in the top
left , the emitters are as large as causality allows while on the top right they are 1/2 of this value and the filling factor is lowered to compensate . in both the overall envelope
is seen clearly .
the sparsity of pulses is apparent when @xmath98 ( bottom left ) and the `` straightening out '' of the envelope is clearly seen ( bottom two panels ) when @xmath92 .
, scaledwidth=50.0% ]
motivated by reconnection in highly magnetized outflow lyutikov ( 2006 ) considers a model in which relativistic sub - jets ( sjs ) are accelerated to a lorentz factor @xmath8 by dissipation of the bulk energy in many different `` mini - engines '' within the relativistically expanding shell .
these `` mini - engines '' or acceleration sites correspond to reconnection sites within the magnetized flow ( e.g. , @xcite ) .
the `` mini - engines '' are at rest in the shell frame .
each mini - engine operates for a time @xmath99 .
the directions of the accelerated sub - jets are random in the shell frame but the opening angle and direction of each is constant while its mini - engine is active . the sub - jet extracts energy from a region of size @xmath100 , where @xmath101 is the speed of extraction of energy from the surrounding region ( relativistic reconnection suggests @xmath102 , @xcite ) .
the observed duration is : @xmath103 where we define the dimensionless parameter @xmath104 . slightly generalizing @xcite we write the probability to observe an emitter as @xmath105 , where @xmath106 and @xmath107 is the sub - jet opening solid angle .
following the notation of [ sec : the - relativistic - turbulence ] , we define , @xmath73 , the filling factor of regions from which energy is extracted into the sub - jets ( @xmath108 ) .
the occupation number of observed pulses is : @xmath109 the condition @xmath110 yields : @xmath111 and @xmath112 this implies that an efficient ( @xmath113 ) highly variable burst requires either a large @xmath8 or a wide shell ( for @xmath114 , @xmath115 . ) .
this constant direction of the emitters and the fact that causality in the shell s frame determines the sub - jet size , @xmath116 , are the main kinematic differences between the sub - jet and the turbulence model ( in which the emitter s direction varies and causality in the emitter s frame determine its size , @xmath16 ) .
for the same @xmath65 and @xmath1 the two models give similar light curves .
in particular , an overall ( rising and falling ) envelope for the light curve is expected in the sub - jet model as well .
we have derived conditions on the parameters of relativistic random emitters needed for producing variable grb light curves .
this is characterized by @xmath117 which ensures that typical pulses do nt overlap and are not too sparse either .
our numerical simulations show that for @xmath118 one obtains light curves that resemble observed grbs ( see fig .
[ cap : lightcurvesf ] ) .
the resulting light curves do not change qualitatively when we introduce a distribution of turbulent lorentz factors and sizes .
causality suggests , for relativistic turbulence , that the relation @xmath119 between the angular size of the emitters , @xmath84 , and the turbulent and the bulk lorentz factors holds naturally @xcite .
but , this condition holds when the turbulent eddies are of the maximal possible size and may be broken by cascade to lower scales .
the condition , @xmath120 arises in the sub - jets model . for high efficiency ,
negligible sub - jet opening angle assuming @xmath121 this reduces to @xmath122 or @xmath123 . while this is similar to the one obtained in the turbulent model , here
there is no apparent physical motivation for proportionality between @xmath99 and @xmath52 and this requires an ad hoc fine tuning . in both models the light curves arising from a single expanding shell with @xmath79
shows a rising and falling underlying envelope .
furthermore , a single shell can not produce bursts which depict long quiescent periods .
the envelope can be erased if @xmath124 , while quiescent periods require an outflow of several shells ( where naturally @xmath124 ) .
these solutions become marginal in the turbulent model if @xmath20 is determined by causality , since @xmath125 requires @xmath126 ( see eq .
[ eq : gamma ] ) .
the sub - jet model , however , may favor @xmath124 as it reduces the required value of @xmath8 .
it seems that with proper conditions ( and rather reasonable in the case of the relativistic turbulence ) these models can produce ( efficiently ) the observed highly variable grb light curves .
we turn now to several shortcomings .
first and foremost is the question how such macroscopic relativistic motions can be generated and sustained .
one needs to convert @xmath127 of the initial total energy to the kinetic energy of the emitters and further dissipation in the emitters frame is needed to generate the radiation .
additional questions involve the shape and other properties of individual pulses versus those seen in observed pulses : ( i ) grbs show a clear difference between the fast rise and the slow decline of individual pulses @xcite . the light curve of an individual pulse results from a combination of the motion of the emitter , its orientation relative to the observer , its width as well as intrinsic inhomogeneities within the emitter . in
the relativistic turbulence model the emitter was radiating long before its velocity pointed towards the observer and it continues to emit long after it moves away from the observer .
there is no reason ( on average ) for a difference between the rising and falling phases of an individual pulse will result in a strong signature differentiating between early and late phases of the overall light curve , which is not observed . on the other hand non - systematic variations ( e.g. , deceleration and acceleration )
are expected to result in similar affects on the temporal structure of rising and decaying parts of pulses . ] .
this is not a problem in the sub - jets model in which the onset of the pulse corresponds to the beginning of the activity of the emitter .
( ii ) the temporal structure of the first and second halves of grb light curves are similar @xcite .
the light curves produced in the two models have an overall envelope that favors stronger pulses earlier and weaker ones later .
this might be resolved by a combination of several emitting shells or with very wide shells , but here fine tuning is required in the turbulent model in order to keep @xmath125 .
( iii ) weaker and denser pulses ( arriving from emitters not moving directly towards the observer ) continues at @xmath128 producing the typical envelope of high latitude emission @xcite .
this is consistent with rapid declines seen in some the early afterglows .
however in many cases the decline is faster . in the internal shocks model
this is attributed to the dominant contribution of a late pulse , that shifts the zero point of the time .
such an option does not arise here unless , once more , we allow for several shells or a single wide shell .
( iv ) the duration of an observed pulse is correlated with the preceding interval @xcite .
there is no reason that such correlation should appear in both models .
( v ) these models predict a doppler induced correlation between the intensity and @xmath129 . while stronger peaks are typically harder , it is not clear whether this specific relationship is satisfied .
we could not find obvious modifications that will address all these issues . while it is not clear that those can not be found , this suggests that the simple versions of these models might not be enough .
a simple extension of a wide shell @xmath124 or several separated shells might resolve some of the issues and it might be essential for the sub - jet model allowing moderate values of the sub - jet s lorentz factor .
nakar , e. , & piran , t.2002,a mnras , 330 , 920 nakar , e. , & piran , t. 2002b , mnras , 331 , 40 nakar , e. , & piran , t. 2002c , ap .
j. lett . , 572 , l139 narayan , r. , & kumar , p. 2008 , arxiv:0812.0018 norris , j. p. , nemiroff , r. j. , bonnell , j. t. , scargle , j. d. , kouveliotou , c. , paciesas , w. s. , meegan , c. a. , & fishman , g. j. 1996 , ap . j. , 459 , 393 | randomly oriented relativistic emitters in a relativistically expanding shell provides an alternative to internal shocks as a mechanism for producing grbs variable light curves with efficient conversion of energy to radiation . in this model the relativistic outflow is broken into small emitters moving relativistically in the outflow s rest frame .
variability arises because an observer sees an emitter only when its velocity points towards him so that only a small fraction of the emitters are seen by a given observer .
models with significant relativistic random motions require converting and maintaining a large fraction of the overall energy into these motions .
while it is not clear how this is achieved , we explore here , using two toy models , the constraints on parameters required to produce light curves comparable to the observations .
we find that a tight relation between the size of the emitters and the bulk and random lorentz factors is needed and that the random lorentz factor determines the variability .
while both models successfully produce the observed variability there are several inconsistencies with other properties of the light curves .
most of which , but not all , might be resolved if the central engine is active for a long time producing a number of shells , resembling to some extent the internal shocks model . |
the one - dimensional nonlinear schrdinger ( nls ) equation in the repulsive or self - defocusing case is usually written as @xcite @xmath1 ) dependences of the wave function @xmath2 are suppressed .
this equation sustains the following dark and grey solitons @xcite : @xmath3 , \end{aligned}\ ] ] with @xmath4 , \\ \phi(x - ct)&=&\tan^{-1}[-2 \xi / c \hskip 0.05 cm \tanh\-\{\xi ( x - ct)\ } ] , \\
\xi & = & \sqrt{(2\eta - c^2)}/2 , \end{aligned}\ ] ] where @xmath5 is the velocity , @xmath6 is the chemical potential , and @xmath7 is related to intensity .
soliton ( [ ds ] ) having a notch " over a background density is grey in general .
it is dark if density @xmath8 at the minimum . at zero velocity
the soliton becomes a stationary dark soliton or a black soliton : @xmath9 $ ] .
the similarity of the nls equation ( [ nls ] ) to the mean - field gross - pitaevskii ( gp ) equation describing a trapped zero - temperature bose - einstein condensate ( bec ) @xcite imply the possibility of a stationary dark soliton , or a black soliton , in a trapped bec or in a degenerate trapped boson - fermion mixture ( dbfm ) @xcite .
dark solitons represent a local minimum in a trapped bec and have been the topic of both experimental @xcite and theoretical @xcite investigations . both dbfm @xcite and
degenerate fermion - fermion mixture ( dffm ) @xcite have been experimentally observed and studied theoretically @xcite .
it has been suggested that the black soliton of a harmonically trapped bec could be a stationary eigenstate @xcite of the gp equation @xcite as in the case of the trap - less nls equation .
we demonstrated that the black soliton in a trapped zero - temperature bec @xcite or a dbfm @xcite is the first stationary vibrational excitation of the nonlinear dynamical equation for the respective systems .
no such vibrational excitation of the nls equation ( [ nls ] ) exists in the absence of a harmonic trap . in this paper
we study the numerical simulation of a black or a stationary dark soliton in a trapped dffm .
we use a coupled time - dependent mean - field - hydrodynamic model for a dffm and consider the formation of black solitons in a quasi - one - dimensional cigar - shaped geometry using numerical and variational solutions .
the present model is inspired by the success of a similar model used recently in the investigation of collapse @xcite , bright @xcite and dark @xcite solitons in a dbfm and of mixing - demixing in a dffm @xcite .
the black soliton of a dffm is demonstrated to be a stationary vibrational excitation of the mean - field - hydrodynamic model .
we suggest a scheme for numerical simulation of a stationary dark soliton in trapped dffm by time evolution of the linear harmonic oscillator equation starting with the analytic vibrational excitation , while the nonlinearities corresponding to the mean - field - hydrodynamic equations are slowly introduced .
the simulation proceeds through successive eigenstates of the dynamical model . to illustrate the stability of our scheme we study the oscillation of the stationary dark soliton upon application of different perturbations @xcite .
collective excitations in the form of solitons and vortices in trapped fermions have also been investigated recently by damski et al .
@xcite and karpiuk et al .
@xcite . however , these authors considered isolated ultra - cold fermions and not a realistic dffm as in the experiments and as discussed in this paper .
also they did not demonstrate the existence of stable dark solitons with a zero at the center of the notch as in the present study .
they identified grey soliton - like structure with a shallow dip in the isolated fermionic density distribution quite distinct from the stable fermionic dark solitons in a dffm as noted in this investigation .
though the present dark solitons are stable in the mean - field formulation , they could be unstable physically due to quantum fluctuations @xcite .
the effect of quantum fluctuations is lost in the mean - free model and can only be studied in a field - theoretic approach . moreover being an excited state they are thermodynamically unstable .
there have been suggestions about how to excite a dark soliton by phase imprinting method @xcite .
the dark soliton is the lowest vibrational excitation of the mean - field model @xcite and there have also been investigations about how to attain such excited states @xcite . nevertheless , despite different suggestions about how to excite a dark soliton @xcite , experiments to date
have not yet generated a stationary dark soliton .
experimentally , so far the dark solitons have been unstable @xcite
. however , considering that they are stationary excitations of the mean - field - hydrodynamic equation their creation _ must _ be possible at least as a non - stationary dark soliton which may turn grey and oscillate before decaying due to quantum fluctuations and thermodynamic effects .
the plan of the paper is as follows . in sec .
ii we present the mean - field - hydrodynamic model which we use in the present study .
we also perform a reduction of the model equations to a quasi - one - dimensional form in a cigar - shaped geometry . in sec .
iii we present a variational analysis of the mean - field equations . in sec .
iv we present the numerical results for stationary dark solitons and compare with the variational results .
the stationary nature of the dark solitons is illustrated through a numerical study of their dynamics under different perturbing forces .
finally , we present a summary in sec . v.
to develop a set of practical time - dependent mean - field - hydrodynamic equations for a dffm , we consider the following lagrangian density @xcite @xmath10 where @xmath11 is the mass of component @xmath12 , @xmath13 a complex probability amplitude , @xmath14 the real probability density , @xmath15 the number of respective atoms , @xmath16 @xmath17 , and @xmath18 is the confining trap . here
the interspecies coupling is @xmath19 with the reduced mass @xmath20 and @xmath21 is the interspecies scattering length .
the interaction between intra - species fermions in spin - polarized state is highly suppressed due to the pauli blocking term @xmath22 and has been neglected in eq .
( [ yy ] ) and will be neglected throughout .
the kinetic energy terms in this equation @xmath23 are derived from a hydrodynamic equation for the fermions @xcite and contribute little to this problem compared to the dominating pauli blocking term in eq .
( [ yy ] ) . however , the inclusion of the kinetic energy terms in eq .
( [ yy ] ) leads to a smooth solution for the probability density everywhere @xcite . to keep the algebra simple and without losing generality , in our calculation
we shall take equal fermion masses : @xmath24 this simulates well the fermion mixtures in two hyperfine states of @xmath25li and @xmath26k atoms observed experimentally @xcite .
the lagrangian density of each fermion component in eq .
( [ yy ] ) is identical to that used in refs .
@xcite .
the mean - field dynamical equations for the system are just the usual euler - lagrange ( el ) equations @xcite @xmath27 where @xmath28 are the three space components . with lagrangian density ( [ yy ] )
the following el equations of motion are derived in a straight - forward fashion @xcite : @xmath29\psi_j=0 , \nonumber \\
j \ne k=1,2 .
\end{aligned}\ ] ] this is essentially a time - dependent version of a similar time - independent model used recently for fermions @xcite . for a stationary state eqs .
( [ e ] ) yield the same result as the formulation of ref .
for a system with large number of fermions both reduce to @xcite the well - known thomas - fermi approximation @xcite : @xmath30^{3/2}$ ] , with @xmath31 the chemical potential .
we reduce eq .
( [ e ] ) to a minimal quasi - one - dimensional form in a cigar - shaped geometry where the confining trap of anisotropy @xmath32 has the form @xmath33 with @xmath34 the radial vector and @xmath35 the axial vector . for a cigar - shaped geometry @xmath36
, we consider solutions of eq .
( [ e ] ) of the type @xmath37 where @xmath38 , @xmath39 , @xmath40 and @xmath41 is the circularly symmetric ground - state linear harmonic - oscillator wave function . in the @xmath36
limit the quasi - one - dimensional functions @xmath42 satisfy the following dimensionless equations @xcite : @xmath43{\phi}_{{j}}(x,\tau)=0,\quad j\ne k = 1,2,\end{aligned}\ ] ] where @xmath44 and @xmath45 in eq .
( [ n ] ) , the normalization condition is given by @xmath46 . in case when @xmath47 , the two equations ( [ n ] ) become identical and reduce to @xmath48{\phi}(x,\tau)=0 , \nonumber \\\end{aligned}\ ] ] which permits the solution @xmath49 , with @xmath50 and @xmath51 .
next we present a variational analysis of eq .
( [ o ] ) based on the stationary trial wave function @xcite @xmath53,\ ] ] where @xmath54 is the amplitude , @xmath55 is the width , @xmath56 the chirp , and @xmath5 the phase .
the lagrangian density for eq . ( [ o ] ) is the one - term version of eq .
( [ yy ] ) , e.g. , @xmath57 + \biggr|\frac{\partial \phi_v}{\partial x } \biggr|^2 + \frac{1}{2 } \gamma n ^2 + \frac{3}{5}\beta n^ { 5/3}\nonumber \\ & + & x^2 n\end{aligned}\ ] ] which is evaluated with this trial function and the effective lagrangian @xmath58 becomes @xmath59 where @xmath60 and the overhead dot denotes time derivative .
the variational lagrange equations @xcite @xmath61 where @xmath62 stands for @xmath5 , @xmath54 , @xmath55 and @xmath56 can then be written as @xmath63 @xmath64 the constant in eq .
( [ con ] ) is fixed by the normalization condition .
eliminating @xmath65 from eqs .
( [ x ] ) and ( [ y ] ) we obtain @xmath66 the use of eqs .
( [ con ] ) , ( [ z ] ) and ( [ eli ] ) leads to the following differential equation for the width @xmath55 : @xmath67 . \nonumber
\\\end{aligned}\ ] ] the quantity in the square bracket is the effective potential of the equation of motion .
small oscillation around a stable configuration is possible when there is a minimum in this potential .
the variational result for width @xmath55 follows by setting the right - hand - side of eq .
( [ min ] ) to zero corresponding to a minimum in this effective potential , from which the variational profile for the soliton can be obtained @xcite using eq .
( [ v ] ) .
the above variational analysis can be extended to the asymmetric case .
however , the algebra becomes quite involved if we take a general variational trial wave function with chirp and phase parameters .
as we are interested mostly in the density profiles , we consider the following normalized stationary trial wave function for fermion component @xmath69 of eq .
( [ n ] ) @xmath70 , \quad j=1,2.\ ] ] using essentially the lagrangian density ( [ yy ] ) in this case we obtain the following effective lagrangian @xmath71 the variational lagrange equations ( [ la ] ) for @xmath72 and @xmath73 now become @xmath74 equations ( [ cp ] ) can be solved for the variational widths @xmath75 and consequently the variational profile of the wave functions obtained from eq .
( [ vx ] ) . when @xmath52 , in eq .
( [ cp ] ) @xmath76 , @xmath77 and @xmath78 ; and in this case it is verified that eq .
( [ cp ] ) yields the same variational widths as from the result obtained in the symmetric case in sec .
iiia given by eq .
( [ min ] ) ; e. g. , @xmath79 by a variational calculation we establish the stability of a black soliton in a trapped bec satisfying the one - dimensional gp equation : @xmath80{\phi}(x,\tau)=0.\ ] ] this equation is identical with eq .
( [ o ] ) with @xmath81 and @xmath82 .
hence the variational analysis of sec .
iiia applies to this case and the differential equation for width @xmath55 is given by eq .
( [ min ] ) with @xmath81 : @xmath83 the variational width in this case is given by setting the right - hand - side of this equation to zero .
of bec from the numerical solution of eq .
( [ p ] ) ( denoted num ) vs. @xmath84 for ( a ) @xmath85 and ( b ) @xmath86 .
the corresponding variational solutions obtained from eq .
( [ q ] ) ( denoted var ) are also shown .
, title="fig : " ] of bec from the numerical solution of eq .
( [ p ] ) ( denoted num ) vs. @xmath84 for ( a ) @xmath85 and ( b ) @xmath86 .
the corresponding variational solutions obtained from eq .
( [ q ] ) ( denoted var ) are also shown .
, title="fig : " ]
we solve eqs .
( [ n ] ) for bright solitons numerically using a time - iteration method based on the crank - nicholson discretization scheme elaborated in ref .
we discretize the coupled partial differential equations ( [ n ] ) using time step @xmath87 and space step @xmath88 in the domain @xmath89 .
the second derivative in @xmath84 is discretized by a three - point finite - difference rule and the first derivative in time by a two - point finite - difference rule .
we start a time evolution of eqs .
( [ n ] ) setting the nonlinear terms to zero , and starting with the eigenfunction of the linear harmonic oscillator problem : @xmath90 @xmath91 during the time evolution the nonlinear terms are switched on slowly and the time evolution continued to obtain the final converged solutions .
in addition to solving the coupled equations ( [ n ] ) , we also solved the single equation ( [ o ] ) in the symmetric case : @xmath52 and eq .
( [ p ] ) for bosons . from the numerical solution of eq .
( [ o ] ) ( denoted num ) vs. @xmath84 for a dffm with @xmath92 , @xmath93 , @xmath94 , while @xmath95 , and @xmath96 . the corresponding variational solution obtained from eq .
( [ q ] ) ( denoted var ) is also shown . ] in our numerical study we take @xmath97 @xmath6 m and consider a dffm consisting of two electronic states of @xmath26k atoms . for a bec we consider @xmath98k atoms .
this corresponds to a radial trap of frequency @xmath99 hz .
consequently , the unit of time is @xmath100 ms . for another fermionic atom the @xmath101 value
gets changed accordingly for @xmath97 @xmath6 m . and
( b ) @xmath102 from the numerical solution of eqs .
( [ n ] ) ( denoted num ) for a dffm with @xmath92 , @xmath103 , @xmath104 , while @xmath105 , @xmath106 , @xmath107 , and @xmath108 . the corresponding variational solutions ( denoted var ) obtained from eqs .
( [ cp ] ) are also shown.,title="fig : " ] and ( b ) @xmath102 from the numerical solution of eqs .
( [ n ] ) ( denoted num ) for a dffm with @xmath92 , @xmath103 , @xmath104 , while @xmath105 , @xmath106 , @xmath107 , and @xmath108 . the corresponding variational solutions ( denoted var ) obtained from eqs .
( [ cp ] ) are also shown.,title="fig : " ] the results for black solitons in a bec are obtained by solving eqs .
( [ p ] ) and ( [ q ] ) and are presented in figs . 1 ( a ) and ( b ) for @xmath85 and 100 , respectively , where the soliton profiles are displayed .
the variational profiles lie close to the numerical results and confirm the existence of a black soliton in this case . for smaller values of nonlinearity ,
the solitons have a probability distribution close to the ansatz given by ( [ v ] ) .
consequently , the variational result is more accurate . for larger nonlinearities the actual probability distribution obtained from the numerical solution deviates from ansatz ( [ v ] ) as can be seen from fig . 1 ( b ) .
by construction the variational ansatz ( [ v ] ) is stationary .
the close agreement between the variational and numerical solutions demonstrates the stationary nature of these dark solitons .
hence , contrary to comments in the literature @xcite , these black solitons of a trapped bec are stationary .
although , we illustrate the present simulation for a black soliton for a repulsive bec , stationary dark soliton also exists for an attractive bec @xcite .
the numerical results for black solitons in a dffm are obtained by solving eqs .
( [ o ] ) and ( [ n ] ) for the symmetric ( @xmath52 ) and asymmetric ( @xmath68 ) cases , respectively .
the corresponding variational results are obtained from eqs .
( [ sim ] ) and ( [ cp ] ) , respectively .
first we present the numerical and variational results for the symmetric situation with @xmath104 , @xmath109 , and @xmath110 in fig .
the nonlinearities here are quite large , e.g. , @xmath111 and @xmath112 , @xmath113 .
the variational and numerical results in this case compare favorably as can be seen in fig .
next we present the numerical and variational results for the asymmetric case with @xmath92 and @xmath103 .
@xmath104 , and @xmath114 corresponding to even larger nonlinearities @xmath105 , @xmath115 , @xmath116 , and @xmath117 the variational and numerical results in this case for black solitons for components 1 and 2 are exhibited in fig .
3 ( a ) and ( b ) . from figs .
1@xmath03 we see that although the variational result is a good quantitative approximation to the exact result for small values of nonlinearity ( figs .
1 ) , it ceases to be so for larger values of nonlinearity ( figs . 2 and 3 ) .
this is true , as for large values of nonlinearity the harmonic - oscillator ansatz for the soliton profile ( [ v ] ) is not a good approximation to the accurate numerical result . in other words
, the variational calculation confirms a stationary black soliton in all cases .
a poor initial variational harmonic - oscillator ansatz for the soliton is responsible for the imprecise variational results for large nonlinearities . at times
@xmath118 @xmath119 and @xmath120 after the soliton has been set into oscillation by displacing the trap through a distance of 2 @xmath6 m . ] m. ] to demonstrate the robustness of the black solitons , we introduced a displacement in the notch of the black soliton upon formation by displacing the center of the harmonic trap by a small distance .
the dark soliton(s ) execute harmonic oscillation for small displacements during a moderate time interval . for the soliton of fig .
2 with @xmath110 the oscillation is illustrated in fig .
4 by exhibiting successive snapshots of the dark soliton at positions of maximum displacement at times @xmath121 and @xmath120 for a trap displacement of 2 @xmath6 m at time 0 .
the soliton remains black at all times with the minimum at the notch equaling zero .
the simple harmonic motion of the position of the notch is illustrated in fig .
5 . the angular frequency of oscillation as obtained from fig .
5 is the trap frequency 500 hz .
the numerical simulations presented in figs .
4 and 5 demonstrate the stationary nature of the black solitons under small oscillations . at times
@xmath122 0 , 40 ms , 80 ms , 120 ms , and 160 ms after the soliton has been set into breathing oscillation by increasing the strength of the trap by @xmath123 to further demonstrate the stationary nature of the black solitons , after their formation we increased suddenly the strength of the harmonic trap by @xmath124 : @xmath125 . the stability of the dark soliton for @xmath126 under this sudden change in the trap frequency is exhibited in fig . 6 where we plot the soliton profile at different times .
the soliton executes reasonably stable breathing oscillation . to study the breathing oscillation in detail we calculate the root - mean - square ( rms ) size @xmath127 of the soliton at different times from a numerical solution of the mean - field equation and the result
is plotted in fig .
we also solved the variational equation ( [ min ] ) , an ordinary differential equation by a fourth - order runge - kutta rule , to calculate the size of the soliton and the time variation of the results is plotted in fig . 7 ( b )
the amplitude and frequency of the variational oscillation are in good qualitative agreement with the precise numerical results .
however , there is quantitative disagreement between the two . after multiplying the variational size by 0.89 , the results in fig .
7 ( b ) are in good agreement with the numerical result in fig . 7 ( a ) .
the period of breathing oscillation from fig . 7
( a ) is 7.3 ms and that from fig .
7 ( b ) is 7.6 ms .
the fair agreement between the variational and numerical results of breathing oscillation illustrates that the variational description of the stationary dark soliton is on the right track and confirms the stationary nature of the black soliton . of the dark soliton
after suddenly increasing the strength of the trap by 10 @xmath128 : ( a ) numerical result , ( b ) variational result .
, title="fig : " ] of the dark soliton after suddenly increasing the strength of the trap by 10 @xmath128 : ( a ) numerical result , ( b ) variational result .
, title="fig : " ]
in conclusion , we used a coupled mean - field - hydrodynamic model for a dffm to study the formation of black solitons in a quasi - one - dimensional geometry by numerical and variational methods .
the existence of stationary dark solitons is predicted by variational study and confirmed by detailed numerical analysis .
the stationary nature of the present solitons is demonstrated numerically through their ( a ) sustained oscillation initiated by a sudden displacement of the harmonic trap , as well as ( b ) breathing oscillation initiated by a sudden increase of the strength of the harmonic trap by @xmath124 : @xmath129 .
here we used a set of mean - field equations for the dffm .
a proper treatment of a degenerate fermi gas or dffm should be done using a fully antisymmetrized many - body slater determinant wave function @xcite as in the case of scattering involving many electrons @xcite . however , in view of the success of a fermionic mean - field model in studies of collapse @xcite , bright @xcite and dark @xcite soliton in a dbfm and of mixing and demixing in a dffm @xcite , we do not believe that the present study on bright solitons in a dffm to be so peculiar as to have no general validity .
f. dalfovo , s. giorgini , l. p. pitaevskii , and s. stringari , rev .
mod . phys . * 71 * , 463 ( 1999 ) ; v. i. yukalov , laser phys .
* 1 * , 435 ( 2004 ) ; v. i. yukalov and m. d. girardeau , laser phys
* 2 * , 375 ( 2005 ) .
j. denschlag _
_ , science * 287 * , 97 ( 2000 ) ; b. p. anderson _
et al . _ ,
* 86 * , 2926 ( 2001 ) ; s. burger _
et al . _ ,
lett . * 83 * , 5198 ( 1999 ) ; s. stock , b. battelier , v. bretin , z. hadzibabic , and j. dalibard , laser phys .
* 2 * , 275 ( 2005 ) .
r. dagosta , b. a. malomed , and c. presilla , phys .
a * 275 * , 424 ( 2000 ) ; n. g. parker , n. p. proukakis , c. f. barenghi , and c. s. adams , j. phys .
b * 37 * , s175 ( 2004 ) ; n. p. proukakis , n. g. parker , d. j. frantzeskakis , c. s. adams , j. opt .
b * 6 * , s380 ( 2004 ) ; p. j. y. louis , e. a. ostrovskaya , y. s. kivshar , j. opt .
b * 6 * , s309 ( 2004 ) ; v. a. brazhnyi , v. v. konotop , phys .
a * 68 * , 043613 ( 2003 ) .
f. schreck _
et al . _ ,
lett . * 87 * , 080403 ( 2001 ) ; a. g. truscott _
et al . _ ,
science * 291 * , 2570 ( 2001 ) ; z. hadzibabic _
et al . _ ,
lett . * 88 * , 160401 ( 2002 ) ; g. modugno _
et al . _ ,
science * 297 * , 2240 ( 2002 ) ; c. ospelkaus , s. ospelkaus , k. sengstock , and k. bongs , phys .
* 96 * , 020401 ( 2006 ) .
k. molmer , phys .
rev . lett . * 80 * , 1804 ( 1998 ) ; r. roth , phys .
a * 66 * , 013614 ( 2002 ) ; t. miyakawa , t. suzuki , and h. yabu , _ ibid . _ * 64 * , 033611 ( 2001 ) ; x .- j .
liu , m. modugno , and h. hu , _ ibid . _ * 68 * , 053605 ( 2003 ) ; x .- j .
liu and h. hu , _ ibid . _ * 67 * , 023613 ( 2003 ) ; m. modugno _
_ , _ ibid _ * 68 * , 043626 ( 2003 ) ; d. m. jezek , m. barranco , m. guilleumas , r. mayol , and m. pi , _ ibid _ * 70 * , 043630 ( 2004 ) . p. k. biswas and s. k. adhikari , j. phys .
b * 33 * , 1575 ( 2000 ) ; * 31 * , l737 ( 1998 ) ; * 31 * , 3147 ( 1998 ) ; * 31 * , l315 ( 1998 ) ; s. k. adhikari , phys . rev .
c * 19 * , 1729 ( 1979 ) ; l. tomio and s. k. adhikari , phys .
c * 22 * , 28 ( 1980 ) . | employing a time - dependent mean - field - hydrodynamic model we study the generation of black solitons in a degenerate fermion - fermion mixture in a cigar - shaped geometry using variational and numerical solutions .
the black soliton is found to be the first stationary vibrational excitation of the system and is considered to be a nonlinear continuation of the vibrational excitation of the harmonic oscillator state .
we illustrate the stationary nature of the black soliton , by studying different perturbations on it after its formation .
instituto de fsica terica , unesp @xmath0 so paulo state university , 01.405 - 900 so paulo , so paulo , brazil |
_ 1.1 illustration and context _ figure [ fig : 1 ] illustrates a simulation of the distribution of small particles suspended in a three - dimensional random flow : the particles are modelled as points , but they are shown as small spheres to make them visible in the figure .
the suspended particles do not interact ( that is , the motion of each particle is independent of the coordinates of the other particles , the equations of motion are given below ) .
we show the initial configuration ( figure [ fig : 1]*a * ) , and two snapshots of the particle positions after a long time , with differing values of the fluid viscosity , ( figures [ fig : 1]*b * and * c * ) . in one case
the particles aggregate , in the sense that the trajectories of different particles coalesce . in the other
their distribution shows some degree of clustering , but their trajectories never coalesce . in this paper
we present an analysis of the transition between aggregating and non - aggregating phases , which we term the path - coalescence transition. illustrating the aggregation of non - interacting particles in a random three - dimensional flow : the motion is defined by equations ( [ eq : 1.1 ] ) to ( [ eq : 1.3 ] ) , ( using a simplified form explained in section [ sec : 3 ] : see equation ( [ eq : 3.23 ] ) ) .
the left panel shows the initial configuration at @xmath0 , the center and right panels show final configurations ( at @xmath1 ) for two different values of @xmath2 . in the case shown in the center panel , trajectories coalesce , until eventually all particles follow the same trajectory .
the particles in the right panel exhibit density fluctuations , but the trajectories do not coalesce .
, title="fig:",width=156 ] illustrating the aggregation of non - interacting particles in a random three - dimensional flow : the motion is defined by equations ( [ eq : 1.1 ] ) to ( [ eq : 1.3 ] ) , ( using a simplified form explained in section [ sec : 3 ] : see equation ( [ eq : 3.23 ] ) ) .
the left panel shows the initial configuration at @xmath0 , the center and right panels show final configurations ( at @xmath1 ) for two different values of @xmath2 . in the case shown in the center panel , trajectories coalesce , until eventually all particles follow the same trajectory .
the particles in the right panel exhibit density fluctuations , but the trajectories do not coalesce .
, title="fig:",width=156 ] illustrating the aggregation of non - interacting particles in a random three - dimensional flow : the motion is defined by equations ( [ eq : 1.1 ] ) to ( [ eq : 1.3 ] ) , ( using a simplified form explained in section [ sec : 3 ] : see equation ( [ eq : 3.23 ] ) ) .
the left panel shows the initial configuration at @xmath0 , the center and right panels show final configurations ( at @xmath1 ) for two different values of @xmath2 . in the case shown in the center panel , trajectories coalesce , until eventually all particles follow the same trajectory .
the particles in the right panel exhibit density fluctuations , but the trajectories do not coalesce .
, title="fig:",width=156 ] there are numerous experimental observations that when small particles are suspended in a complex and apparently random flow , their density becomes non - uniform .
clustering of particles into regions of high density has been observed in experiments on particles floating on the surface of liquids with a complex or turbulent flow @xcite , and also turbulent flow in channels @xcite .
the conditions under which this occurs are not yet fully understood .
the aggregation effect illustrated in figure 1 is an extreme form of clustering .
there appears to be less experimental work on aggregation , but coalescence of suspended water droplets is clearly very important in the formation of rain drops in clouds @xcite . even less is known about aggregation . in particular
it is not clear when a model of particles suspended in a random flow can exhibit aggregation , and when additional physical phenomena ( such as differential drift velocities under gravitational forces , or brownian diffusion ) must be invoked .
earlier theoretical work on clustering in random flows has used fokker - planck equations determining moments of the particle density of advected particles @xcite .
this passive scalar approach does not allow for the effects of inertia of the particles .
it has been supplemented by a perturbative analysis of the deviations of particles from advected trajectories , as proposed by @xcite .
these two approaches are combined in papers by @xcite and @xcite .
numerical work indicates that clustering in solenoidal flows occurs most readily when the correlation time of the velocity field is comparable with the time constant associated with the viscous drag @xcite .
the aggregation ( as opposed to clustering ) of particles by random flows has received relatively little attention .
@xcite appears to have been the first to propose that particles subjected to a smooth random flow can coalesce , and showed numerical evidence that this can happen in one dimension .
he argued that there is a transition between coalescing and non - coalescing phases , and identified the dimensionless parameter which determines the phase transition in one dimension .
this paper describes results obtained from a new approach to characterising particle aggregation in random flows .
it is based upon calculating a lyapunov exponent describing the rate of separation of nearby particles from a solution of a system of stochastic differential equations : the lyapunov exponent is the expectation value of one of the variables . in the limit
as the correlation time of the flow approaches zero , the stochastic differential equations can be reduced to a pair of langevin equations . our results for three - dimensional flows which are discussed here build upon earlier work by two of the present authors in one dimension @xcite ( where we solved deutsch s model exactly ) and two dimensions @xcite .
the three - dimensional case is most important for physical applications , but involves substantial additional technical complications .
we remark that @xcite also considered the two - dimensional case , and quotes an analytic expression for the maximal lyapunov exponent .
his expression is incorrect in two dimensions , because the generating function that he uses is divergent for the equilibrium distribution .
his calculation can be adapted to give the correct expression in the one dimensional case , as quoted in @xcite .
the definition of lyapunov exponents is explained by @xcite , who also discuss the method we use for extracting the lyapunov exponents from direct numerical simulations .
we note that in the case where inertia of the particles can be neglected , our results reduce to calculating the lyapunov exponent for a spatially correlated brownian motion , which was discussed from a mathematical point of view by @xcite and @xcite .
_ 1.2 the model _
the most natural model for theoretical investigation is the motion of spherical particles ( radius @xmath3 , mass @xmath4 ) moving in a random velocity field @xmath5 with specified isotropic and homogeneous statistical properties .
the particles are assumed not to affect either the flow , or each other s motion , and to experience a drag force given by stokes s law : the force @xmath6 on a particle moving with velocity @xmath7 relative to the fluid is @xmath8 , where @xmath9 is the viscosity of the fluid .
we will simplify the problem by assuming that the particles are made of a material which is much denser than the fluid in which they are suspended : this enables us to neglect the inertia of the displaced fluid .
accordingly , we consider a large number of suspended particles , initially with random positions and zero velocity , having equations of motion @xmath10\,.\end{aligned}\ ] ] the random velocity field @xmath5 could be either externally imposed ( for example , if a gas is driven by an ultrasonic noise source ) , or self - generated ( as in the case of turbulence ) .
the equations of motion with displaced mass effects included are discussed by @xcite .
our neglect of displaced mass effects is justified for aerosol systems . in order to fully specify the problem
we must define the statistical properties of the random velocity field @xmath5 .
the random force @xmath11 is generated from a vector potential @xmath12 and a scalar potential @xmath13 : @xmath14 the scalar fields @xmath15 have isotropic , homogeneous and stationary statistics .
we assume that these fields are statistically independent , and that they all have the same correlation function , except that the intensity of @xmath13 exceeds that of the other fields by a factor @xmath16 : @xmath17 c(\vert{\bf r}-{\bf r}'\vert , t - t ' ) \ .\ ] ] the random force @xmath18 is characterised by its typical magnitude @xmath19 , and by the correlation length @xmath20 and correlation time @xmath21 of the correlation function @xmath22 .
in the case of a well - developed turbulent flow , the velocity field has a power - law spectrum with upper and lower cutoffs @xcite . if a random velocity field modeling fully - developed turbulence is used in our model , it is most appropriate to take the correlation length and time to be those of the dissipation scale , that is , the cutoff with the smaller length scale .
the system of equations is characterised by three independent dimensionless parameters , which we take as @xmath23 the parameter @xmath24 is a dimensionless measure of the degree of damping , @xmath25 is a dimensionless measure of the strength of the forcing term and @xmath26 measures the relative magnitudes of potential and solenoidal components of the velocity field ( which is purely potential when @xmath27 , and purely solenoidal in the limit as @xmath28 ) . _
1.3 description of our results _
we show that the phase transition is determined by a lyapunov exponent , @xmath29 , describing the separation of nearby particles : their trajectories coalesce if @xmath30 . here
we describe a new and general approach to this problem , reducing the determination of the lyapunov exponent to the analysis of a simple dynamical system , described by a system of ordinary differential equations containing stochastic forcing terms : the lyapunov exponent is found to be proportional to the expectation value of one coordinate .
these stochastic differential equations are derived in section [ sec : 2 ] .
they introduce an apparent paradox : the structure of the equations appears to be identical in two - dimensional and three - dimensional flows , suggesting that the path - coalescence transition is fundamentally the same in two and three dimensions . that would be a very surprising conclusion . in order to demonstrate and illuminate our method , in sections
[ sec : 3 ] and [ sec : 4 ] we pursue the solution of this problem in considerable depth in one limiting case , namely the limit where the correlation time @xmath31 of the random velocity field is small and the random force is sufficiently weak ( strictly , we consider the case where @xmath32 ) . in this limit ,
the system of ordinary differential equations described in section [ sec : 2 ] becomes a system of two coupled langevin equations . in section [ sec : 3 ] we show that there is in fact a difference between the three - dimensional problem and the two - dimensional case studied in @xcite , which is a rather subtle example of the difficulties in applying langevin approaches to nonlinear equations @xcite . in section [ sec : 4 ]
we discuss a perturbation theory for the lyapunov exponent describing the phase transition , expanded in powers of a parameter @xmath33 which is a dimensionless measure of the inertia of the particles .
the perturbation theory is constructed by transforming the langevin equation first into a fokker - planck equation , and then into a non - hermite an perturbation of a three - dimensional isotropic quantum harmonic oscillator .
we are then able to use the harmonic - oscillator creation and annihilation operators @xcite to express the perturbation theory in a purely algebraic form , enabling us to compute the coefficients to any desired order .
we investigate the phase diagram ( the line in parameter space separating coalescing and non - coalescing phases ) , using both monte carlo averaging of the langevin equation and the results of our perturbation theory .
we find that aggregation can only occur if the random flow has a certain degree of compressibility , which increases as the effects of inertia increase , until there is no coalescing phase even for a purely potential flow field .
the analysis in sections [ sec : 2 ] to [ sec : 4 ] considers the case where all of the suspended particles have the same mass @xmath4 and damping rate @xmath2 .
this ideal can be approached quite accurately in model experiments , but in most applications in the natural world and in technology , the suspended particles will have different masses , sizes and shapes . in section [ sec : 5 ] we discuss the effect of dispersion in the distribution of masses in the one - dimensional case : these arguments can be adapted to treat higher dimensions and dispersion of the damping constant @xmath2 .
we argue that path coalescence is not destroyed by mass dispersion ( although of course it is no longer a sharp transition ) . in this paper
we give a quite comprehensive discussion of the the case where the correlation time of the random flow is very short , and the stochastic differential equations derived in section 2 can be approximated by langevin equations .
section 6 discusses how our approach can be extended to other cases .
to determine whether particles cluster together , we consider two nearby trajectories with spatial separation @xmath34 and momenta differing by @xmath35 .
the linearised equations of motion derived from ( [ eq : 1.1 ] ) are @xmath36 here @xmath37 is proportional to the strain - rate matrix of the velocity field , with elements @xmath38 it is convenient to parameterise @xmath34 and @xmath35 as follows : @xmath39 where @xmath40 and @xmath41 are orthogonal unit vectors , which depend upon time .
the parameter @xmath42 is a scale factor : trajectories coalesce if @xmath42 decreases with probability unity in the long - time limit . in the three - dimensional case , we find it convenient to introduce the third element @xmath43 of a time - dependent orthonormal basis so that @xmath44 differentiating ( [ eq : 2.2 ] ) , and substituting the resulting expressions into ( [ eq : 2.1 ] ) gives @xmath45 projecting @xmath46 onto the unit vectors @xmath47 gives the following three scalar equations of motion @xmath48 the last of these equations implies that @xmath49 , so that @xmath50 is proportional to @xmath41 : we write @xmath51 so that the equation for @xmath52 gives @xmath53 the first equation of ( [ eq : 2.6 ] ) indicates that @xmath42 is a product of random variables , and therefore has a log - normal distribution , that is , the logarithm of @xmath42 has a gaussian probability density . in the limit as @xmath54 ,
the mean and variance of @xmath55 are both linear functions of time : @xmath56 where @xmath29 , @xmath57 , @xmath58 and @xmath59 are constants . if @xmath30 , the probability of @xmath55 exceeding any specified value approaches zero as @xmath54 , implying that trajectories of nearby particles almost always coalesce .
the lyapunov exponent @xmath29 is the mean value of the derivative @xmath60 , so that the first equation of ( [ eq : 2.6 ] ) gives @xmath61 now consider the three projections of @xmath62 , as given by equation ( [ eq : 2.5 ] ) : @xmath63 we introduce the notation @xmath64 and note that the statistics of the transformed matrix elements @xmath65 are the same as those of the original elements @xmath66 , because the statistics of the velocity field are isotropic .
using eqs .
( [ eq : 2.6 ] ) to ( [ eq : 2.8 ] ) and @xmath67 to simplify , we find the following equations of motion for the variables @xmath68 @xmath69 finally , the equation for @xmath70 gives @xmath71 eqs .
( [ eq : 2.9 ] ) and ( [ eq : 2.12 ] ) are the principal results of this paper .
( [ eq : 2.9 ] ) shows that the lyapunov exponent ( the sign of which determines whether or not path coalescence occurs ) is given by the expectation value of a random variable @xmath72 of a simple , finite dimensional stochastic dynamical system , described by eqs .
( [ eq : 2.12 ] ) .
this dynamical system is almost completely de - coupled from the other variables : the equations for @xmath72 and @xmath73 do not depend upon @xmath42 , and the vectors @xmath74 only enter these equations through the evaluation of the random matrix elements @xmath75 .
we pointed out that statistics of these elements are independent of the orientation of the orthogonal triplet @xmath76 .
in the two - dimensional case the analysis leading to ( [ eq : 2.12 ] ) proceeds along similar lines , and leads to the same pair of equations @xcite .
the only difference is that the equation ( [ eq : 2.13 ] ) is absent in the two - dimensional case .
this suggests that the expression for the lyapunov exponent @xmath77 should be the same in two and three dimensions .
this would be a surprising conclusion , but it is not obvious how it can be averted .
however , it does prove to be false , as will be demonstrated in the next section for the limiting case where @xmath32 .
let us now consider how to treat eqs . ( [ eq : 2.12 ] ) in the limit where the correlation time @xmath21 of @xmath37 is very short . because the random field @xmath18 is fluctuating very rapidly , the position @xmath78 of a particle at time @xmath79 is independent of the instantaneous value of the force @xmath18 , so that the value if @xmath80 at the position of the particle is statistically indistinguishable from a random sample of the field @xmath81 .
we also assume that the gradients of the fluctuating forces ( the quantities @xmath82 and @xmath83 in equations ( [ eq : 2.12 ] ) ) are sufficiently small that the typical magnitude of the displacement of the variables @xmath72 , @xmath73 occurring during the correlation time @xmath21 is small compared to the typical magnitude of these variables .
this condition is expressed in terms of the dimensionless variables @xmath84 and @xmath85 later in this section . under these conditions of short correlation time and small amplitude , equations ( [ eq : 2.12 ] )
may be replaced by langevin equations . at first sight , this would appear to lead to @xmath86{\rm d}t + { \rm d}f_1 \\
{ \rm d}y_2&=&\biggl[-\gamma y_2-{2\over m}y_1y_2\biggr]{\rm d}t+{\rm d}f_2 \nonumber\end{aligned}\ ] ] where the @xmath87 are increments of a brownian process , satisfying @xmath88 and @xmath89 , for some constant diffusion coefficients @xmath90 . in the two - dimensional case ,
this expectation is correct @xcite , and eqs .
( [ eq : 3.1 ] ) are the appropriate langevin equations . in the three - dimensional case
, we will see that an additional drift term must be added to the second of these equations .
this is a consequence of the fact that , in the three - dimensional case , @xmath73 is a non - linear function of the components of the vector @xmath91 because it is the magnitude of this vector .
this is an example of the difficulties that arise when treating langevin equations involving non - linear functions of noise terms @xcite . in order to determine the correct langevin equations to model ( [ eq : 2.12 ] ) ,
let us consider the integral of the stochastic forcing terms @xmath87 over a time interval @xmath92 which is long compared to @xmath21 , but short enough that the change in the variables @xmath68 occurring in time @xmath92 can be neglected .
we define @xmath93 and find @xmath94 where @xmath95 the calculation of @xmath96 is more subtle .
we have @xmath97 we must take account of the fact that the unit vectors @xmath74 are rotating : we write @xmath98 where @xmath99 are elements of a rotation matrix .
we now write ( [ eq : 3.5 ] ) in the form @xmath100 now consider the rotation of the unit vectors : using ( [ eq : 2.9 ] ) and ( [ eq : 2.13 ] ) we have @xmath101 so that @xmath102 we obtain @xmath103 this yields @xmath104 the langevin equations therefore contain an additional drift term due to the fact that @xmath105 is non - zero : the correct langevin equations in three dimensions are @xmath106{\rm d}t + { \rm d}\zeta_1\ , , \nonumber \\
{ \rm d}y_2&=&\biggl[-\gamma y_2+{d_{31}\over{y_2}}-{2\over{m}}y_1y_2\biggr ] { \rm d}t+{\rm d}\zeta_2\ , \end{aligned}\ ] ] with @xmath107 the diffusion constants @xmath90 were defined in equation ( [ eq : 3.4 ] ) .
note that in two dimensions , however , ( [ eq : 3.1 ] ) remains valid because the term arising from the rotation of @xmath108 is absent .
now consider the evaluation of the diffusion constants in terms of the statistics of the force @xmath18 .
the elements of the force - gradient matrix @xmath109 are @xmath110 where @xmath111 is the `` kronecker @xmath112-symbol '' describing the parity of the permutation of the indices @xmath113 . define @xmath114 then define @xmath115 , @xmath116 , so that @xmath117 we introduce a more convenient dimensionless measure of the relative importance of solenoidal and potential fields @xmath118 and find @xmath119 in the three - dimensional case because @xmath120 .
it is convenient to re - scale the langevin equations into dimensionless form : write @xmath121 and define @xmath122 with these changes of variables , the langevin equations become @xmath123{\rm d}t'+{\rm d}w_1 \\
{ \rm d}x_2&=&[-x_2+x_2^{-1}-2\epsilon x_1x_2]{\rm d}t'+{\rm d}w_2 \nonumber\end{aligned}\ ] ] with @xmath124 eqs .
( [ eq : 3.20],[eq : 3.21 ] ) must be solved to determine the expectation value of @xmath125 in the steady state .
the lyapunov exponent is then given by @xmath126 figure [ fig : 2]*a * compares the lyapunov exponent obtained from a monte carlo simulation of equations ( [ eq : 3.20][eq : 3.22 ] ) with a direct numerical simulation of a random flow described by equation ( [ eq : 1.1 ] ) .
the lyapunov exponents determined from eqs .
( [ eq : 3.20 ] ) and ( [ eq : 3.21 ] ) for @xmath127 and @xmath128 are plotted as red lines .
the results are compared to numerical simulations of ( [ eq : 1.1 ] ) , using a method described in @xcite to determine the lyapunov exponent . because we are concerned with the limit where the correlation time @xmath21 is taken to zero , the random flow was generated using a discrete series of uncorrelated random impulses , acting over a small time step @xmath129 : the impulse @xmath130 at time @xmath131 is taken to be of the form ( [ eq : 1.2 ] ) in terms of scalar fields @xmath132 and @xmath133 satisfying @xmath134 and similarly for @xmath133 .
this implies @xmath135 .
now we discuss the conditions under which the langevin equations ( [ eq : 3.20 ] ) and ( [ eq : 3.21 ] ) are a valid approximation of ( [ eq : 2.12 ] ) and ( [ eq : 2.13 ] ) . for this purpose
it is sufficient to consider the one - dimensional version of equations ( [ eq : 3.13 ] ) , namely @xmath136 ( this equation appears with a different notation in @xcite ) .
the langevin equations are valid provided the changes in the value of @xmath137 over the correlation time @xmath21 is small compared to the typical values of this quantity
. this criterion can obviously only be satisfied if the correlation time is sufficiently short that @xmath138 .
the criterion also requires the stochastic force @xmath139 to be sufficiently weak .
the deterministic part of the velocity , @xmath140 , is positive in the interval from @xmath141 to @xmath142 .
the criterion on the strength of @xmath143 is that the displacement over time @xmath21 should be small compared to the width of that interval , that is @xmath144 . using the fact that @xmath145 , we obtain the following criteria for the validity of the langevin approximation : @xmath146 for completeness , we end this section by mentioning how equations ( [ eq : 3.20 ] ) and ( [ eq : 3.21 ] ) differ in one and -two dimensions .
the one - dimensional case was considered in @xcite : the lyapunov exponent is given by @xmath147 , with @xmath137 satisfying ( [ eq : 3.25 ] ) . in two dimensions ,
as we have already remarked , the term @xmath148 is absent from the second equation of ( [ eq : 3.20 ] ) , and @xmath149 @xcite ) .
we now show how to obtain an asymptotic approximation for the lyapunov exponent using eqs .
( [ eq : 3.20 ] ) and ( [ eq : 3.21 ] ) .
these equations are equivalent to a two - dimensional fokker - planck equation ( @xcite ) for a probability density @xmath150 , of the form @xmath151 here the diffusion constant @xmath152 and the drift velocity is @xmath153 with components @xmath154 and @xmath155 .
we write @xmath156 , and seek a steady - state solution satisfying @xmath157 by perturbation theory in @xmath112 . in order to simplify the application of perturbation theory , it is convenient to make a transformation so that the unperturbed fokker - planck operator @xmath158 is transformed into a hermitian operator . rather than proceeding to the hermitian form directly , we first map the two - dimensional fokker - planck equation to a three - dimensional equation with a rotational symmetry ( we seek a solution which is invariant under rotation ) . after making this transformation , we find that the corresponding hermitian operator in three - dimensional space is the schr " odinger operator of an isotropic three - dimensional harmonic oscillator .
the perturbation analysis can then be performed very easily , using the algebra of harmonic - oscillator raising and lowering operators , described in @xcite . to shorten equations
, we will use a variant of the dirac notation scheme : in summary , functions @xmath3 , @xmath159 are symbolised by vectors @xmath160 , @xmath161 , linear operators are denoted by a hat , e.g. @xmath162 , and the integral over all space of the product of two functions is denoted by the inner product @xmath163 . in the original form , the action of the unperturbed part of the fokker - planck operator on a function @xmath164 is @xmath165 \ .\ ] ] we transform this by defining the action of @xmath166 on a function @xmath167 as follows @xmath168 \nonumber \\ & = & \partial_1[(\partial_1+x_1)p ' ] + { 1\over{x_2}}\partial_2[x_2(\partial_2+x_2)p ' ] \ .\end{aligned}\ ] ] we now consider @xmath169 to be an operator acting in three - dimensional space , with cylindrical polar coordinates @xmath170 .
we identify @xmath171 , and @xmath172 , and take @xmath173 to be a function which is restricted so that it has cylindrical symmetry , being independent of @xmath174 . with this interpretation
, we can add differentials with respect to @xmath175 to the definition of @xmath176 , and write @xmath177+{1\over{r^2 } } \partial_\varphi^2+\partial_z(\partial_z+z ) = \nabla.({\bf x}+\nabla)\end{aligned}\ ] ] which is the fokker - planck operator for isotropic diffusion in three - dimensional space ( with @xmath152 ) , with a drift velocity @xmath178 .
thus we have transformed the two - dimensional fokker - planck equation to a three - dimensional one with a very simple unperturbed velocity field .
it is convenient to work with cartesian coordinates @xmath179 in the three - dimensional space , having the usual relation to the cylindrical polar coordinates @xmath170 .
the fokker - planck equation is then @xmath180 = \hat { \cal f}'p'=[\hat { \cal f}_0+\epsilon \hat { \cal f}_1]p'\end{aligned}\ ] ] where , in cartesian coordinates , @xmath181 has components @xmath182 we now transform the fokker - planck @xmath183 operator so that @xmath176 is transformed into a very simple hermitian operator , by writing @xmath184 we find ( on writing @xmath185 ) @xmath186\ ] ] so that @xmath187 is ( apart from a negative multiplicative factor ) the hamiltonian operator for a three - dimensional harmonic oscillator .
the spectrum of @xmath187 is the set of non - positive integers @xmath188 .
the eigenfunctions of @xmath187 are generated by raising and lowering operators @xcite : @xmath189 the transformed perturbation operator is @xmath190 \ .\end{aligned}\ ] ] instead of solving the fokker - planck equation @xmath191 we attempt to solve @xmath192 , where @xmath193 .
now consider how to obtain the lyapunov exponent from the function @xmath194 .
we have @xmath195 .
we calculate the average of @xmath196 as follows @xmath197 now we change the notation , using a variant of the dirac notation to represent the function @xmath194 by a ket vector @xmath198 . allowing for the possibility that @xmath198 is not normalised , we write @xmath199 where @xmath200 is the ground - state eigenfunction of @xmath201 , given by the function @xmath202 $ ] .
we calculate @xmath198 by perturbation theory : writing @xmath203 we find that the functions @xmath204 satisfy the recursion relation @xmath205 at first sight this appears to be ill - defined because one of the eigenvalues of @xmath201 is zero , so that the inverse of @xmath201 is only defined for the subspace of functions which are orthogonal to the ground state , @xmath200 . however , because all of the components of @xmath206 have a creation operator as a left factor , the function @xmath207 is orthogonal to @xmath208 for any function @xmath209 , so that ( [ eq : 4.14 ] ) is in fact well - defined .
the iteration starts with @xmath210 . the functions @xmath204 should all have rotational symmetry about the @xmath211-axis .
the angular - momentum operator @xmath212 commutes with both @xmath187 and @xmath213 ( that is , @xmath214=0 $ ] and @xmath215=0 $ ] where we use square brackets for the commutator , @xmath216= \hat { \cal a}\hat { \cal b}-\hat { \cal b}\hat { \cal a}$ ] ) . the operators @xmath187 and @xmath213 can therefore be simultaneously reduced to block diagonal form , with blocks labelled by eigenvalues of @xmath217
. we can restrict ourselves to the subspace where the eigenvalue of @xmath217 is zero .
the functions of this subspace are generated from the ground state using transformed raising and lowering operators , defined as follows : @xmath218 the transformed operators @xmath219 and @xmath220 satisfy @xmath221=\hat i$ ] ( where @xmath222 is the identity operator ) , which is the fundamental relation describing harmonic - oscillator raising and lowering operators . expressing @xmath187 and @xmath217 in terms of these operators ,
we find @xmath223 using results from @xcite , we see that both @xmath187 and @xmath217 are linear combinations of harmonic - oscillator hamiltonians , @xmath224 , @xmath225 and @xmath226 . the @xmath227th eigenfunction @xmath228 of a harmonic - oscillator hamiltonian @xmath229 is obtained from its ground state @xmath230 by repeated application of the raising operator @xmath231 : @xmath232 and this eigenfunction has eigenvalue @xmath227 .
thus eigenfunctions of @xmath233 and @xmath234 with zero angular momentum are constructed as follows @xmath235 for @xmath236 and @xmath237 .
the corresponding eigenvalues of @xmath187 are @xmath238 .
the functions @xmath204 are expanded in terms of the @xmath239 , with coefficients @xmath240 : @xmath241 by projecting equation ( [ eq : 4.14 ] ) onto the vector @xmath242 and using the fact that the eigenvectors @xmath243 of @xmath187 form a complete basis , the iteration can be expressed as follows [ for @xmath244 : @xmath245 the matrix elements @xmath246 are readily computed using the algebraic properties of the raising and lowering operators , as discussed in @xcite .
the coefficients @xmath247 are then calculated recursively .
this allows us to obtain the functions @xmath248 .
the lowest order is @xmath249 .
its contribution to @xmath29 vanishes in view of ( [ eq : 4.10 ] ) .
the leading order is @xmath250 the next order , @xmath251 , does not contribute to @xmath29 since @xmath252 does not contain @xmath253 .
in fact , only odd orders contain @xmath253 and thus give non - zero contributions to @xmath29 .
we also find that the denominator in ( [ eq : 4.13 ] ) is unity at all orders .
the final result is : @xmath254 where the first five coefficients @xmath255 are @xmath256 the coefficients in ( [ eq : 4.22 ] ) have a growth which is typical of asymptotic series , as discussed by @xcite .
figure [ fig : 2]*b * shows approximations to the lyapunov exponent @xmath29 for @xmath257 . shown
are six different partial sums of the series ( [ eq : 4.22 ] ) , including terms up to @xmath258 , with @xmath259 . for a given value of @xmath112
, there is an optimal value of @xmath258 , which we term @xmath260 , defined by the criterion that the term in ( [ eq : 4.22 ] ) with index @xmath260 is smallest in magnitude .
the function @xmath261 can be inverted , its inverse @xmath262 is the value of @xmath112 for which the @xmath258 term is optimal . for values of @xmath112
less than the @xmath262 the results are shown as solid lines . beyond this optimal value of @xmath112 , the results are shown as dashed lines .
the results show that the series agrees well with the numerical simulation up @xmath263 , as would be expected for an asymptotic series .
* a * lyapunov exponent as a function of @xmath112 : results from eqs .
( [ eq : 3.20 ] ) , ( [ eq : 3.21 ] ) and ( [ eq : 3.22 ] ) are shown as solid lines , those from direct simulations as symbols , @xmath264 ( @xmath265 ) , @xmath266 ( @xmath267 ) , and @xmath268 ( @xmath269 ) . * b * lyapunov exponent @xmath270 versus @xmath112 for @xmath271 . shown
are results from simulations ( @xmath265 ) as well as from the asymptotic series ( [ eq : 4.22 ] ) summed to orders @xmath272 : up to @xmath262 full lines and from @xmath262 dashed lines .
* c * phase diagram in the @xmath112-@xmath273-plane . shown
are results from numerical simulations of ( [ eq : 2.1 ] ) , @xmath265 , summation of the asymptotic series ( [ eq : 4.22 ] ) summed to @xmath274 , blue line , as well as results from langevin simulations ( red line ) . ]
the phase boundary in the @xmath112-@xmath273-plane is determined by the condition @xmath275 .
figure [ fig : 2]*c * shows this phase boundary as determined from truncating the series ( [ eq : 4.22 ] ) at the optimal order @xmath261 ( blue line ) .
this asymptotic result is shown for values of @xmath112 up to @xmath276 . beyond this range ,
the asymptotic approximation becomes increasingly inaccurate . also shown , in the same plot ,
are results obtained from the langevin equations [ eqs .
( [ eq : 3.20 ] ) , ( [ eq : 3.21 ] ) and ( [ eq : 3.22 ] ) ] , and from direct simulations ( @xmath265 ) .
the results show that the coalescing phase disappears as the effect of inertia is increased : the coalescing disappears altogether for the case of pure potential flow ( @xmath277 ) at @xmath278 . one notable difference between the three - dimensional calculation presented here and the two - dimensional case considered in @xcite is that in the two - dimensional case the phase boundary has a power series in @xmath112 which vanishes identically ( and the phase line is therefore non - analytic ) .
in most naturally occurring aerosols the suspended particles have different mass @xmath4 , and particles of differing sizes will also have different values of the damping coefficient @xmath2 .
it is important to consider whether particles still have a tendency to coalesce even when the particles have differing values of @xmath4 and @xmath2 : we argue that the path coalescence effect is not destroyed by small mass dispersion .
the argument can be adapted to dispersion of the damping coefficient , reaching the same conclusion .
assume that the path - coalescence effect occurs for particles of mass @xmath4 .
compare the motion of this reference particle with that of an initially nearby particle with mass @xmath279 .
the reference particle has equation of motion @xmath280 writing @xmath281 , the equation of motion for the other particle is @xmath282 collecting the terms which are first order in @xmath283 , we obtain a linearised equation of motion for @xmath283 : @xmath284 this is an inhomogeneous differential equation for the separation @xmath283 between two particles , with a driving term proportional to their mass difference @xmath285 . the solution of this equation can be constructed from a green s function satisfying @xmath286 @xmath287 with @xmath288 for @xmath289 .
the solution of equation ( [ eq : 5.3 ] ) is @xmath290 for @xmath291 , equation ( [ eq : 5.4 ] ) is the equation for small displacements of trajectories of particles with the same mass . we know that in the path - coalescing phase the solutions have a negative value of the lyapunov exponent @xmath29 , and that they therefore decay exponentially at large time . in the case
where @xmath292 is bounded by an exponentially decreasing function , such that @xmath293 , equation ( [ eq : 5.5 ] ) remains finite as @xmath54 . for sufficiently large @xmath294 ,
the probability of this inequality being violated is extremely small .
this indicates that in the path - coalescing phase the solution ( [ eq : 5.5 ] ) remains finite as @xmath54 , except for very rare events .
the conclusion is that , when @xmath295 , two initially close particles with nearly equal mass will remain in close proximity for a very long time .
in this paper we described the path - coalescence transition , and showed that the transition point is determined by the change of sign of a lyapunov exponent .
we showed that in general the lyapunov exponent is determined from an expectation value of a variable of a simple dynamical system , equations ( [ eq : 2.12 ] ) , which is driven by stochastic forcing functions .
we considered the solution of these equations in a particular limiting case , where the dimensionless parameters satisfy @xmath32 , by mapping the continuous differential equations into a pair of coupled langevin equations .
we used these langevin equations to produce a rather complete description of the phase transition in that limit .
the remainder of these concluding remarks discuss how equations ( [ eq : 2.12 ] ) can be used in the case where these inequalities are not satisfied . in order to solve these differential equations
it is necessary to characterise the statistics of the stochastic driving terms @xmath65 .
these terms contain information about the strain - rate of the field evaluated at a point along the reference particle trajectory .
there are two possibilities : case a1 : the trajectory @xmath78 is sufficiently slowly moving that the displacement over time @xmath21 is small compared to @xmath20 . in this case
the statistics of @xmath75 are those of a randomly chosen static point , and the correlation time of @xmath65 will be @xmath21 . case a2 : if the trajectory @xmath78 is moving sufficiently rapidly that its displacement in time @xmath31 is large compared to @xmath20 , then the correlation time of @xmath65 will be smaller than @xmath21 because the loss of correlations results primarily from changing the position at which @xmath297 is sampled .
the limit which was investigated in detail in this paper ( @xmath298 ) is an example of case a1 . in cases where @xmath84 and @xmath85 approach different limits however
, all three possibilities can occur in the system described by equations ( [ eq : 1.1 ] ) to ( [ eq : 1.3 ] ) . | we describe a criterion for particles suspended in a randomly moving fluid to aggregate .
aggregation occurs when the expectation value of a random variable is negative .
this variable evolves under a stochastic differential equation .
we analyse this equation in detail in the limit where the correlation time of the velocity field of the fluid is very short , such that the stochastic differential equation is a langevin equation . |
bnard - marangoni convection ( bmc ) in a plane fluid layer with an open surface is the surface - tension - driven analog to buoyancy - driven rayleigh - bnard convection ( rbc ) in a layer with rigid top and bottom walls .
both systems present classic examples of hydrodynamic instabilities , and are generic models for convective flows driven by buoyancy and marangoni forces , respectively .
many works on bmc are devoted to the regular cellular patterns as dissipative structures , which are observed in highly viscous fluids such as silicone oils , e.g. @xcite . in these experimental and theoretical investigations , bmc is either stationary or only weakly time - dependent , and the prandtl number @xmath0 is large compared with unity because of the high kinematic viscosity .
marangoni convection in low prandtl number fluids , i.e. liquid metals or semiconductor melts ( with @xmath0 of order @xmath1 ) , plays an important role in industrial processes such as crystal growth @xcite or electron beam evaporation @xcite .
theoretical and experimental studies as well as numerical simulations have therefore been largely focused on configurations resembling actual industrial setups such as the floating zone @xcite .
by contrast , bmc in liquid metals has received comparatively little attention .
the only experimental work known to the author is the study of ginde et al .
@xcite , where liquid tin was used .
theoretical studies of dauby et al . @xcite and of thess & bestehorn @xcite predict the occurrence of inverted " hexagons at the onset of convection for sufficiently small prandtl numbers .
the orientation of the flow in such hexagonal cells is opposite to that found at high prandtl numbers .
numerical studies of two- and three - dimensional bmc were undertaken by the present author in collaboration with a. thess @xcite . in @xcite , partiulcar attention was paid to the limit of vanishing prandtl number .
earlier studies of rbc had already explored low prandtl numbers , and noted interesting behavior of two - dimensional roll solutions , termed inertial convection @xcite .
the analog of inertial convection in rbc was observed and analyzed in @xcite for bmc .
this type of convection is characterized by the dominance of inertia over viscous forces and a low energy dissipation .
however , the two - dimensional rolls are susceptible to three - dimensional perturbations , and inertial convection is therefore typically not realized in three - dimensional simulations @xcite .
some indications for the existence of inertial convection come from rbc experiments with mercury and sodium @xcite , but those are only based on integral heat flux measurements for a narrow range of prandtl numbers .
in contrast to the non - conducting high - prandtl - number liquids , the bnard - marangoni instability in liquid metals can be influenced by magnetic fields . for a vertical applied field
the linear stability theory has been thoroughly explored by several authors @xcite . in this configuration
the magnetic field does not introduce horizontal anisotropy .
the magnetic field suppresses the instability , i.e. , the necessary temperature difference for convective instability grows with the magnetic induction @xmath2 . for sufficiently strong @xmath2 ,
the vertical structure of the unstable mode exhibits hartmann layers at the bottom and the free surface , and the corresponding critical wavelength shrinks @xcite .
a horizontal component of the magnetic field introduces horizontal anisotropy , but would not suppress the instability because the roll axis of the normal mode can always be aligned with the field . in this case no lorentz force is produced .
the resulting anisotropy of the flow pattern has been discussed in the context of linear theory @xcite , but not on the basis of nonlinear simulations .
the nonlinear behavior with a vertical magnetic field has so far only been examined using perturbation methods , which only apply near the onset of convection @xcite .
the present paper continues the investigations of nonlinear bmc with a vertical magnetic field using the direct simulation approach of @xcite , whereby large marangoni numbers and complex flows can be realized numerically .
both two- and three - dimensional simulations will be performed in order to analyze the qualitative effects of the magnetic field on the particular solutions described in @xcite .
the choice of parameters and boundary conditions is therefore guided by these previous works .
the paper is organized as follows . in the following section the basic equations and
the numerical method will be described .
after that , the two - dimensional and three - dimensional simulation results are presented and analyzed in separate sections .
the paper ends with some conclusions and suggestions for future work .
the system under study is a planar liquid layer of thickness @xmath3 with a free upper surface , which is heated from below .
the liquid has a finite electric conductivity @xmath4 .
buoyancy and surface deflections are neglected .
the isothermal bottom of the layer is located at @xmath5 and the free surface at @xmath6 .
periodic boundary conditions apply in the @xmath7- and @xmath8-directions .
the thickness @xmath3 is chosen as lengthscale for non - dimensionalization . the dimensionless quantities @xmath9 and @xmath10 denote the periodicity length with respect to @xmath7 and @xmath8 . as in @xcite ,
free - slip conditions are assumed at the bottom of the layer in order to compare with the non - magnetic case .
the heat flux density at the free surface is assumed fixed in the present work , which corresponds to the limit of zero biot number .
it is convenient to introduce the deviation @xmath11 from the the conductive temperature profile by writing @xmath12 here @xmath13 is the prescribed bottom temperature and @xmath14 the conductive temperature difference .
notice that @xmath15 since the fluid can not become hotter than the bottom wall .
the shear stresses at the free surface are @xmath16 where the coefficient @xmath17 characterizes the linear decrease in surface tension @xmath18 with temperature .
the symbols @xmath19 , @xmath20 and @xmath21 are the fluid velocity , density and kinematic viscosity .
the vectors in eq .
( [ marangoni1 ] ) are understood to be projected onto the free surface .
the lorentz force and the induced current due to the external homogeneous magnetic field @xmath2 will be treated in the limit of low magnetic reynolds number . as explained in @xcite , this approximation
is justified for length scales and velocities realized in the laboratory and in industrial applications .
it takes into account the action of the magnetic field on the velocity field but ignores the reaction of the flow on the magnetic field .
the densities of induced current and lorentz force in this approximation are given by @xmath22 the divergence of the induced current density @xmath23 vanishes because of the extremly short charge relaxation time in comparison with the time scales of magnetohydrodynamic flows ( * ? ? ?
* section 2.2 ) .
for this reason , the electric potential @xmath24 satisfies @xmath25 it is also assumed that the liquid is bounded by electrical insulators , i.e. the normal component @xmath26 of the current density vanishes at the top and the bottom of the layer . using ohm s law , @xmath27 translates into boundary conditions on @xmath24 .
the equations will be given in a form based on viscous velocity scaling , i.e. @xmath28 is the velocity scale and @xmath29 the time scale .
furthermore , @xmath30 serves as scale of @xmath11 , where @xmath31 denotes the thermal diffusivity of the fluid .
the magnetic field is oriented in along the @xmath32 unit vector , and the scale of the electric potential is given by @xmath33 .
the dimensionless equations are @xmath34 at the bottom @xmath5 , the conditions @xmath35 apply .
the boundary conditions at the top surface @xmath36 are @xmath37 with the marangoni number @xmath38 as control parameter for the forcing .
the remaining parameters are the prandtl number @xmath39 and the hartmann number @xmath40 the scale for @xmath11 has been chosen in such a way that a coupling of velocity and temperature is maintained also in the limit @xmath41 , i.e. when the left hand side in ( [ heatviscous ] ) vanishes . this approximation is called the ( viscous ) zero prandtl number limit . the same limit has been considered in the work of thual @xcite for rbc .
other limiting cases based on different temperature and velocity scalings are also discussed in this work , and it is argued that these other cases either do not provide a uniform approximation throughout the fluid domain , or model transients or flows where thermal convection is merely a side effect .
convective heat transport is measured by the nondimensional nusselt number @xmath42 defined as the ratio between the total heat flux and the conductive heat flux .
in contrast to convection between isothermal plates with fixed temperatures , convection reduces the temperature drop across the layer .
since the contribution of heat conduction to the total heat transport is thereby reduced , one has to compute the conductive heat flux for the present , reduced surface temperature .
the result reads @xmath43 where @xmath44 denotes the mean perturbation of the surface temperature . in the zero prandtl number limit @xmath42 equals unity .
results for @xmath45 can be extrapolated from @xmath41 if one regards the solution of the zero prandtl number equations as the leading term in an expansion in @xmath0 .
the calculations given in @xcite lead to the following results for mean surface temperature perturbation @xmath46 and @xmath47 : @xmath48 in these equations , @xmath49 and @xmath11 are the solutions of the zero prandtl number equations . due to its connection with the nusselt number the quantity @xmath50 ( where the overbar symbol denotes the volume average )
will be referred to as reduced nusselt number .
finally , the alternative scaling of the basic equations based on the units @xmath51 for velocity , @xmath52 for the electric potential and @xmath53 for the temperature perturbation @xmath11 will be needed for later discussion .
the equations ( [ nsviscous]-[heatviscous ] ) then take the form @xmath54,\\
\nabla \cdot{\mbox{\boldmath$v$\unboldmath } } & = & 0,\\ \nabla^2 \phi & = & \nabla \cdot\left({\mbox{\boldmath$v$\unboldmath } } \times { \mbox{\boldmath$e$\unboldmath}}_z\right),\\ \label{heatthermal } \frac{\partial \theta}{\partial t } + ( { \mbox{\boldmath$v$\unboldmath}}\cdot \nabla)\theta & = & \nabla^2 \theta + v_z . \end{aligned}\ ] ] the boundary conditions ( [ bcbotfreeviscous],[bctopviscous ] ) remain unchanged , but for the nusselt number one has to use @xmath55 instead of ( [ newnusselt ] ) .
the numerical code for two- and three - dimensional direct simulations is based on equations ( [ nsviscous]-[heatviscous ] ) and a pseudospectral fourier - chebyshev discretization .
it has been described in @xcite for the non - magnetic case .
the modifications for the present case with vertical magnetic field are analogous to the implementation of the electric potential and lorentz force terms for the channel code discussed in @xcite .
the linear stability of the basic conductive state has been treated in several papers , including asymptotic limits of large hartmann number and biot numbers as well as zero and finite crispation numbers @xcite . for the purposes of the present paper the critical marangoni and wavenumbers for free - slip boundary conditions at the bottom
are required as a function of the hartmann number .
the velocity and temperature perturbations for the neutral stability problem are @xmath56 whereby one obtains the stability equations @xmath57 w(z)=0,\hskip5 mm \left[d^2-k^2\right]\theta(z)=-w(z)\ ] ] and the boundary conditions @xmath58 the symbol @xmath59 denotes the @xmath60-derivative .
notice that the prandtl number is absent from these equations since the neutral conditions are independent of @xmath0 .
because of the linearity of the system ( [ stabilityequations]-[stabilitybcs ] ) , one can choose the normalization @xmath61 to solve these equations without the marangoni boundary condition @xmath62 eq . ( [ ma - crit - eq ] ) is then used to calculate the marangoni number @xmath63 for neutral conditions at the chosen wavenumber @xmath64 .
the solution of the fourth - order equation for @xmath65 and the second - order equation for @xmath66 is obtained simultaenously by a chebyshev collocation method .
critical conditions are found by searching for the minimum @xmath63 for the given value of @xmath67 , which provides the critical parameters @xmath68 and @xmath69 .
the results are shown in fig .
[ fig : linearstability ] for free - slip ( @xmath70 ) and no - slip ( @xmath71 ) boundary conditions at the bottom .
the critical marangoni and wavenumber are smaller for the free - slip case , but the difference vanishes as @xmath67 increases .
the asymptotic relations @xmath72 shown in fig . [
fig : linearstability ] have been obtained by wilson @xcite for the no - slip case and are discussed and interpreted by thess & nitschke @xcite .
they appear to be valid irrespective of the bottom boundary condition on the velocity .
the vertical structure of the neutral modes differs only in a zone of width @xmath73 at the bottom between the free- and no - slip cases . near the free surface
, the unstable modes display a boundary layer structure with a width of order @xmath73 as explained in @xcite .
-13 mm two - dimensional solutions will be considered first as they display several interesting features in the non - magnetic case @xcite . in particular
, large marangoni numbers can be explored more easily than in three - dimensional simulations .
the computational domains are chosen on the basis of the linear stability analysis in the previous section for hartmann numbers @xmath74 and @xmath75 .
for these values we have the critical parameters @xmath76 , @xmath77 and @xmath78 , @xmath79 .
the simulations are therefore performed with a periodicity length @xmath80 , where @xmath81 takes the value corresponding to @xmath74 and @xmath75 .
this way , two roll cells fit into the computational domain at the onset of convection , i.e. for @xmath82 . in order to delay instabilities and to maintain this flow topology at larger values of @xmath83 the mirror symmetry between the rolls
is enforced by the numerical method through symmetry conditions on the fourier coefficients .
simulation results are shown in fig .
[ fig : snapshots - ha20-pr02 ] for @xmath75 and @xmath84 for several values of the marangoni number .
the simulations evolve to a steady flow in all cases shown . near the onset of convection at @xmath85
nonlinear effects are weak , and the flow field reflects the structure of the neutrally stable mode associated with the critical marangoni number .
one can clearly identify the boundary layer at the free surface originating from the lorentz force in the streamline pattern . for @xmath86 ( fig .
[ fig : snapshots - ha20-pr02](b ) ) the vorticity remains no longer confined near the free surface , but is swept into the bulk because of the stronger advection .
a significant difference to the non - magnetic case is the change of sign of the vorticity in the domain indicated by the dashed isolines ( middle column ) .
this change of sign is an indication of the lorentz force impeding the advection of vorticity in the bulk of the liquid layer . for larger values of @xmath83 the advection dominates and the negative vorticity region disappears .
moreover , the streamfunction and vorticity distributions become approximately symmetric with matching isolines near the center of the roll ( fig .
[ fig : snapshots - ha20-pr02](d ) ) . in this case
, the flow approaches an inviscid balance between pressure gradient and advection term in the bulk region since the gradients of streamfunction and vorticity become aligned .
the viscous and lorentz force terms are then weak by comparison with the other terms in the navier - stokes equation .
the tendency of the flow to approach such a nearly inviscid configuration has a profound consequence for the limit of vanishing prandtl number . when the forcing of the flow by the surface tension gradients is strong enough to establish such a nearly inviscid balance , then the approximate cancellation of pressure and advection term imply that only the linear terms control the dynamics in the momentum equation ( [ nsviscous ] ) .
the forcing is provided by the marangoni boundary condition coupling temperature perturbation and shear stress , and the temperature perturbation becomes a linear functional of the velocity field in the limit @xmath41 ( eq . ( [ heatviscous ] ) ) .
one can then expect the velocity to grow exponentially in time since the dynamics is governed by an effectively linear equation with a forcing term proportional to the velocity itself .
this is illustrated by fig .
[ fig : reynoldsvstime ] , which shows the growth of the mean velocity with time for @xmath75 and @xmath87 .
the simulation was started from the steady solution for @xmath88 with the prandtl number set to zero at @xmath89 .
after a short initial transient , the streamfuction and vorticity fields approach the symmetric state with an approximate inviscid balance similar to fig .
[ fig : snapshots - ha20-pr02](d ) , and the spatial rms velocity grows exponentially with time .
this behavior is completely analogous to the non - magnetic case @xcite . for finite prandtl numbers ,
the unbounded growth of the velocity eventually stops since the temperature difference across the layer is reduced by the fast advection , and the shear stress diminishes because of the correspondingly reduced temperature difference on the free surface . because the saturation of the velocity is provided by the heat transport it turns out that the appropriate scaling of the velocity and temperature are provided by thermal units @xmath51 and @xmath53 .
if velocity and temperature perturbations are measured in these units , then these quantities become independent of @xmath0 for @xmath90 , and equations ( [ nsthermal]-[heatthermal ] ) are appropriate . this mode of convection
is the inertial convection mentioned in the introduction .
conversely , when the forcing by the surface - tension gradient is not too strong , then the velocity field does not attain the symmetric streamlines required for the approximate cancellation of pressure gradient and advection term , and the momentum transport is responsible for nonlinear saturation of the convective instability . in this case
, the viscous units provide the appropriate scales for velocity and temperature perturbation in the limit @xmath90 .
this mode of convection has been termed weak convection in ref.@xcite .
the change from viscous to thermal scaling appears in a certain range of marangoni numbers .
[ fig : weaktoinertial ] illustrates the switch from viscous to thermal scaling by the reynolds number , i.e. the spatial rms velocity in viscous units , and the quantity @xmath47 , which scales as @xmath91 when the viscous scaling applies , and which becomes independent of @xmath0 when the thermal scaling holds .
the steady flows were computed for five finite values of the prandtl number from @xmath92 up to @xmath93 and for @xmath41 , and for @xmath74 and @xmath75 with the corresponding domain sizes @xmath80 .
in addition , the computations were repeated for the non - magnetic case ( @xmath94 ) with the same domain sizes @xmath80 as for @xmath74 and @xmath75 . the numerical resolution for the simulations
was adjusted depending on the reynolds number , and convergence was verified for several parameter sets by doubling the number of collocation points in both directions .
typical resolutions at the lowest value of @xmath0 were @xmath95 and @xmath96 collocation points
. the left column of fig.[fig : weaktoinertial ] shows that the spatial rms velocity ( reynolds number ) approaches the limiting curve for @xmath41 when the marangoni number is not too large .
when scaled in thermal units , the rms velocity would tend to zero as @xmath90 , i.e. the curves would look similar to @xmath47 in the right column . for sufficiently large marangoni numbers , the curves for @xmath47 show a tendency to converge onto a limiting curve as @xmath0
is reduced . the rms velocity of the steady solution branch for @xmath41 shows either a very rapid growth over a narrow @xmath83-range or appears to end abruptly for @xmath75 . in the non - magnetic case
there is an additional bifurcation in the @xmath41 solution branch @xcite , which appears to be absent in the four combinations of @xmath67 and @xmath9 investigated here .
this bifurcation allows one to clearly distinguish between the weak convection regime with viscous scaling for @xmath90 and the inertial regime with thermal scaling in the non - magnetic case @xcite . in the present cases ,
the transition from weak to inertial convection at fixed @xmath0 is gradual except for @xmath75 , where a discontinuity is present at sufficiently low @xmath0 .
concerning the apparently different behaviors of the solution branches for @xmath41 one has to bear in mind that these cases are very hard to resolve numerically .
the solutions converge very slowly , and are rather sensitive to the spatial resolution , which must be increased with the reynolds number to resolve the increasingly smaller viscous boundary layers .
the asymptotic behavior with @xmath83 of the branches with @xmath41 can not be reliably determined in the present numerical approach .
the extensive computations summarized by fig . [ fig : weaktoinertial ] were performed for the non - magnetic and magnetic cases with the same periodicity length in order to quantify by how much the lorentz forces delays the transition from weak to inertial convection .
to do so , the transition from weak to inertial convection must be characterized by a typical marangoni number .
for the present purposes , the characteristic marangoni number @xmath97 will be identified by the crossing between the branches @xmath98 for @xmath92 and @xmath99 , i.e. @xmath100 is used to describe the transition from weak to inertial convection .
table [ table1 ] lists the corresponding values for the four cases presented in fig .
[ fig : weaktoinertial ] .
the quantity @xmath101 is larger for the magnetic than for the non - magnetic case , but the relative increase in @xmath102 between @xmath74 and @xmath75 is relatively moderate when compared with the increase in @xmath102 on account of the changed domain size , i.e. comparing the two @xmath102 values for @xmath94 .
the behavior of inertial convection at large values of @xmath83 is characterized by a boundary layer structure of temperature and vorticity fields in the non - magnetic case , which leads to characteristic power laws @xmath103 for the rms velocity @xmath104 ( in thermal units ) and nusselt number @xmath42 .
since the reynolds number is @xmath105 , it becomes very difficult to resolve the small structures in the flow field for small @xmath0 .
the calculations for large @xmath83 have therefore only been done for @xmath93 and for selected values of @xmath83 for @xmath106 .
the results are shown in fig .
[ fig : powerlaws ] for @xmath93 .
the two different domain sizes with @xmath94 show identical power law behavior in agreement with eq .
( [ powerlaws ] ) , but have slightly different prefactors . for non - zero @xmath67 there is no power - law scaling seen for @xmath42 in fig .
[ fig : powerlaws ] . for the rms velocity
there is some indication , but with a smaller exponent than for @xmath94 .
this may , however , be misleading because the temperature field still lacks an isothermal core in the bulk , as can be seen in fig .
[ fig : snapshots - ha20-pr02](d ) .
the absence of power - law scaling for @xmath42 is therefore not surprising .
moreover , the estimates for the kinetic energy dissipation given in ref . @xcite have to be corrected because of the additional lorentz force .
it is not clear if a modified scaling would emerge from an attempt to take the lorentz force into account , which will not be made in this paper .
the lorentz force has another interesting effect on the vorticity distribution in inertial convection . in the non - magnetic case
, one finds an essentially constant vorticity inside the convection roll .
the physical argument for this constant value has been given by batchelor @xcite .
batchelor considers steady flow in a region of closed streamlines with an approximate balance of pressure gradients and advection terms in the momentum equation , i.e. high - reynolds - number flow . in this case ,
convection of vorticity dominates along the streamlines , and the vorticity is therefore approximately constant along streamlines . since the motion is assumed to be steady , viscous diffusion of vorticity across streamlines must vanish because the vorticity distribution would otherwise change with time .
the vorticity therefore can not change between neighboring streamlines , and is therefore constant in the entire region of high speed flow with closed streamlines . in magnetic inertial convection
the additional lorentz force is small when compared with pressure gradient and advection term , but it nevertheless dissipates energy . in a steady state there must therefore be an influx of energy into the region of closed streamlines .
since this influx must be provided by viscous diffusion , the vorticity can not be constant as in the non - magnetic case .
the scatter plots of vorticity and streamfunction shown in fig .
[ fig : psiomega ] demonstrate that this is indeed the case .
the roll center is associated with the highest values of @xmath107 .
viscous effects dominate near the roll boundaries where @xmath108 , and no functional relation between @xmath107 and @xmath109 exists for small values of @xmath107 .
most of this region has therefore been omitted from the scatter plots of fig .
[ fig : psiomega ] by adjusting the @xmath107 axes . for larger values of @xmath107
there is a clear functional relation between vorticity and streamfunction , and @xmath109 changes almost linearly with @xmath107 .
the scatter is somewhat higher for @xmath75 , which is probably due to the lower reynolds number caused by the stronger magnetic damping .
the approximately linear behavior of @xmath109 with @xmath107 can be understood from a modification of the derivation given by batchelor @xcite .
the starting point is the steady momentum ( navier - stokes ) equation in dimensional form , @xmath110 where the lorentz force is denoted by @xmath111 and @xmath112 the ( approximate ) inviscid balance implies that @xmath113 following @xcite , one now introduces orthogonal curvilinear coordinates @xmath114 , where the lines of constant @xmath115 are orthogonal to the streamlines with @xmath116 .
differentials associated with @xmath107 and @xmath115 are @xmath117 and @xmath118 in physical space , where @xmath119 is unknown .
the value of @xmath120 is constant along streamlines ( bernoulli equation ) , and the gradient of @xmath120 is therefore simply @xmath121 .
equation ( [ inviscidbalance ] ) then takes the form @xmath122 whereby the vorticity is a function of @xmath107 only .
the next step is to integrate eq .
( [ ns - dim ] ) around a closed streamline .
the left hand side vanishes since the line element @xmath123 is everywhere parallel to @xmath19 and therefore orthogonal to @xmath124 .
the first term on the right also integrates to zero , i.e. one is left with @xmath125 the curl of @xmath109 is parallel to the streamline because @xmath126 , and the left hand side of eq .
( [ ns - dim2 ] ) simplifies to @xmath127 when the lorentz force is zero , this equation implies that @xmath109 is constant . for two - dimensional convection
the lorentz force is given by @xmath128= -\sigma b^2 v_x { \mbox{\boldmath$e$\unboldmath}}_x\ ] ] since the electric potential vanishes in this case .
the integral on the right hand side of eq .
( [ ns - dim2 ] ) then becomes @xmath129 the integral on the right hand side of eq .
( [ lineintlorentz ] ) represents a fraction @xmath130 of the circulation around the streamline .
one can therefore write eq .
( [ ns - dim2 ] ) as @xmath131 i.e. the local slope of the vorticity - streamfunction dependence is determined by @xmath132 and the prefactor @xmath130 , which depends on the streamline geometry and the velocity distribution along the streamline .
the almost linear decrease of @xmath109 with @xmath107 indicates that @xmath130 changes only weakly with @xmath107 .
as in the two - dimensional case , the focus of the present section is on the comparison with the non - magnetic case considered earlier @xcite .
the shape of the computational domain is therefore chosen in agreement with this previous work , which focused on the smallest rectangular domain compatible with a perfectly hexagonal pattern at the onset of convection .
it has the aspect ratios @xmath133 , @xmath134 with periodic boundary conditions in the @xmath7- and @xmath8-directions . because of the considerable computational expense of three - dimensional simulations covering a significant interval of marangoni numbers
the other parameters are fixed to @xmath41 and @xmath74 .
the corresponding wavenumber is @xmath135 as in the previous section . as in the non - magnetic case ,
the limit @xmath41 is expected to provide a consistent approximation for sufficiently small prandtl numbers @xmath45 .
the simulations were performed by systematically increasing the marangoni number using the final state of previous runs as initial condition . near @xmath82 ,
the marangoni number was also decreased to explore the subcritical range .
some simulations were started from random initial conditions to check for additional solution branches , which may be missed by the continuation approach .
the main results of the simulations are summarized in the plots of fig .
[ fig:3d - bif ] , which show the reduced nusselt number @xmath50 for different solutions as a function of the dimensionless parameter @xmath136 measuring the distance from the linear instability threshold .
the results for the non - magnetic case from @xcite with @xmath137 and @xmath138 are shown for comparison .
two different types of stationary solutions , identified as perfect hexagons ( hx ) and deformed hexagons ( dhx ) are found in both non - magnetic and magnetic cases .
these solutions are found up to @xmath139 for @xmath74 whereas they only exist up to @xmath140 for @xmath94 .
the perfect hexagons have hexagonal symmetry , which is lost in the deformed state . as in the non - magnetic case ,
the hexagonal convection cells are characterized by downflow in the center of the hexagons .
this flow orientation is different from the hexagonal cells at @xmath141 , which have upflow in the center of the hexagon @xcite .
the hexagons exist subcritically down to @xmath142 at @xmath74 and down to @xmath143 at @xmath94 .
time - dependent solutions with oscillatory dynamics are marked by os in fig .
[ fig:3d - bif ] .
as discussed in @xcite , the branch marked by os1 corresponds to an expansion / contraction of the cells in the @xmath7-direction .
the solution branch os2 displays an additional periodic excitation of the mean flow component @xmath144 in the @xmath8-direction .
this solution branch slightly overlaps with os1 .
the os2 branch leads to chaotic dynamics at @xmath145 , which was explored in @xcite .
for @xmath74 there are three different oscillatory branches with regular dynamics . on
the branch marked os1 , the hexagons oscillate in @xmath8-direction and in antiphase about their position as shown in fig .
[ fig : oscill - hex ] . during the oscillation
, the hexagons exchange their @xmath8-position relative to each other , i.e. the oscillation amplitude in the @xmath8 coordinate is large even at the onset . the translatory motion in @xmath8
is accompanied by a size oscillation of the cells . on the os2 branch ,
the oscillatory motion in the @xmath8 direction changes its character .
the hexagons do not return to their initial position but continue their motion in their respective direction . in effect , each cell performs a size oscillation and travels in either the positive or negative @xmath8-direction through the periodic domain . finally , the os3 branch is accompanied by an oscillation of the @xmath8 component of the mean flow . on the os2 branch
the mean flow remains zero .
chaotic dynamics appears beyond @xmath146 , i.e. , @xmath147 .
it is illustrated by fig .
[ fig : chaos ] , which shows as snapshot of the surface temperature perturbation , and the time series of the reduced nusselt number .
in contrast to the non - magnetic case , no overlap could be found for the different oscillatory branches , i.e. the simulations provided a unique oscillatory solution at given marangoni number .
the onset of time dependent flow requires a higher supercriticality parameter @xmath148 .
the same applies for the excitation of the mean flow and for chaos , i.e. the magnetic field provides a considerable damping influence .
details of the transition to chaos have not been explored so far due to the considerable cost of such simulations . for @xmath149 and larger ( @xmath150 ) a numerical resultion of @xmath151 , @xmath152 and @xmath153 collocation points
was needed in order to resolve the structures of the velocity field .
the integration times were of the order @xmath154 with a typical timestep @xmath155 .
the two- and three - dimensional simulations have systematically explored the influence of a vertical magnetic field on low - prandtl - number bmc .
in addition to delaying the instability , the magnetic field also extends the range of marangoni numbers where the zero - prandtl - number limit works for two - dimensional convection .
the transition from weak to inertial convection shows some differences in comparison with the non - magnetic case of ref.@xcite , but the phenomenon itself remains unchanged .
only the asymptotic power - law scaling of inertial convection with @xmath83 could not be detected at finite hartmann numbers , but may nonetheless be present when @xmath83 is increased further .
the novel feature of inertial convection with magnetic field is the modified vorticity distribution due to the magnetic damping .
the three - dimensional simulations were restricted to @xmath41 and @xmath74 .
the appearance of time - dependent flow and chaos is again delayed by the magnetic damping , and the oscillatory solution branches differ from those found at @xmath94 . as already done for @xmath94 , it could be interesting to examine the chaotization in more detail . throughout the paper only the free - slip boundary condition at the bottom was considered .
this was motivated by the observation in ref.@xcite that the free - slip condition permits the existence of stationary rolls up to very large values of @xmath83 , and thereby allows one to study this asymptotic case rather easily .
no - slip conditions will lead to more complicated behavior since additional vorticity is produced at the bottom wall .
however , with the additional magnetic damping the critical parameters for the free- and no - slip cases approach for large values of @xmath67 , and the difference between free- and no - slip could therefore be less significant than at @xmath94 .
the free - slip condition in the present paper was chosen mainly in order to facilitate the comparison with the non - magnetic case .
the work presented here suggests at least two directions for future simulations .
first , the no - slip condition and larger domain sizes should be explored in order to see if the phenomena analyzed here remain relevant in a more realistic configuration .
second , a horizontal magnetic field should lead to interesting pattern dynamics due to the imposed anisotropy .
ultimately , buoyancy and surface deformability should also be included in future numerical work .
lllll @xmath67 & @xmath64 & @xmath82 & @xmath97 & @xmath102 + @xmath156 & @xmath157 & @xmath158 & @xmath159 & @xmath160 + @xmath161 & @xmath157 & @xmath162 & @xmath163 & @xmath164 + @xmath156 & @xmath165 & @xmath166 & @xmath167 & @xmath168 + @xmath169 & @xmath165 & @xmath170 & @xmath171 & @xmath172 + the author acknowledges financial support from the deutsche forschungsgemeinschaft in the framework of the emmy noether program ( grant bo 1668/2 ) .
computer resources were provided by the computing centers of tu ilmenau and tu dresden . | the effect of a homogeneous magnetic field on surface - tension - driven bnard convection is studied by means of direct numerical simulations .
the flow is computed in a rectangular domain with periodic horizontal boundary conditions and the free - slip condition on the bottom wall using a pseudospectral fourier - chebyshev discretization .
deformations of the free surface are neglected .
two- and three - dimensional flows are computed for either vanishing or small prandtl number , which are typical of liquid metals .
the main focus of the paper is on a qualitative comparison of the flow states with the non - magnetic case , and on the effects associated with the possible near - cancellation of the nonlinear and pressure terms in the momentum equations for two - dimensional rolls . in the three - dimensional case ,
the transition from a stationary hexagonal pattern at the onset of convection to three - dimensional time - dependent convection is explored by a series of simulations at zero prandtl number . |
the penning trap @xcite is widely used to confine charged particles in an evacuated region of space through the use of static electric and magnetic fields . a versatile tool for enabling isolation and manipulation of charged particles in a well - controlled environment
, it finds applications in various disciplines , including physics of non - neutral plasmas , mass spectrometry , biomolecular chemistry , precision spectroscopy , antimatter science , quantum information and fundamental metrology.@xcite an iconic example is the creation of an artificial atom called
geonium,"@xcite consisting of a single elementary particle isolated in a penning trap;@xcite the most precise value of the fine structure constant @xmath0 is determined from the spin flip and cyclotron oscillation of one electron in such a system.@xcite a penning trap is made of a stack of cylindrically - symmetric electrodes biased to generate a restoring force along the trap axis . for radial confinement this electrode stack and its vacuum envelope are typically inserted into a coaxial solenoid which can generate a strong magnetic field ( @xmath1 tesla ) along the common symmetry axis . a superconductive magnet , such as used in nuclear magnetic resonance ( nmr ) experiments , can provide very uniform magnetic fields greater than 5 tesla .
this classic architecture separates the electrode structure and the source of magnetic field , allowing the two subsystems to be independently removed , modified or adjusted when necessary . for applications demanding high precision ,
a superconductive solenoid can be designed to provide ( 1 ) field homogeneity as high as 1 part in @xmath2 over 1 @xmath3 , and ( 2 ) self - shielding via flux conservation to screen out ambient field fluctuations.@xcite on the other hand , a superconductive magnet is costly and may not be entirely advantageous for some applications .
nmr magnets can occupy as much as @xmath4 m@xmath5 of space and require some form of refrigeration .
this is not compatible with small instrument development , nor with facilities or missions that have severe space constraints .
rare - earth permanent magnets with high remnant fields have enabled the construction of a variety of compact structures with strong magnetic fields.@xcite compact penning traps have been developed which utilize arrays of radially - oriented magnets ( wedges ) to replace a solenoidal magnet.@xcite in this work we introduce a simplified , unitary architecture for compact penning traps and discuss experiments using such traps for storage of highly charged ions extracted from an ebit ion source . in this unitary architecture , the rare - earth magnets that generate the magnetic field
are naturally integrated within the electrode structure that generates the electric field .
details regarding the slowing and capture of highly charged ions from the nist ebit will be presented in a separate publication .
here we focus upon the design and the observed properties of unitary penning traps with embedded magnets .
section [ sec : designconstruct ] presents two architectures utilizing axially - oriented rare - earth magnets . in section [ sec : observations ] we describe an experimental apparatus deployed at the nist ebit facility and present the first results from tests using stored highly charged ions .
we are interested in isolating ions extracted from the nist ebit for spectroscopic experiments.@xcite space constraints preclude the use of a multi - tesla penning trap@xcite . since compact penning traps using rare - earth magnets occupy less than 1000 @xmath3 of space , they are an attractive alternative .
such compact penning traps have been used to study light ions in a liquid - nitrogen cooled apparatus@xcite and to store molecular anions in a room - temperature apparatus@xcite . for our planned experiments ,
however , the trap must be operable at the high voltages necessary to capture highly charged ions .
very low background gas pressure must be attainable at room temperature , and access must be provided for laser or atomic beams to interact with the stored ions . such constraints and requirements , together with known conditions favorable for ion confinement , have led us to explore non - traditional architectures which could help simplify construction .
the simplest penning trap using one permanent magnet is discussed in sec .
[ sec : onemag ] , illustrating the basic features of a unitary architecture .
this is followed by a more intricate design involving two rare - earth permanent magnets ( sec . [
sec : twomag ] ) to provide better magnetic field homogeneity and radial access for laser or atomic beams .
schematic of the one magnet penning trap ; the symmetry axis is horizontal .
the ndfeb magnet is the central ring electrode , with its axial magnetization @xmath6 indicated by the arrows .
characteristic dimensions of the trapping volume : @xmath7 mm is the inner radius of the ndfeb magnet ; @xmath8 mm is the distance from the midplane to one of the endcaps.,scaledwidth=47.0% ] a simple design for a penning trap using an embedded permanent magnet is shown in the schematic diagram fig . [ fig:1magtrapcad ] . to allow ion passage , open endcap electrodes ( ofhc copper )
are used , separated from the central ring " electrode by insulating spacers made of machinable glass ceramic(macor@xcite ) .
the ring electrode is a composite structure consisting of an axially - magnetized , annular neodymium magnet sandwiched tightly between two annulated copper discs . the annulated copper discs on each side are used to align the magnet with the axis of the electrode stack , and have tapped holes for attaching wires .
rare - earth magnets are fabricated from compounds containing an element in the lanthanide series , such as praseodymium ( @xmath9pr ) , neodymium ( @xmath10nd ) or samarium ( @xmath11sm ) .
the neodymium magnets , chemically denoted by nd@xmath12fe@xmath13b or ndfeb , are chosen for our work because they are readily available in various grades and shapes with triple - layer plating to protect the magnets from corrosion ; the first layer is nickel , followed by copper and finally nickel again .
hence , an electrical potential can be applied to the full surface of the composite ring electrode via one of the press - fitted copper discs .
the integration of the rare - earth magnets into the electrode structure for providing the trapping fields is an essential feature of a unitary architecture .
this coupling of the magnetic and electric components is fully exploited in section [ sec : twomag ] wherein the trap electrodes are made of iron to yoke the fields emanating from the rare - earth magnets .
the dual role played by the ndfeb magnets and electrodes reduces the number of trap components and the overall size of the ion trap while still providing a useful magnetic field of @xmath14 0.317(1 ) t in the trapping region .
the ndfeb magnet ( n42 grade ) has dimensions of 19.05 mm inner diameter , 38.10 mm outer diameter , and 19.05 mm length .
the trapping region has characteristic dimensions of @xmath7 mm and @xmath15 mm .
the overall assembly diameter and length of the trap are @xmath16 mm and @xmath17 mm , respectively .
figure [ fig : bfieldmap ] shows the magnetic flux density ( b - field ) produced by a single ndfeb magnet , computed using quickfield@xcite , a commercial software which implements a well - known finite element method ( fem)@xcite .
the magnetic b - field ( @xmath18 ) has a saddle - point at the center .
the magnet field is 0.317 tesla at the trap center , growing radially to 0.381 t at the inner wall of the magnet . on axis
, the field drops to zero at a distance of 12.94 mm from the center , reversing direction and reaching a local maximum of 0.089 t at 20.75 mm before attenuating with increasing distance .
this field pattern is identical to that of a model based on a pair of solenoids at the inner and outer walls of the magnet , carrying counter - rotating currents.@xcite near the center of the trap the ion motions are well approximated by calculations for an ideal penning trap@xcite wherein a uniform magnetic field is superimposed upon a quadrupole potential . in a plane
perpendicular to a uniform magnetic field , a charged particle will undergo circular motion with a frequency , commonly called the cyclotron frequency , given by @xmath19 where @xmath20 is the charge of the particle , @xmath21 is the magnitude of the magnetic field , and @xmath22 is the mass of the particle . for a bare ne nucleus in a magnetic field of @xmath23 0.3 t , the cyclotron frequency @xmath24 2.3 mhz .
along the trap axis of symmetry ( z - axis ) , the center of cyclotron orbit bounces between the endcaps , which are positively biased relative to the ring electrode ; to lowest order , the electric field is given by a quadrupole potential @xmath25 the field coordinates @xmath26 and @xmath27 are defined from the center of the trap and @xmath28 is determined by trap dimensions @xmath29 where @xmath30 is the inner radius of the ring electrode , and @xmath31 is the distance of reentrant endcaps from the midplane .
the coefficient @xmath32 is the potential difference between the endcaps and the central ring electrode ; it is used interchangeably with @xmath33 in other sections .
the dimensionless parameter @xmath34 ( sometimes denoted c@xmath12)is a geometrical factor of order unity ; if the electrode surfaces near the center closely approximate hyperbola of revolution , then @xmath35 with @xmath36 . small amplitude motion along the trap axis
is described by a simple harmonic oscillator with frequency @xmath37 for the one - magnet trap in fig .
[ fig:1magtrapcad ] with @xmath38 v , an axially - bound @xmath39ne@xmath40 ion oscillates with @xmath41 414.8 khz . for this trap , @xmath42 .
in contrast to the stable axial oscillation , transverse motion is localized only if dynamical equilibrium is possible . the axial restoring force provided by the quadrupole electric field
is accompanied by an outward radial force on the ion , pulling the ion towards the ring electrode .
this tendency to leave the trap radially ( i.e. , to roll off the saddle point of eq.[pennv ] ) must be balanced by the lorentz force due to the magnetic field if the ion is to remain trapped .
consequently , a third component of the ion motion , the magnetron motion , arises from the @xmath43 interaction@xcite . in equilibrium
the cyclotron orbit drifts slowly around the center of the trap , as illustrated in sec.[sec : twomag ] , with a magnetron frequency given by @xmath44 .\ ] ] considering a bare ne nucleus with the trapping parameters described above , @xmath45 38.03 khz .
the condition for equilibrium is more restrictive in weaker magnetic field .
as discussed in ref.@xcite , magnetron motion requires @xmath46 hence , the existence of localized motions is determined by the trap geometry , applied fields and ion properties ( charge and mass ) . in an ideal penning trap ,
if this condition is satisfied at one location , then it holds over the entire volume bounded by the electrodes . in a one - magnet penning trap the allowed region could be reduced by its magnetic field gradient ; useful axial well depth is restricted to a narrower range for a one - magnet trap than for a high - field solenoid system . nevertheless , the trap illustrated in fig .
[ fig:1magtrapcad ] proves sufficient for capture of highly charged ions from an ebit .
we observe ion storage times of order 1 second , limited primarily by background gas collisions(sec .
[ sec : observations ] ) .
the one - magnet penning trap allows straightforward assembly and operation at high voltages , but it has some drawbacks .
the ndfeb magnet blocks laser or atomic beams propagating in the midplane , obstructing access to the stored ions .
since the entire inner surface of the ndfeb magnet is in the line of sight from the trap center , magnet imperfections could degrade ion confinement stability .
furthermore , the uniformity of its magnetic field is not optimized .
a possible solution is to construct a compact penning trap using the classic architecture but with the solenoid magnet replaced by two radially - magnetized rings , each formed from 8 ndfeb wedges.@xcite however , the construction of such magnet arrays requires considerable effort and care to ensure good alignment and axial symmetry .
in addition , adhesives used in bonding the wedges(e.g .
, epoxies ) generally degrade in high temperature bake - outs commonly used in improving base pressure for room - temperature applications . in this section
, we present a simpler alternative : an architecture employing two identical ndfeb magnets ( @xmath6 parallel to their aligned symmetry axes ) and electrodes made of soft iron in an integrated structure for producing the magnetic field .
the design work is guided by fem simulations of magnetic structures using quickfield .
the precision of the algorithm depends upon the coarseness of the grid and the accuracy of the magnetization ( b - h ) curves ( for materials such as ndfeb@xcite , soft iron , etc . ) .
diagram of a penning trap with two embedded ndfeb magnets ; the symmetry axis is horizontal .
the two magnets are seated tightly on opposite sides of the iron ring electrode with inner radius @xmath47 mm .
holes in this ring allow beam access .
the iron endcaps have reentrant tubes extending into the ndfeb magnets and ending at @xmath48 mm from the midplane .
the outer diameter is 5.08 cm , and the overall length is 6.10 cm.,scaledwidth=48.0% ] figure [ fig:2magtrapcad ] shows a diagram of the two - magnet penning trap .
the two open - access endcaps and the central ring electrode are made of electrical iron , with inner surfaces polished to mirror - like finish . to enable radial beam access to the stored ions , the central ring electrode has four equidistant holes , along two orthogonal directions . the top hole in the vertical direction houses an aspheric lens for light collection .
the characteristic trap dimensions are @xmath49 mm ( inner radius of central ring ) and @xmath50 mm ( distance from the midplane to an endcap ) .
the overall assembly diameter and length of the trap are 50.8 mm and 60.974 mm , respectively . near the center of the trap , a bare ne nucleus bound in a @xmath38
v well would oscillate with axial frequency @xmath41 597.04 khz , corresponding to @xmath51 in eq.[pennv ] .
the rare - earth magnets are of the same dimensions as the one used in the one - magnet design ( sec .
[ sec : onemag ] ) , but are manufactured for continuous operation at temperatures up to 453 k or 180 @xmath52c ( n40uh grade ) .
n40uh grade magnets are chosen to withstand higher bakeout temperatures without loss of magnetization .
the two magnets are oriented to have parallel magnetization .
one at a time , the magnets are fitted tightly onto the ring electrode ; since there is strong magnetic attraction between the magnet and the iron ring , an installation device with a threaded rod is used to control the docking process .
a similar installation procedure is used to insert the reentrant endcaps into the magnets .
each endcap is aligned with the central ring electrode directly , using a copper ring and macor spacer in series .
this two - magnet architecture with soft - iron electrodes has several useful features .
first , the alignment of this trap is better than the one - magnet design because it relies only upon fitting precision - machined components .
since the ndfeb magnets are secluded behind the reentrant endcaps , magnet imperfections which could degrade ion confinement are attenuated .
the electrical - iron endcaps help to reduce the fringing magnetic field , and their reentrant shape yokes the field towards the trap center . by redirecting the magnetic flux around the magnets to close the magnetic circuit via the trapping region , the soft - iron endcaps largely account for the strength and uniformity of the magnetic field near the trap center . a comparison of the calculated and measured axial magnetic field is shown in fig . [
fig : 2mbfield ] .
an axial hall probe is used to measure the field along the trap axis .
the hall sensor has dimensions 1.9 mm by 2.3 mm wide , and @xmath53 mm thick ; the overall instrument accuracy is about 2 % , but the resolution is @xmath4 part in 300 in the range of the measurements reported here .
two grades of ndfeb magnets were used : n42 ( red circle ) and n40uh ( green square ) .
figure [ fig : 2mbfield](a ) compares measurements with calculation ( solid line ) for the on - axis magnetic field .
an axi - symmetric model is used , neglecting perturbations due to the 4 holes in the ring electrode .
when magnetization ( bh ) curve parameters are optimized within typical manufacturing tolerance ( 5 % to 8 % ) , @xcite the model calculation fits well to data over the range @xmath54 cm @xmath55 cm .
figure [ fig : 2mbfield](b ) gives a detailed view of the trapping region of the two - magnet penning trap .
the error bars represent @xmath56 uncertainty , combining in quadrature the probe resolution with the uncertainty @xmath57 due to the field gradient . the scatter of measured b - field values in the region @xmath58 mm from the trap center has a standard deviation that is smaller than the sensor resolution .
this suggests that the homogeneity is about 1 part in 300 in the optimized region ; higher uniformity should be obtainable using small shim coils . compared with a one - magnet trap ( dash line ) , the magnetic field homogeneity in the two - magnet trap is better within @xmath58 mm from the trap center .
( a)comparison of the on - axis magnetic field calculated for the one - magnet trap ( dash line ) and the two - magnet trap ( solid line ) ; ( b)magnification for finer comparison of the homogeneity within the trapping region .
the reentrant edges of the trap endcaps are indicated by a pair of vertical lines ; green dashed line for the one - magnet trap and blue solid line for the two - magnet trap .
hall probe measurements for the two magnet trap are also plotted : circle for n42 grade neomagnets , and square for n40uh neomagnets.,scaledwidth=49.0% ] the electrostatic field near the center of the trap is approximately that of a quadrupole potential .
notwithstanding holes for admitting ions or radial beams , the electrode surfaces facing the trap center are chosen to help minimize deviations from a quadrupole potential.@xcite different geometries were studied numerically using a boundary element method ( bem)@xcite originally developed for calculating properties of electrostatic lenses . in the two - magnet trap
the reentrant endcaps are positioned close to the trap center in order to make the magnetic field more uniform , causing the electric field due to the endcaps to penetrate the trap center . as a result
, the axial trapping well has a depth ( maximum - to - minimum potential difference ) equal to 38.8 % of @xmath59 , the applied voltage . having obtained the electric and magnetic fields of the trap , the motions of an ion are computed by integrating the equations of motion using standard runge - kutta techniques .
an example of an ion trajectory is shown in fig .
[ fig : xysim ] .
similar computations were undertaken to investigate the conditions under which energetic ions extracted from an ebit ion source can be slowed and captured .
circular motions of a ne@xmath40 ion in the midplane of the two - magnet penning trap .
the initial condition is chosen to illustrate the fast cyclotron motion undergoing a counter - clockwise magnetron drift ( dashed circle ) around the center of the trap.,scaledwidth=43.0% ]
[ fig : apparatus ] shows a simplified diagram of a room - temperature apparatus with the two - magnet penning trap ( centered on the six - way cross ) , ion optics , and several detectors . the vacuum chamber ( illustrated in very light shades )
is built from readily - available commercial components , evacuated using turbo - molecular and ion pumps ( not shown ) .
ions entering the apparatus are first steered by orthogonal pairs of deflectors ( rightmost ) and then focused with an einzel lens .
as the ions approach the penning trap in the six - way cross , they are slowed down by the electric field from a pair of rings with tapered inner surfaces .
the two - magnet penning trap is oriented so that a pair of holes in the ring electrode is aligned with the center of a reentrant window at the top of the six - way cross ; a lens system installed on top of this cross collects light emitted by the stored ions onto a photomultiplier .
ions can be counted by ejecting them to a time - of - flight ( tof ) micro - channel plate ( mcp ) centered on the six - way cube ; alternatively , a position - sensitive mcp detector ( leftmost ) can be used if the tof detector is retracted with a translator .
this experimental set - up is constrained to fit within @xmath60 of vacant space in the ion extraction area of the nist ebit .
the ion source is an electron beam ion trap ( ebit ) which produces highly charged ions by electron impact ionization of an injected gas .
the nist ebit is equipped with a beamline for extracting the highly charged ions in pulses , analyzing the charge states , and transporting them to the user area.@xcite for the work reported here , neutral gas of neon or argon is injected into the ebit , which is operated with an electron beam energy as high as 4000 ev to produce various charge states of interest .
extracted ions have an energy of @xmath61 , where @xmath62 is the ion charge and @xmath63 is the electron beam energy . a simple diagram illustrating the transport , charge state selection by an analyzing magnet , and injection of extracted ions into a penning trap is provided in ref.@xcite .
we have recently demonstrated capture of highly charged ions from the ebit using both the one - magnet penning trap in fig.[fig:1magtrapcad ] and two - magnet penning trap in fig.[fig:2magtrapcad ] .
a detailed discussion of the transport , slowing and capture of ions with extraction energies of @xmath64 ev will be presented elsewhere . in brief , an ion pulse is admitted into the penning trap by momentarily lowering the potential on the entrance endcap during its transit from the ion source . at an optimal time corresponding to the arrival of the ions in the penning trap the potential of this endcap is raised to close the trap . to effectively slow ions ,
the penning trap is floated to a high voltage that matches the ebit ion beam energy . in these early tests ,
captured ions included ne xi , ne x , ne ix , as well as ar xvii , ar xvi , ar xv and ar xiv . in the discussion
below , we focus on bare neon nuclei ( i.e. , ne@xmath40 or ne xi ) . to evaluate the trap operation , the captured ions are kept in the trap for a particular `` storage time , '' and then ejected through the hole in the exit endcap towards a micro - channel plate detector(see fig .
[ fig : apparatus ] ) . upon hitting the tof detector ,
the ion pulse is converted by the micro - channel plate into an electrical pulse with a gain in the range from @xmath65 to @xmath66 .
the timing and detection scheme is shown in fig .
[ fig : detectscheme ] . the electron pulse at the output of the detector is capacitively coupled to a fast pre - amplifier which converts it into a voltage for digital acquisition .
this signal is recorded by a digital oscilloscope that is triggered to start data acquisition at the same time when the trap is triggered to release the stored ions . the trigger pulse from a gate / delay generator
is delayed from the instant of ion capture by the desired storage time .
the tof detector signals for some representative storage times are shown in figure [ fig : tof_ionlossdecaytime ] ( right side ) . for storage times much shorter than a second
, the tof detector signal has only one strong peak , corresponding to ne@xmath40 , as illustrated in fig.[fig : tof_ionlossdecaytime](a ) .
after a storage time of about half a second ( b ) , a second peak is clearly observable , with its arrival time corresponding to ne@xmath67 . with increasing storage time from ( c ) through ( e ) , more tof signal peaks are observed .
the production of lower charge states is due to electron capture from background gas atoms in charge exchange collisions .
the vacuum chamber pressure was about @xmath68 pa . the overall tof signal size decreases with storage time .
the vertical scale is magnified in going from figure [ fig : tof_ionlossdecaytime](a ) through ( e ) , by a factor of 2 each time .
this pattern suggests that the ion cloud is expanding radially , thus reducing the number of ions that can pass through the aperture ( 8 mm diameter ) of the open - access endcap when ejected.@xcite it is possible , particularly for long storage times , that some ions may have already collided with the ring electrode by the time of ejection .
the tof detector signal can be converted to ion number since the charge states and signal gain calibration are known . with the fast response time of the tof detector ( 8 mm active diameter , @xmath69 ns rise time ) the peaks corresponding to the arrival of different charge states
are well resolved , allowing the number of ions to be determined with good accuracy .
we observe exponential decay in the ion count as a function of the storage time , as illustrated in figure [ fig : tof_ionlossdecaytime](f ) .
the logarithmic - linear plot shows a good fit to an exponential function for the number of detected bare ne ions , as well as for the sum of all detected ions including charge exchange products .
the decay time constants ( e@xmath70 time ) are 1.09(2 ) s for the detected bare ne nuclei , and 2.41(6 ) s for the sum over charge states .
the decay rate for the sum of all detected ions is presumably due to the slow expansion of the ion cloud as energy is removed due to collisions with the background gas.@xcite to the extent that the detected charge states have roughly the same detection efficiency , the decay rate for the sum of charge states provides an estimate of the expansion rate . subtracting the expansion rate from the decay rate for bare ne nuclei ,
one obtains an estimate of the charge exchange rate of conversion to lower charge states ; this gives a time constant of @xmath71 s for the charge - exchange loss of ne xi ions stored in a trap with background gas pressure of @xmath72 pa .
collisions with the residual background gas is a dominant factor for the retention of the captured ions in this room - temperature trap .
the background gas cools the axial motion .
however , as discussed in sec .
( [ sec : onemag ] ) , the magnetron drift that keeps an ion on the potential hill is metastable ; background gas collisions ( drag ) tend to increase the magnetron radius until the ion hits the ring electrode.@xcite we measured the decay rates of stored ions for various background gas pressures .
figure [ fig : decaypressuredependence ] plots the measurements for two well - depths ; for each well - depth , the dependence of the injected ions ( ne@xmath40 ) only and the sum of detected ions are presented . in the observation range below @xmath73
pa , all observed decay rates are proportional to the gas pressure . for the base pressure of @xmath74 pa obtained in this room - temperature apparatus , a decay time constant of 3.8
s was observed for the sum of detected ions coming from a @xmath75 v well .
storage times @xmath76 s are useful for a variety of experiments . from the trend in fig .
[ fig : decaypressuredependence ] , we anticipate further improvements at lower pressures , e.g. in a cryogenic apparatus .
tests indicate that the sintered ndfeb magnets can be thermal - cycled between liquid nitrogen and room temperatures without damage or degradation .
in addition to lowering the residual gas pressure via cryopumping , operating at 77k would strengthen the remnant field of the ndfeb magnets by @xmath77 percent . with the trap electrodes and magnets heat - sunk to a cryogenic bath
, thermal fluctuations can be expected to be significantly smaller .
however , the best performance in the low pressure regime would require good surface uniformity and conductivity to minimize patch effects which contribute to the ion cloud expansion rate .
for the experiments discussed here , all surfaces of magnets and iron electrodes that bound the trapping volume are polished to near optical flatness . the electrical conductivity of the surfaces can be improved by deposition of a thin layer of gold or carbon ( such as aquadag or aerodag)@xcite .
ion temperature is important also .
kinetic / transport theories of an ion swarm moving in a neutral gas indicate that ion mobility@xcite improves with lower temperature.@xcite this is consistent with our observation that ion cloud radius expansion tends to be slower for colder ions .
[ fig : decaypressuredependence ] shows a significant variation in the decay rate of the total number of detected ions depending upon the trapping well depth : the decay rate in a shallower trapping well ( @xmath75 v ) is a factor of @xmath78 lower than the decay rate for @xmath79 v. to correlate this with temperature , we note that the tof peaks for the @xmath75 v well are narrower than corresponding peaks in a @xmath79 v well , by almost a factor of 2 , indicating that the ions stored in the shallower @xmath75 v well are colder .
the expected higher ion mobility could account for a slower ion cloud expansion and consequently a lower decay rate of the ion count from the @xmath75 v well .
dependence of observed ion number decay rate upon background gas pressure , for two potential well depths : @xmath80 v ( circle ) and @xmath81 v ( square ) .
filled symbols are for ne@xmath40 only .
unfilled symbols are summed over charge states .
error bars represent @xmath82 uncertainty .
, scaledwidth=42.0% ]
penning traps with unitary architecture are presented which fully integrate rare - earth magnets and electrical conductors to jointly provide the electric and magnetic fields necessary for ion confinement . in the one - magnet penning trap , the simplest case , a ni - cu - ni - plated ndfeb
magnet serves as the ring electrode . in the two - magnet penning trap , occupying as little as 125 @xmath83 of space , the rare - earth magnets are secluded and yoked by soft - iron electrodes ; the reentrant endcaps become the dominant source of the magnetic field in the trapping region . the magnetic field at the center is about 0.32 t with better than 1 percent uniformity over a distance of @xmath58 mm .
the simplicity of such compact penning traps contributes to ease of assembly , self - alignment , operation at high voltages , and beam access to the trapping region .
we have demonstrated the use of a one - magnet penning trap and a two - magnet penning trap for storing highly charged ions extracted from the nist ebit .
ions captured in these room - temperature traps include ne@xmath40 , ne@xmath67 , ne@xmath84 , as well as ar@xmath85 , ar@xmath86 , ar@xmath87 and ar@xmath88 .
experiments with captured bare ne nuclei indicate that the one - magnet and two - magnet penning traps are well aligned ; the stored ion clouds in both traps expand slowly , with comparable ion counts and loss rates dominated by collisions with background gas . in initial studies with the one - magnet penning trap
, the ion count rate for short storage time was so high as to exceed the limit of the position - sensitive mcp detector , which exhibits pile - up effects when the ion count rate exceeds @xmath89 khz . using data for ion storage times
@xmath90 second to avoid pile - up effects in the detector for the one - magnet trap , we obtained ion cloud expansion time - constants of order 1 second at base pressure @xmath91 pa .
subsequently , a much faster time - of - flight mcp detector was added in the modified set - up for the two - magnet penning trap to allow observations in the short storage - time regime , analysis of the ion charge states , and better fine - tuning of ion injection .
another improvement was the use of more vacuum pumps ( not shown in fig.[fig : apparatus ] ) to obtain lower residual background gas pressure .
a base pressure of @xmath74 pa was attained in the two - magnet penning trap ; based upon observed decay rates , we estimate that the ion cloud expansion time - constant is about 3 s under optimal conditions , sufficiently long for a variety of spectroscopic experiments .
improvement in storage lifetime is anticipated at lower pressures , e.g. in a cryogenic apparatus .
active stabilization of the ion cloud rotation to prolong ion confinement is also possible by incorporating rotating - wall " electrodes.@xcite our work is motivated by efforts at nist to produce one - electron ions in circular rydberg states.@xcite other potential applications include portable mass spectrometers ( for space - borne or earth - based field instruments ) , laser spectroscopy , and studies of rydberg or metastable states in stored highly charged ions . | penning traps are made extremely compact by embedding rare - earth permanent magnets in the electrode structure .
axially - oriented ndfeb magnets are used in unitary architectures that couple the electric and magnetic components into an integrated structure .
we have constructed a two - magnet penning trap with radial access to enable the use of laser or atomic beams , as well as the collection of light .
an experimental apparatus equipped with ion optics is installed at the nist electron beam ion trap ( ebit ) facility , constrained to fit within 1 meter at the end of a horizontal beamline for transporting highly charged ions .
highly charged ions of neon and argon , extracted with initial energies up to 4000 ev per unit charge , are captured and stored to study the confinement properties of a one - magnet trap and a two - magnet trap .
design considerations and some test results are discussed .
[ http://link.aip.org/link/?rsi/83/023103 ] copyright 2012 american institute of physics |
let @xmath9 , @xmath10 , be a bounded , connected domain with a @xmath11 smooth boundary @xmath12 .
a complex number @xmath13 will be called an interior transmission eigenvalue ( ite ) if the following problem has a non - trivial solution : @xmath14 where @xmath15 denotes the exterior euclidean unit normal to @xmath16 , @xmath17 , @xmath18 are strictly positive real - valued functions .
the spectral problem for ( ite ) is related to a non self - adjoint operator @xmath19 ( see section 3 ) and in the isotropic case @xmath20 the boundary problem ( 1.1 ) is not parameter - elliptic . for these reasons many well - known techniques developed for self - adjoint operators or for parameter - elliptic boundary problems are not applicable .
the positive ( ite ) are related to the inverse scattering problems .
more precisely , if @xmath21 is a real ( ite ) , then the far - field operator @xmath22 with kernel the scattering amplitude @xmath23 is not injective and its range is not dense .
this is crucial for the so - called linear sampling method ( see @xcite , @xcite ) which works if we avoid the real ( ite ) . for this reason
the problem of the existence and the discreteness of ( ite ) draw the attention of many authors ( see the survey @xcite for a comprehensive review and a more complete list of references ) .
secondly , it was proved that we can determine the ( ite ) from the far - field operator .
finally , it was established that in some cases the knowledge of all complex ( ite ) determines the index of refraction of the scattering obstacle ( see @xcite , @xcite ) .
this explains the increasing interest toward ( ite ) and the fact that a lot of papers concerning the existence and the spectral properties of ( ite ) in relation with the inverse scattering problems of reconstruction have been recently published .
on the other hand , the analysis of the ( ite ) leads to some interesting and difficult mathematical spectral problems for non self - adjoint operators .
these problems are connected with two * major questions * : + ( a ) describe the eigenvalue - free regions in the complex plane .
+ ( b ) find a weyl asymptotic of the counting function of the eigenvalues .
+ in contrast to the case of self - adjoint operators these questions are much more difficult and there are no general results . as far as the weyl asymptotics are concerned
, one may study the leading term of the counting function and one can search an optimal remainder . on the other hand , even in the case of boundary problems for non self - adjoint operators which are parameter - elliptic , the weyl asymptotics in the literature concern mainly the leading term ( see @xcite for some results in this direction for non self - adjoint operators ) .
the question ( a ) has been investigated by the second author in @xcite ( see also @xcite for a weaker result ) and the result in @xcite plays an important role in our analysis . in the present paper
our purpose is to study the question ( b ) . under some conditions the ( ite )
form a discrete set in @xmath24 and they have as an accumulation point only infinity ( see for instance @xcite , @xcite ) .
introduce the counting function @xmath25 where the eigenvalues are counted with their multiplicity ( see section 3 for the precise definition of the multiplicity ) .
recently , many works concerning the weyl asymptotics of @xmath26 have been published both in the isotropic @xmath27 and anisotropic cases ( see @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) . in @xcite
the case when @xmath28 and @xmath29 has been investigated and for @xmath30 a sharp asymptotics of @xmath26 with remainder @xmath31 has been established . in
all other works only the leading term of @xmath26 was obtained .
we should mention that in @xcite the anisotropic case has been studied and the asymptotics of @xmath26 with a remainder is stated .
however , the proof has a gap and only the asymptotics with leading term seems to be correct .
the isotropic case is more difficult since the boundary problem is not parameter - elliptic and the tools for elliptic boundary problems can not be applied . in the isotropic case when @xmath32 , @xmath33 , @xmath34 , it has been recently established in @xcite , @xcite the asymptotics @xmath35 where @xmath36 and @xmath37 are defined below .
it is important to remark that in @xcite , @xcite the analysis is based on the study of some trace class operators leading to an asymptotics @xmath38 where @xmath39 is sufficiently large . combining this asymptotics with the tauberian theorem of hardy and littlewood , one obtains ( 1.2 ) and
the remainder is given by the principal part divided by a logarithmic factor . to obtain a sharper remainder
one could apply a finer tauberain theorem ( see @xcite ) , but for this purpose it is necessary to establish asymptotics like ( 1.3 ) with sharper remainder for @xmath40 lying on certain parabola in @xmath41 .
this , however , seems to be a very difficult problem . in the present work
we follow another approach inspired by the paper @xcite , where asymptotics have been established for the number of the resonances associated to an exterior transmission boundary problem .
the purpose is to study the asymptotic behavior of @xmath26 under the condition @xmath42 our main result is the following assume @xmath43 fulfilled .
assume also either the condition @xmath44 or the condition @xmath45 then , the @xmath46 form a discrete set in @xmath47 and we have the following asymptotics @xmath48 for every @xmath49 , where @xmath50 @xmath51 being the volume of the unit ball in @xmath52 , and @xmath53 if @xmath54 holds , @xmath55 if @xmath56 holds . moreover ,
if in addition to @xmath56 we assume either the condition @xmath57 or the condition @xmath58 then @xmath59 holds with @xmath53 . to prove this theorem we use in an essential way the eigenvalue - free regions obtained in @xcite . in fact , we prove in the present paper a * more general result * saying that if there are no interior transmission eigenvalues in a region of the form @xmath60 then the asymptotics ( 1.7 ) with remainder @xmath61 is true . on the other hand , it is proved in @xcite that under the assumptions of theorem 1.1 , we have indeed an eigenvalue - free region ( 1.10 ) with @xmath62 replaced by @xmath63 , where @xmath62 is given by theorem 1.1 .
note that the parametrix construction of the dirichlet - to - neumann map in @xcite , section 11 , suggests that for strictly concave domains there are reasons to believe that ( 1.10 ) is true with @xmath64 .
hence in this case our argument will imply immediately a better bound of the remainder in ( 1.7 ) .
it is also worth noticing that if we have an eigenvalue - free region of the form ( 1.10 ) with @xmath65 , we get asymptotics with an almost optimal remainder term @xmath66 in ( 1.7 ) .
however , the existence of such an eigenvalue - free region is a very difficult open problem . nevertheless ,
according to our result , the problem of bounding the remainder in the weyl formula for the ( ite ) is reduced to that of getting an eigenvalue - free region in @xmath41 , and a larger eigenvalue - free region yields a sharper bound for the remainder . to our best knowledge
, it seems that our paper is the first one where such a relationship is established . for reader s convenience , in what follows in this section we will discuss the main steps in the proof of theorem 1.1 .
the starting point of our argument is a trace formula ( see section 3 and ( 3.5 ) ) which allows us to relate the number of the ( ite ) with the number of the eigenvalues , @xmath67 , of two self - adjoint operators for which the weyl asymptotics are known to hold , together with a trace of an operator given by an integral involving a meromorphic operator - valued function , @xmath68 , and its inverse @xmath69 ( see formula ( 3.6 ) ) . the main problem to deal with is to estimate the trace of this integral and it yields the bound @xmath70 of the remainder .
we apply this formula to obtain an asymptotics for the difference @xmath71 , and by a standard argument it is easy to see that this is sufficient to prove ( 1.7 ) .
since it is more convenient to work in the semi - classical setting , we reduce our problem to a semi - classical one by introducing a small parameter @xmath72 .
thus we are going to count the number of points @xmath73 , @xmath74 being an ( ite ) , in a region of the form @xmath75 provided we have an eigenvalue - free region ( 1.10 ) with @xmath63 in place of @xmath62 ( see proposition 3.7 ) .
this requires to make a change of variables @xmath76 in the trace formula ( 3.6 ) and to study the behavior of the integral term when @xmath77 and @xmath78 next we construct a meromorphic function @xmath79 with poles among the points @xmath80 and such that if an ( ite ) , @xmath81 , does not belong to the set @xmath82 , then @xmath83 is a zero of @xmath79 and the multiplicities of the corresponding zeros of @xmath79 and ( ite ) agree .
it should be mentioned that the construction of the function @xmath79 is not trivial and it requires to build a semi - classical parametrix for the corresponding dirichlet - to - neumann map @xmath84 in the elliptic zone .
this is carried out in section 2 by using the parametrix construction in @xcite .
the estimate of the remainder is reduced to that of the integral @xmath85 where @xmath86 is a suitable closed contour chosen so that on @xmath87 we have neither zeros nor poles of @xmath79 .
the main property of the function @xmath79 is the estimate @xmath88 provided the distance between @xmath89 and the set @xmath80 is greater than @xmath90 , @xmath91 being arbitrary ( see lemma 3.4 ) .
this estimate plays a crucial role in the estimate of ( 1.11 ) .
next in lemma 3.5 we show that for @xmath92 we also have @xmath93 moreover , the function @xmath94 is holomorphic in @xmath95 , @xmath96 and satisfies the bound @xmath97 in the domain @xmath98 the next step consists of choosing a closed contour @xmath99 , where @xmath100 are linear segments parallel to the real axis .
for the integrals over @xmath101 we apply ( 1.12 ) and one gets @xmath102 we take @xmath103 \cup \tilde{\gamma}_j \cup [ \tilde{w}_j^{+ } , w_j^{+}],\ : j = 1,2 $ ] with suitable contours @xmath104 ( see section 3 for the notation ) .
the estimates of the imaginary parts of the integrals over @xmath105 are more delicate since these contours cross the positive real axis and we must avoid the points @xmath106 .
our argument is similar to the choice of the contour in @xcite and the details are given in section 3 .
the main point is lemma 3.8 , where the contours @xmath104 are constructed so that @xmath107 combining this with ( 1.13 ) , we obtain the statement of proposition 3.7 and by scaling we get the asymptotics of @xmath108 + * acknowledgment . * thanks are due to the referee for the comments and remarks concerning the initial version of the paper .
let @xmath109 and consider the problem @xmath110 where @xmath111 @xmath112 , @xmath113 , @xmath114 being strictly positive functions .
the dirichlet - to - neumann map is defined by @xmath115 where @xmath116 , @xmath117 and @xmath118 denotes the restriction on @xmath16 .
denote by @xmath119 the dirichlet self - adjoint realization of the operator @xmath120 on the hilbert space @xmath121 .
it is well - known that the spectrum of @xmath119 consists of a discrete set of positive eigenvalues which are also poles of the resolvent @xmath122 .
moreover , if @xmath123 , we have @xmath124 modulo an operator - valued function holomorphic at @xmath125 , where @xmath126 is a finite rank projection .
the multiplicity of @xmath125 is defined as being the rank of @xmath126 .
let @xmath127 .
the following properties of the dirichlet - to - neumann map are more or less well - known but we will give a proof for the sake of completeness .
the dirichlet - to - neumann map @xmath128 is a meromorphic operator - valued function in @xmath95 with poles at @xmath129 , @xmath130 .
moreover , @xmath131 modulo an operator - valued function holomorphic at @xmath129 , where @xmath132 is of rank @xmath133 .
if @xmath134 , then we have the bound @xmath135 where @xmath8 is a constant which may depend on @xmath136 .
_ clearly , there exists an extension operator @xmath137 such that @xmath138 and @xmath139 is supported near @xmath16 .
if @xmath140 and @xmath141 does not belong to @xmath142 , it is easy to see that the solution @xmath143 of ( 2.1 ) can be expressed by the formula @xmath144 hence @xmath145 it follows from ( 2.4 ) that @xmath128 is a meromorphic operator - valued function in @xmath95 with poles among the poles of @xmath146 and that ( 2.2 ) holds with @xmath147 this implies @xmath148 as desired . by ( 2.4 ) we also have @xmath149 @xmath150 @xmath151 clearly , we have @xmath152 @xmath153 on the other hand , the coercive estimate @xmath154 implies the bounds @xmath155
@xmath156 @xmath157 @xmath158 therefore , ( 2.3 ) follows from the above estimates and the proof is complete .
@xmath159 let @xmath160 be coordinates on @xmath161 and denote by @xmath162 the principal symbol of the laplace - beltrami operator , @xmath163 , on @xmath16 equipped with the riemannian metric induced by the euclidean one in @xmath52 .
it is well - known that @xmath164 is a polynomial function in @xmath165 , homogeneous of order 2 , and @xmath166 with constants @xmath167 .
set @xmath168 .
let @xmath169 , @xmath170 for @xmath171 , @xmath172 for @xmath173 , and set @xmath174 where @xmath175 . for @xmath176 , introduce the function @xmath177 since @xmath178 the functions @xmath179 and @xmath180 are holomorphic in @xmath95 and @xmath181 with some constant @xmath8 . in what follows in this section we will construct a parametrix for the operator @xmath182 , where @xmath183 denotes the @xmath184do with symbol @xmath185 .
in fact , this construction is carried out in @xcite and here we will only recall the main points . first , notice that it suffices to make the construction locally and then to glue up all pieces by using a partition of the unity on @xmath16 .
given an arbitrary point @xmath186 , there exists a small neighborhood @xmath187 of @xmath188 and local normal coordinates @xmath189 such that @xmath190 , @xmath191 is defined by @xmath192 , @xmath193 being coordinates in @xmath191 , @xmath194 in @xmath195 , and in these coordinates the operator @xmath196 can be written in the form @xmath197 here we have set @xmath198 , @xmath199 , @xmath200 , @xmath201 being a symmetric @xmath202 matrix - valued function with smooth real - valued entries , @xmath203 , @xmath204 and @xmath205 being smooth functions .
moreover , we have @xmath206 , @xmath207 being the principal symbol of @xmath163 written in the coordinates @xmath160 .
let @xmath208 , @xmath209 in a neighborhood of @xmath188 . in @xcite , it was constructed a parametrix , @xmath210 , of ( 2.1 ) satisfying the condition @xmath211 and having the form @xmath212 where @xmath213 is as above and @xmath214 is a small constant independent of @xmath215 .
the phase @xmath216 is a complex - valued function such that @xmath217 and the amplitude @xmath218 satisfies @xmath219 . more generally , the functions @xmath216 and @xmath218 are of the form @xmath220 @xmath221 @xmath222 being an arbitrary integer . the phase @xmath216 satisfies the eikonal equation mod @xmath223 : @xmath224 and @xmath218 satisfies the equation @xmath225 where @xmath226 , @xmath227 and @xmath228 are smooth functions .
it was shown in section 4 of @xcite that @xmath229 , @xmath230 , @xmath231 , @xmath232 , @xmath233 , @xmath234 , uniformly in @xmath95 and @xmath235 .
recall that @xmath236 are the spaces of all functions @xmath237 satisfying the estimates @xmath238 for all multi - indices @xmath239 and @xmath240 .
moreover , the functions @xmath241 , @xmath227 , @xmath228 are polynomials in @xmath179 , @xmath180 and @xmath89 , and therefore they are holomorphic in @xmath95 . as in @xcite
, it is easy to see that @xmath242 where the function @xmath243e^{-\frac{i}{h } \langle x',\xi'\rangle}e^{\frac{i}{h}\widetilde\varphi}a + e^{\frac{i}{h}\widetilde\varphi}\phi\left(\frac{x_1}{\delta_1}\right)\left(x_1^na_n+h^nb_n\right)\ ] ] is holomorphic in @xmath89 and satisfies the bounds @xmath244 with some @xmath245 independent of @xmath246 and @xmath239 .
the parametrix , @xmath247 , of the operator @xmath248 is defined by @xmath249 where @xmath250 @xmath251 since @xmath252
we deduce that mod @xmath253 the function @xmath254 is given by the expression @xmath255 @xmath256 with some function @xmath257 independent of the functions @xmath258 and @xmath259 .
let @xmath260 be a partition of the unity on @xmath16
. set @xmath261 the operator @xmath262 is an @xmath184do on @xmath16 with a principal symbol @xmath263 , holomorphic in @xmath95 .
let @xmath264 be the solution of ( 2.1 ) with @xmath265 .
then @xmath266 is the solution of ( 2.1 ) with @xmath267 .
moreover , it is easy to see that , if @xmath141 does not belong to spec@xmath268 , we have @xmath269 which yields the identity @xmath270
it follows from ( 2.7 ) that if @xmath246 is taken large enough , the operator @xmath271 is meromorphic with values in the space of trace class operators on @xmath272 . let @xmath273 be the characteristic values of @xmath274 . recall that @xmath273 are defined as being the eigenvalues of the self - adjoint operator @xmath275 .
if @xmath141 does not belong to @xmath142 , then for every integer @xmath276 we have the bound @xmath277 where the constant @xmath8 depends on @xmath136 and @xmath246 but is independent of @xmath89 , @xmath278 , @xmath279 , and @xmath280 is defined in lemma @xmath281 _ proof .
_ we will use the well - known fact that the characteristic values of the laplace - beltrami operator on a compact riemannian manifold without boundary ( in our case @xmath16 , dim@xmath282 ) satisfy @xmath283 for every integer @xmath116 . on the other hand , by using the trace theorem and lemma 2.1 , we obtain @xmath284 @xmath285 @xmath286 since the function @xmath287 is compactly supported , we have the bound @xmath288 in view of ( 2.7 ) we also have @xmath289 with some @xmath290 independent of @xmath136 and @xmath246 , provided @xmath276 and @xmath246 being large enough . by ( 2.12 ) and ( 2.13 ) we conclude @xmath291 clearly , ( 2.10 ) follows from ( 2.11 ) and ( 2.14 ) and the proof is complete . @xmath159
for @xmath292 define the operator @xmath293 , where @xmath294 and @xmath295 solve the problem @xmath296 denote by @xmath297 , @xmath18 , the dirichlet self - adjoint realization of the operator @xmath298 on the hilbert space @xmath299 .
set @xmath300 and define also the operators @xmath301 , where @xmath143 is the solution of the problem @xmath302 differentiating this equation with respect to @xmath303 , one obtains easily the identity @xmath304 introduce the operator @xmath305 if @xmath306 is a meromorphic operator - valued function with residue of finite rank , the same is true for @xmath307 and we have the formula @xmath308 where @xmath309 @xmath310 @xmath311 @xmath312 moreover , if @xmath313 is a simple closed positively oriented curve which avoids the eigenvalues of @xmath297 , @xmath18 , as well as the poles of @xmath306 , then we have the identity @xmath314
@xmath315 _ proof .
_ clearly , if @xmath316 satisfies ( 3.1 ) and @xmath303 does not belong to @xmath317 , we have @xmath318 where @xmath319 . the boundary condition in ( 3.1 )
implies the identity @xmath320 hence @xmath321 which clearly implies ( 3.4 )
. moreover , if @xmath306 is meromorphic , so are the operators @xmath322 , and by ( 3.4 ) the operator @xmath307 is meromorphic , too . using ( 3.3 ) and the cyclicity of the trace ( see lemma 2.2 of @xcite ) ,
we get @xmath323 @xmath324 @xmath325 @xmath326 @xmath327 @xmath328 @xmath329 which implies ( 3.5 ) .
if @xmath307 is a meromorphic operator - valued function with residue of finite rank , we define the multiplicity of a pole @xmath330 of @xmath307 by @xmath331 let the curve @xmath332 be as in proposition 3.1 and denote by @xmath333 and @xmath334 , @xmath18 , the number ( counted with the multiplicity ) of the poles of @xmath307 and the eigenvalues of @xmath297 , respectively , in the interior of @xmath332 .
proposition 3.1 implies the following _ proof .
_ it is easy to see that @xmath336 , where the operator @xmath19 is defined by @xmath337 with domain @xmath338 @xmath339 hence the finite - rank operator @xmath340 is in fact a projection ( e.g. see @xcite ) , and therefore the rank coincides with the trace .
thus , ( 3.6 ) follows from ( 3.5 ) .
let @xmath89 and @xmath278 be as in the previous section and denote by @xmath341 , @xmath342 , @xmath343 , @xmath18 , the operators defined by replacing in the definition of @xmath344 , @xmath345 , @xmath274 introduced in section 2 the pair @xmath346 by @xmath347 .
clearly , we have the relationship @xmath348 @xmath349 in what follows @xmath350 will denote the sobolev space @xmath351 equipped with the semi - classical norm .
there exist an invertible , bounded operator @xmath352 , with an inverse @xmath353 , @xmath354 , and trace class operators @xmath355 and @xmath356 such that @xmath357 @xmath358 where @xmath359 if @xmath54 holds , @xmath360 if @xmath56 holds .
moreover , the operators @xmath361 are holomorphic with respect to @xmath89 for @xmath362 _ proof . _ set @xmath363 , @xmath364 and let the real - valued function @xmath365 be as in section 2 , with a sufficiently large support .
it follows from the parametrix construction in section 2 that @xmath366 with a symbol @xmath367 , where @xmath368 are holomorphic in @xmath95 , and @xmath369 let @xmath370 be a real - valued compactly supported function such that @xmath371 and @xmath372 on supp@xmath287 .
it suffices to show that the operator @xmath373 is invertible .
indeed , this would imply ( 3.8 ) and ( 3.9 ) with @xmath374 and @xmath375 , @xmath376 .
an easy computation shows that @xmath377 where @xmath378 and @xmath379 are the restrictions on @xmath16 of the functions @xmath380 respectively .
let us see that @xmath381 with some constants @xmath382 , where @xmath359 if @xmath383 and @xmath360 if @xmath384 , @xmath385 . since @xmath386 we have with some positive constants @xmath387 , @xmath388 , @xmath389 ,
@xmath390 @xmath391 @xmath392 which yields the lower bound in ( 3.10 ) .
the upper bound is obvious .
it follows from ( 3.10 ) that @xmath393 .
moreover , when @xmath394 , @xmath395 on @xmath16 , by ( 2.8 ) one concludes that @xmath396 .
hence the operator @xmath373 is invertible with an inverse which is an @xmath184do with a symbol belonging to the class @xmath397 . in particular , we have @xmath398 , @xmath399 , @xmath354 .
set @xmath400 , @xmath18 .
define the operator @xmath401 as follows : @xmath402 @xmath403 we obtain easily that @xmath404 @xmath405 clearly , the operator @xmath406 is trace class and meromorphic in @xmath95 with poles @xmath407 , @xmath408 , and residue of finite rank , so we can define the meromorphic function @xmath409 _ proof .
_ it follows from lemma 2.2 and the properties of the characteristic values that @xmath412 satisfy the bound ( 2.10 ) with a new constant @xmath8 and @xmath413 replaced by @xmath414 .
in fact , for @xmath415 we have @xmath416 since the operator @xmath417 is bounded , for the first term on the right hand side we apply lemma 2.2 . on the other hand , @xmath418 and for @xmath419 we obtain easily ( 2.10 ) with @xmath420 since @xmath421 has compact support .
next , if @xmath422 , then we have @xmath423 or @xmath424 the case @xmath425 is similar .
+ therefore , we have @xmath426 @xmath427 @xmath428 @xmath429 now , given any @xmath430 we can take @xmath431 and @xmath432 , and ( 3.11 ) follows from ( 3.12 ) .
let @xmath62 be as in theorem @xmath433 .
then , given any @xmath434 the operator @xmath435 is invertible on @xmath272 for @xmath95 , @xmath436 and its inverse satisfies in this region the bound @xmath437 with some constants @xmath438 . for these values of @xmath89
we also have @xmath439 moreover , the function @xmath94 is holomorphic in @xmath95 , @xmath96 and satisfies the bound @xmath440 in @xmath441 .
_ it follows from the analysis in section 5 of @xcite that , under the assumptions of theorem 1.1 , the operator @xmath442 is invertible for @xmath95 , @xmath96 and @xmath443 @xmath444 now ( 3.13 ) follows from these bounds and lemma 3.3 because @xmath445 @xmath446 the bound ( 3.14 ) can be obtained in precisely the same way as ( 3.11 ) by using ( 3.13 ) and the formula @xmath447 note that the norm @xmath448 will add a factor @xmath449 which for sufficiently large @xmath136 yields a factor @xmath450 .
clearly , it follows from the fredholm theorem that , under the assumptions of theorem 1.1 , the operator - valued function @xmath451 is meromorphic in @xmath452 with finite rank residue and holomorphic with respect to @xmath95 for @xmath96
. therefore the functions @xmath79 and @xmath453 are holomorphic in @xmath95 , @xmath96 , and hence so is @xmath94 .
fix an arbitrary @xmath454 .
then the function @xmath455 is holomorphic in @xmath95 , @xmath96 and @xmath456 .
it follows from the bounds ( 3.11 ) and ( 3.14 ) that @xmath457 for @xmath95 , @xmath96 in particular , the later estimate holds on the circle @xmath458 since for every @xmath459 we have @xmath460 applying the caratheodory theorem ( e.g. see 5.5 in @xcite ) , we get @xmath461 this implies ( 3.15 ) because @xmath462 .
let @xmath86 be a simple closed positively oriented curve which avoids the eigenvalues of @xmath463 , @xmath18 , as well as the poles of @xmath464 . denote by @xmath465 the number of the poles , @xmath466 , of @xmath307 such that @xmath83 are in the interior of the domain @xmath467 with boundary @xmath332 . similarly , we denote by @xmath468 the number of the eigenvalues , @xmath469 , of @xmath297 such that @xmath129 are in @xmath467 .
corollary 3.2 implies the following _ proof .
_ we apply ( 3.6 ) and use the identities @xmath471 combined with the analyticity of @xmath472 in @xmath89 and the following well - known formula @xmath473 the above formula for @xmath474 is classical for finite rank perturbations of the identity . for trace class ones this formula follows by an approximation with finite rank operators
( see for example , section 5 , @xcite ) .
it follows from ( 3.16 ) that @xmath475 is a zero of @xmath79 if and only if @xmath476 is a pole of @xmath477 ( and hence @xmath478 is an interior transmission eigenvalue ) and the multiplicities coincide .
similarly , one can see that if @xmath479 is a pole of @xmath79 with multiplicity @xmath480 , then @xmath481 and @xmath482 , where @xmath483 is the multiplicity of @xmath484 as an eigenvalue of @xmath297 . in what follows
we will use the formula ( 3.16 ) to prove the following _ proof .
_ we will consider only the case @xmath488 , since the case @xmath489 is similar ( and even simpler since the function @xmath79 does not have poles in @xmath490 ) .
consider the points @xmath491 , @xmath492 , @xmath493 , @xmath494 and set @xmath495 @xmath496 the following lemma will be proved later on .
there exist positively oriented piecewise smooth curves @xmath497 and @xmath498 , where @xmath499 connects the point @xmath500 with @xmath501 , while @xmath502 connects the point @xmath503 with @xmath504 , such that @xmath505 now we apply lemma 3.6 with a contour @xmath506 , where @xmath507 is the segment @xmath508 $ ] on the line passing through the points @xmath509 and @xmath510 , and @xmath511 is the segment @xmath512 $ ] on the line passing through the points @xmath513 and @xmath514 .
@xmath515\cup\widetilde\gamma_1\cup[\widetilde w_1^+,w_1^+]$ ] , @xmath516\cup\widetilde\gamma_2\cup[\widetilde w_2 ^ -,w_2 ^ -]$ ] ( see figure 1 ) .
since @xmath517 , @xmath518 , @xmath519 , by ( 3.15 ) we have @xmath520 @xmath521 applying ( 3.15 ) once more , we have @xmath522 } \frac{d}{dz}\log g_h(z)dz\right|\le c_\epsilon h^{-d+1 - 2\epsilon}\int_{3h^{\kappa-\epsilon}}^{1/2}\frac{d\sigma}{\sigma}\le c_\epsilon h^{-d+1 - 3\epsilon},\:\quad j = 1,2.\eqno{(3.20)}\ ] ] on the other hand , since the counting function of the eigenvalues of @xmath297 satisfies the weyl law , we deduce @xmath523 @xmath524 @xmath525 and similarly @xmath526 @xmath527 consequently , @xmath528 taking together ( 3.16 ) , ( 3.18 ) , ( 3.19 ) , ( 3.20 ) and ( 3.21 ) , we obtain @xmath529 thus , to establish ( 3.17 ) , it remains to show that the counting function @xmath530 satisfies @xmath531 given a parameter @xmath532 , independent of @xmath278 , introduce @xmath533 .
clearly , there exists @xmath534 such that @xmath535 , @xmath536 .
let @xmath537 be the zeros ( repeated with their multiplicities ) of @xmath79 in @xmath538 and let @xmath539 be the poles ( repeated with their multiplicities ) of @xmath79 in @xmath540 .
therefore the function @xmath541 is holomorphic in the interior of @xmath540 .
obviously , @xmath539 are among the eigenvalues of the operators @xmath542 and @xmath543 in an interval of the form @xmath544\cup [ 2-{\cal o}(h^{\kappa-\epsilon } ) , 2+{\cal o}(h^{\kappa-\epsilon})].\ ] ] hence , by the weyl law for the counting function of these eigenvalues , as in the proof of ( 3.21 ) , we get @xmath545 by ( 3.14 ) and ( 3.24 ) , we have @xmath546 @xmath547 on the other hand , applying ( 3.11 ) and ( 3.24 ) , for @xmath548 , @xmath549 , @xmath550 , we obtain @xmath551 @xmath552 we claim that there exists @xmath553 such that the distance between @xmath539 and the circle @xmath554 is greater than @xmath90 , provided @xmath555 . indeed , if we suppose the contrary , this would imply that the length of the interval @xmath556 is upper bounded by @xmath557 , which is impossible if @xmath558 is taken large enough .
this proves the claim .
thus , by ( 3.26 ) we have the estimate @xmath559 on @xmath554 , which in turn implies @xmath559 on @xmath560 . combining this with ( 3.25 ) and the jensen theorem
( see for example 3.6 in @xcite ) , yields for the zeros @xmath561 in @xmath562 the following bound @xmath563 since the left - hand side of ( 3.23 ) is upper bounded by the number of the zeros and the poles of the function @xmath79 in @xmath564 , the estimate ( 3.23 ) follows from ( 3.24 ) and ( 3.27 ) .
the bound @xmath565 of the number of the zeros @xmath561 of @xmath79 in @xmath566 does not depend on the statement of lemma @xmath567 but only on the application of the jensen theorem based on @xmath568 , @xmath569 .
we will use @xmath565 in the proof of lemma @xmath567 below .
_ proof of lemma @xmath567 .
_ we will consider only the case @xmath570 , since the case @xmath571 is similar .
introduce the function @xmath572 where @xmath573 is the set of all zeros of @xmath79 in @xmath574 and @xmath575 is the set of all poles of @xmath79 in @xmath576 .
since @xmath577 does not have zeros and poles in @xmath574 , the function @xmath578 is holomorphic in @xmath574 .
we need the following _ proof . _
set @xmath581 , where @xmath555 and @xmath582 .
clearly , @xmath583 , where every @xmath584 is a domain with a piecewise smooth boundary and @xmath585 if @xmath586 .
moreover , we have @xmath587 let @xmath588 be independent of @xmath278 .
let @xmath589 be the set of all @xmath584 such that @xmath590 we now construct a closed curve , @xmath591 as follows : we keep all arcs on @xmath592 having no common points with @xmath589 and replace the arc @xmath593 with arcs on @xmath594 connecting the corresponding end points .
thus we can guarantee that @xmath595 belongs to an @xmath596 neighborhood of @xmath597 and , moreover , the distance between @xmath595 and the set @xmath598 is greater than @xmath90 . in the same way , as in the proof of ( 3.25 ) and ( 3.26 ) above , by using ( 3.11 ) , ( 3.14 ) , ( 3.24 ) and ( 3.27 ) , we get @xmath599 since @xmath600 is in the interior of the domain bounded by @xmath595 , the estimate ( 3.29 ) implies ( 3.28 )
. we will now construct the curve @xmath499 .
let @xmath601 be the set of all @xmath584 such that @xmath602 \neq \emptyset.$ ] we keep all segments on @xmath603 $ ] having no common points with @xmath601 and replace the segments on @xmath603\cap u_{\nu_i}$ ] with arcs on @xmath604 connecting the corresponding end points .
thus we get a piecewise smooth curve @xmath499 belonging to an @xmath596 neighborhood of @xmath603 $ ] and the distance between @xmath499 and the set @xmath598 is greater than @xmath605 .
hence @xmath497 .
now we can write @xmath606}\frac{d}{dz}\log\zeta_h(z)dz\ ] ] @xmath607 we will show that @xmath608 @xmath609 since the length of the interval @xmath610 $ ] is @xmath611 , the estimate ( 3.31 ) implies that the absolute value of the first integral on the right - hand side of ( 3.30 ) is @xmath612 .
thus , ( 3.18 ) would follow from ( 3.30 ) , ( 3.31 ) , ( 3.32 ) and the bounds ( 3.24 ) and ( 3.27 ) . to prove ( 3.31 )
, we apply the caratheodory theorem ( see 5.5 .
@xcite ) for the derivative of the function @xmath613 .
note that @xmath614 can be bounded from below in the same way as in ( 3.25 ) above .
therefore , applying ( 3.28 ) , we get for the real part of @xmath615 the estimate @xmath616 since @xmath617 , we conclude by the caratheodory theorem that @xmath618 in the disc @xmath619 , which clearly implies ( 3.31 ) . to establish ( 3.32 ) , observe that if @xmath620 does not lie on the line connecting the points @xmath500 and @xmath621 and if @xmath622 denotes the distance from @xmath620 to this line , after a suitable change of variables , we have @xmath623 since the integral in the left - hand side of ( 3.32 ) differs from the integral in the left - hand side of ( 3.33 ) either by @xmath624 or @xmath625 , the estimate ( 3.33 ) implies ( 3.32 ) in this case .
if @xmath620 lies on the line connecting the points @xmath500 and @xmath621 , then the integral on the left - hand side of ( 3.32 ) is a limit of integrals of the first kind , and hence ( 3.32 ) will be true in this case , too .
this completes the proof of lemma 3.8 .
_ proof of theorem @xmath433_. let @xmath62 be as described in theorem 1.1 .
let @xmath626 and @xmath627 be arbitrary real numbers , independent of @xmath278 , and let @xmath628 be independent of @xmath278 .
it follows from the proof of proposition 3.7 ( see ( 3.27 ) ) that @xmath629 @xmath630 therefore , by ( 3.17 ) we get for every @xmath631 @xmath632 choose @xmath633 the above asymptotics yields @xmath634 @xmath635 recall that according to the results in @xcite , there are no ( ite ) in the region @xmath636 for every @xmath485 , provided @xmath637 . on the other hand ,
it is clear that the region @xmath638 is contained in the region @xmath639 and contains the region @xmath640 thus we get the asymptotics @xmath641 for every @xmath49 . here
we replace @xmath642 by @xmath643 , which is not important since our argument works for every @xmath644 the asymptotics ( 3.34 ) yields @xmath645 for every integer @xmath234 such that @xmath646 .
let @xmath647 be the smallest integer such that @xmath648 .
it is clear that we have @xmath649 with a constant @xmath650 independent of @xmath651 moreover , @xmath652 with @xmath653 independent of @xmath654 .
summing up the asymptotics ( 3.35 ) and using ( 3.36 ) , we get ( 1.7 ) .
thus the proof of theorem 1.1 is complete . | we prove weyl asymptotics @xmath0 , @xmath1 , for the counting function @xmath2 , @xmath3 , of the interior transmission eigenvalues ( ite ) , @xmath4 . here
@xmath5 denotes the space dimension and @xmath6 is such that there are no ( ite ) in the region @xmath7 for some @xmath8 .
interior transmission eigenvalues , weyl formula with remainder , eigenvalue - free regions + * msc : * primary 35p20 , secondary 35p15 , 35p25 |
the experimental realization of magnetically trapped bose - einstein condensates ( bec ) of alkali - metal atoms @xcite has led to a broad interest in the properties of this state of quantum matter .
one aspect of particular interest is to study the superfluid behaviour of these systems .
an ideal tool to examine this is to rotate the condensate and to observe whether such a rotation can result in the creation of vortex states .
theoretically many aspects of vortex states in a trapped bec have been discussed in the literature .
the critical angular velocity required to create a vortex state in trapped becs was derived by baym and pethick @xcite .
dalfovo and stringari determined numerically the shape of the vortex state @xcite and the moment of inertia in harmonic traps@xcite . ho and shenoy @xcite discussed the influence of the internal atomic structure on vortex states and barenghi @xcite studied vortex waves .
@xcite considered the fragmentation of a rotating bec with attractive interaction .
et al._@xcite have numerically studied excitations around a vortex state .
rokshar showed that the stability of a vortex state depends on the shape of the trap @xcite and considered the properties of vortex states on a torus @xcite . despite recent interest in the rotational properties of becs ,
the experimental realization of a vortex state remains to be done @xcite .
in contrast to superfluid helium , the trapped becs are not in direct contact to an external container .
how to rotate the trapped becs into vortex states is still an open question .
we have recently proposed to transfer angular momentum to a bec by using multiple travelling wave laser beams @xcite and to create vortex states by means of laguerre - gaussian laser beams using a raman transition @xcite . a related idea based on raman transitions
has also independently been proposed by bolda and walls @xcite . in addition , a combination of multiple laser beams with raman transition has recently been employed to discuss the creation of dark solitons and vortices in trapped becs @xcite . in the present paper
we study different aspects of the quantized circular motion of a trapped bec driven by a spatially homogeneous rotating force . in ref .
@xcite we showed that such a force can be created by four travelling wave laser beams .
in particular , we analyse the reason why in a harmonic trap the rotating force only leads to a rotational state of becs which is a coherent superposition of vortex states . on this basis
, we further demonstrate how to create a pure vortex state by introducing an anharmonic trapping potential .
the paper is organised as follows . in sec .
ii we derive the gross - pitaevskii equation for trapped becs composed of two - level atoms in the presence of highly detuned laser fields with different frequencies . in sec .
iii , we review the proposal of ref . @xcite and apply the general model to the transfer of angular momentum to becs by a configuration of four travelling - wave laser beams . the force induced by these laser beams is spatially homogeneous and rotates with the frequency difference between the laser beams . in sec .
iv we discuss the center - of - mass motion and the shape of a harmonically trapped bec driven by the rotating force .
further we analyze the physical mechanism of why rotating a harmonically trapped bec can not create a pure vortex state . the possibility to create a pure vortex state of an ideal bec
is demonstrated in sec .
v by introducing an anharmonic trapping potential . in sec .
vi we consider an interacting bose gas in an anharmonic trap . with a sinusodial modulation of the frequency difference between the laser beams the nonlinear energy shift due to
the interatomic interaction can be compensated .
we numerically show that an almost pure vortex state can be created .
the conlusion is included in sec .
vii with a short discussion of the relation between the present proposal and that of refs .
@xcite based on raman transitions .
we consider a condensate of @xmath1 two - level atoms at zero temperature confined in a trapping potential . in the formalism of quantum field theory
this system can be described by the hamiltonian @xmath2 \psi_i ( \vec { x } ) \big \ } + { 1\over2 } \int d^3 x \sum_{i , j = e , g } g_{ij}\psi_i^\dagger ( \vec { x})\psi_j^\dagger ( \vec{x } ) \psi_j(\vec{x } ) \psi_i ( \vec { x } ) \ ; . \label{h0}\ ] ] the symbol @xmath3 denotes the chemical potential .
the center - of - mass hamiltonian @xmath4 contains the kinetic energy and the trapping potential @xmath5 , which is assumed to be cylindrically symmetric around the z - axis .
@xmath6 denote the field operator for atoms in the internal ground state ( @xmath7 ) and excited state ( @xmath8 ) with internal energy @xmath9 .
@xmath10 dentoes the atomic mass . without affecting the generality of our final results we have assumed that the trap potential is the same for both internal states .
the second integral in eq .
( [ h0 ] ) describes the interatomic interaction in the absence of laser fields .
the factors @xmath11 are defined as @xmath12 , with @xmath13 being the s - wave scattering lengths for atoms with different internal states . the interaction of the atoms with the laser beams is incorporated by the electric dipole coupling in the rotating wave approximation , @xmath14 here @xmath15 is the frequency of the @xmath16th laser beam , and @xmath17 is defined as @xmath18 with @xmath19 being the dipole moment of the atoms and @xmath20 denoting the positive frequency part of the @xmath16th laser s electric field .
it has been discussed at length in the literature @xcite how this hamiltonian can be reduced to an effective hamiltonian for atoms being in their internal ground state in the presence of a single laser beam . for this reason
we will only briefly discuss how this can be done in the presence of several laser beams with different frequencies . in order to avoid spontaneous emission
all laser beams have to be detuned far off resonance so that the population of the excited - state atoms is negligible . in this case
the heisenberg equation of motion for the excited - state field operator can be adiabatically eliminated by assuming that the detunings @xmath21 of the laser beams from the atomic resonance frequency are all in the same order of magnitude and are the largest characteristic frequencies of the system @xcite . the resulting expression for @xmath22
is given by @xmath23 the kernel @xmath24 describing the photon - exchange induced dipole - dipole interaction is defined in ref .
the parameter @xmath25 denotes the sponataneous emission rate from excited atoms and is assumed to be much smaller than the detuning of the laser beams in this paper .
the expression ( [ psie ] ) for @xmath22 can be inserted into the heisenberg equation for @xmath26 thus leading to the effective equation @xmath27 + v_l(\vec{x},t ) \psi_g(\vec{x},t ) + v_{nl } ( \vec{x},t ) \psi_g ( \vec{x},t ) \ ; , \label{qftdgl}\ ] ] with the light - induced linear potential being given by @xmath28 note that in the denominator we have replaced @xmath29 by the average detuning @xmath30 of all lasers .
this has to be done for consistency since otherwise @xmath31 would be non - hermitian even in absence of spontaneous emission .
mathematically this can be made because the difference @xmath32 between two detunings has to be treated as of higher order in @xmath33 .
assuming that the condensate has a coherence length larger than the characteristic distance of the photon - exchange induced dipole - dipole interation , we derive the nonlinear potential @xmath34 the strength of this potential depends on the intensity of the applied laser fields .
as we will be concerned only with very weak laser beams @xmath35 can safely be neglected in the present case . to describe the collective evolution of a condensed gas of bosonic atoms we will adopt the mean field approach in eq .
( [ qftdgl ] ) .
this approximation describes a trapped bose - einstein condensate quite well @xcite and amounts in replacing the field operator @xmath36 in eq .
( [ qftdgl ] ) , after the calculation of the commutator , by a collective wavefunction @xmath37 of all atoms in the condensate , with @xmath38 being normalized to one .
this procedure results in the gross - pitaevskii equation for @xmath38 , @xmath39 for brevity we have introduced @xmath40 .
this equation determines the nonlinear dynamics of trapped becs in multiple laser beams and defines the effective hamiltonian @xmath41 for the collective wavefunction @xmath38 .
the design of an optical potential @xmath31 which can rotate the bec by transferring angular momentum was addressed in ref .
@xcite to which we refer for all details of the construction . here
we will only briefly sketch the important features of this optical potential .
the potential @xmath42 of ref .
@xcite is created by four traveling wave laser beams of equal intensity and width much larger than the size of the condensate ( this is usually the case ) so that they can be considered as plane waves .
the laser beams 1 and 2 propagate along the x- and y - direction have the same frequency @xmath43 .
their phases are chosen to be equal at the origin so that the rabi frequencies of the two lasers have the form @xmath44 $ ] and @xmath45 $ ] .
the other two laser beams have the frequency @xmath46 and propagate in directions slightly different from the x- and y- direction with a small angle @xmath47 . the frequency difference @xmath48 is assumed to be so small that the difference between @xmath49 and @xmath50 for any @xmath51 is negligible .
the wavevector of the third laser beam is given by @xmath52 , where @xmath53 denotes the deviation of the wavevectors .
similiarly the 4th laser s wavevector is given by @xmath54 .
the phases are chosen so that laser 4 is in phase with lasers 1 and 2 at the origin whereas the phase of laser 3 is shifted by @xmath55 .
the third rabi frequency is then given by @xmath56 \omega_0 \exp [ i ( k x -\delta k y)]$ ] and the fourth by @xmath57 $ ] .
the described configuration of laser beams leads to the optical potential @xmath58 \nonumber \\ & & + e^{-i \phi(t ) } \big [ e^{i \delta k x } + i e^{i\delta k y } + i e^{-i k(x - y ) + i \delta k y } + e^{ik(x - y)+i\delta kx } \big ] \big \ } + h.c.\ ; , \label{vl2 } \end{aligned}\ ] ] where we have introduced the effective rabi frequency @xmath59 and the time dependent phase @xmath60 . in this section and in the rest of the paper we neglect c - number terms in the optical potential since they only produce an overall phase shift in the wavefunction . since the wavelength of an optical laser is much smaller than the size of a typical condensate ( several micrometers ) all exponentials in eq . ( [ vl2 ] ) containing the wavevector @xmath61
are rapidly varying over the condensate and therefore average to zero @xcite .
the remaining exponentials , @xmath62 $ ] and @xmath63 $ ] , can be approximated by expressions linear in @xmath64 if the angle @xmath47 is small enough .
the resulting optical potential then takes the form .
@xmath65 it obviously produces a rotating spatially homogeneous force over the size of the condensate .
the rotating force can transfer angular momentum to the condensate .
this can be seen easily by rewriting it in the form @xmath66 because of the commutation relation @xmath67 = \pm \hbar ( x\pm iy)$ ] the operator @xmath68 increases the orbital angular momentum by @xmath69 .
thus , if the frequency difference @xmath70 is nearly resonant with a transition between two trap eigenstates , the angular momentum of the condensate is increased by the force @xcite . since @xmath31 is produced by laser beams we have to estimate the rate of decoherence of the condensate due to the spontaneous emission of photons which is given by the probability that an atom is excited times the spontaneous emission rate @xmath25 .
the excitation probability for off - resonant light is of the order of @xmath71 . the decoherence time is therefore given by @xmath72 and is therefore much larger than @xmath73 since the detuning is much larger than the natural linewidth .
for instance , if @xmath74 is about 10 ms then for @xmath75 the decoherence time is in the order of a second .
one also has to address the question of how stable the frequencies of the laser beams must be so that @xmath70 is well defined .
it turns out that this poses no problem at all since only the frequency _ difference _
@xmath70 is required to be stable .
since in the proposal of ref .
@xcite all four laser beams originate from the same laser source , and since the frequency difference is created by modulating the laser beams , it is clear that only the modulator frequency must be stable .
the stability requirement on the laser beams can therefore easily be fulfilled by any common laser source .
we also point out that laser beams are of course only one possibility to produce a rotating homogeneous force . the same potential could be created by spatially varying and rotating magnetic fields , for instance .
the following analysis of the condensate s motion in the potential ( [ vlvort ] ) therefore covers this case as well .
we now turn to the effect of the optical potential ( [ vlvort ] ) on a condensate trapped in a harmonical potential of the form @xmath76 @xmath77 and @xmath78 denote the trap frequency in the z - direction and in the x - y - plane , respectively .
we have shown in ref .
@xcite that in the case of an ideal bose gas and a resonant optical potential ( i.e. , @xmath79 ) the condensate rotates around the center of the trap with increasing mean radius but preserving its shape ( comp .
it was also demonstrated that this motion corresponds to a coherent superposition of vortex states @xmath80 ( @xmath81 $ ] , where @xmath82 is the azimuthal angle in the x - y - plane ) with different winding number @xmath83 . in this section
we extend this result to an interacting bose gas .
we start from eq .
( [ gpeham ] ) including the resonant optical potential ( [ vlvort ] ) .
as is well - known the time evolution of the expectation value of any operator @xmath84 is governed by the equation @xmath85 \rangle \ ; .\ ] ] it is straightforward to derive from this relation the corresponding equations for the position operators . observing that @xmath86 for ( @xmath87 ) , where the light - induced force @xmath88 is given by @xmath89 . for
a bec in the ground state the condensate s mean position and mean momentum are initially zero ( @xmath90 ) so that we find for the solution @xmath91 + \vec{e}_y \omega_\perp t \sin ( \omega_\perp t ) \ } \ ; , \label{trajeclsg}\ ] ] with @xmath92 .
again the condensate performs a rotation around the trap center whereby the radius of the rotation increases linearly in time .
this result was to be expected since usual interatomic potentials are translation - invariant and therefore affect the center - of - mass motion only indirectly .
it provides an extension of ehrenfest s theorem for interacting bose condensates .
we remark that the result ( [ trajeclsg ] ) deviates from the corresponding result of ref .
@xcite by the factor @xmath93 inside the square brackets .
this difference arises because in ref .
@xcite a rotating wave approximation ( rwa ) has been made .
as usual for rwa the difference is negligible for large times @xmath94 . to demonstrate that , as in the noninteracting case , the shape of the condensate is not affected by the optical potential we consider the time evolution of the variance of the position operators . using @xmath95 and defining @xmath96 we arrive at @xmath97 because neither the center - of - mass motion nor the optical potential enter into these equations we can infer that the shape of the condensate remains unaffected by the optical potential .
it is worth noting that one can derive from eq .
( [ varianz ] ) the mean energy conservation law @xmath98 + \omega_z^2 ( \delta z)^2\ } + g \int d^3 u |\psi ( \vec{u})|^4 = \mbox { const . } \ ; , \ ] ] which nicely demonstrates the relation between the interaction energy and the widths of the condensate s wavefunction .
we conclude that , albeit the potential ( [ vlvort ] ) seems to be perfectly suited for this task , neither a pure vortex state nor even a vortex line far away from the center of the trap is created .
however , this does not rule out the possibility to form a vortex state during the formation of the condensate by directly cooling the gas in a rotating trapping potential
. our result may also be helpful to understand why the recent experiments at mit @xcite attempting to observe persistent flows around a potential wall at the trap center by rotating the trap can not achieve the expected result . to understand why the potential ( [ vlvort ] ) produces a superposition of vortex states rather than a pure one we observe that it couples only neighboring eigenstates of the trap .
this can be seen by expressing the coordinates @xmath99 in terms of the corresponding annihilation operators of the trap ( i.e. , @xmath100 , for instance ) .
since the potential is linear in these operators it couples states with energy @xmath101 ( in the noninteracting case ) to states with energy @xmath102 .
if the coupling is resonant ( @xmath103 ) the potential ( [ vlvort ] ) not only resonantly transforms the ground state to the first excited vortex state , but it also resonantly transfers the @xmath83th vortex state ( with energy @xmath101 ) to the @xmath104th vortex state , because of the equal energy interval between these energy eigenstates . as a result , the long - time evolution of the system driven by the rotating force will lead to a superposition of vortex states with different winding number .
this implies that to create a pure vortex state one must destroy the equal interval between the trap energy levels to avoid the simultaneous excitations of different vortex states .
we will analyze this in the next section
in the preceding sections we have presented a scheme to transfer a harmonically trapped condensate into a rotating state which is a coherent superposition of vortex states .
we also analysed why the scheme does not create a pure vortex state if the trapping potential is harmonic : the equal interval between the energy levels in a harmonic trap causes the condensate to climb up a ladder of resonant transitions into different vortex states .
the situation is different if an anharmonic ( yet still rotationally symmetric ) potential @xmath105 is used in eq .
( [ gpe ] ) for which the interval between the energy levels is not equal .
therefore , if the frequency @xmath106 in the optical potential ( [ vlvort ] ) is choosen to be in resonance with the transition from the ground state to the first excited vortex state , the transition to the next vortex state will in general be out of resonance ( see fig . 2 )
this fact can be exploited to prepare a pure vortex state in an anharmonic trap . to estimate this
quantitatively we first consider the ideal bose gas and expand the wavevector in terms of the simultaneous eigenstates @xmath107 of the hamiltonian ( [ hcms ] ) and the orbital angular momentum @xmath108 , @xmath109 c_{m , s}(t ) \psi_{m , s}$ ] .
the eigenstates fulfill the relations @xmath110 and @xmath111 .
it should be noted that the inclusion of the phase factor @xmath112 $ ] in this expansion is equivalent to the description of the system in a frame of reference rotating with angular velocity @xmath113 ( where we left open the possibility that the frequencies can vary in time ) . in this description
the broken symmetry between states of positive ( @xmath114 ) and negative ( @xmath115 ) angular momentum becomes obvious .
since the latter are counterrotating for @xmath116 transitions to these states are highly suppressed . using this expansion the schrdinger equation ( [ gpe ] ) for the ideal condensate with @xmath117
can be reduced to @xmath118 with @xmath119 .
we assume that , as it is the case for the harmonic potential , @xmath120 has a unique ground state with energy @xmath121 and that the energy levels @xmath122 are well separated , with @xmath123 being the lowest vortex excitation . in this case , if we choose @xmath124 the transition from @xmath125 to @xmath126 is in resonance while the transition from @xmath126 to @xmath127 is detuned by the amount @xmath128 . furthermore ,
if the condition @xmath129 is fulfilled , the transition from @xmath130 to @xmath127 will be highly supressed so that we effectively have a two - level system consisting of the states @xmath125 and @xmath130 .
it is obvious that after a time @xmath131 the complete population will have been transferred to the first excited vortex state , where we have defined the rabi frequency @xmath132 of the transition between @xmath125 and @xmath126 this situation is mathematically very similar to the resonant excitation of the first excited state in an atom by a laser beam . in this case
one can safely neglect all states beside the ground and the first excited state , too . as a specific example we consider a two - dimensional anharmonic oscillator for which the trapping potential is given by @xmath133 , where @xmath134 is the two - dimensional radial variable and @xmath135 determines the strength of the potential .
the neglection of the z - direction does not alter the predictions for the ideal bose gas and is justified for the interacting bose gas if excitations along the z - axis are sufficiently supressed .
approximate eigenstates of the resulting hamiltonian can be obtained by application of ritz s variational method . for later use
we include the nonlinearity and minimize the expression @xmath136 \psi + \frac{g_{2d}}{2 } |\psi|^4 \right \}\ ] ] for the conserved energy .
the coupling constant @xmath137 that we use for an effective description of the vortex creation in two dimensions is related to the three - dimensional coupling parameter @xmath138 by @xmath139 . using the ansatz @xmath140 we find for the variational energy eigenvalues the expression @xmath141 , with @xmath142^{1/3 } \quad , \quad e^{(nl)}_{n,0 } = { \cal e } { \cal g}f_n ( n+1 ) \left [ \frac{2(n+2)}{1+{\cal g } f_n } \right ] ^{1/3 } \ ; .\ ] ] @xmath143 represents the contribution of the linear part of the hamiltonian ( kinetic and potential energy ) and @xmath144 denoted the contribution of the nonlinear interaction energy .
we have introduced the energy scale @xmath145 , the dimensionless interaction parameter @xmath146 , and the numerical factor @xmath147 $ ] .
the characteristic length scale of the system is given by @xmath148^{1/6}$ ] .
for the ideal bose gas we set @xmath149 and find for the lowest energy eigenvalues the values @xmath150 , @xmath151 , @xmath152 , and @xmath153 . we have determined the energy @xmath154 by using the ansatz @xmath155 for the wavefunction .
the largeness of @xmath154 allows us to neglect all transitions to this state .
the interaction matrix elements are of the order of @xmath156 for these four states , but we will need only the value for @xmath157 and @xmath158 .
the transition from @xmath159 to @xmath160 will be in resonance if we choose the frequency difference @xmath161 .
the transition from @xmath160 to @xmath162 is then supressed if the interaction matrix element @xmath163 is much smaller than the energy difference @xmath164 .
this is guaranteed if we set the transition energy @xmath165 equal to @xmath166 .
the laser scheme presented above then will create a vortex state after a time @xmath131 .
we have depicted this physical situation ( to scale ) in fig .
3 . to estimate its experimental realizability and to compare it with the interacting bose gas we consider the case that the numerical value of the trapping potential strength @xmath135 is given by @xmath167 j / m@xmath168 so that for @xmath169rb atoms ( @xmath170 kg ) the trap size turns out to be @xmath171 m ( these numerical values turn out to be convenient for the numerical calculations presented below ) .
we then find for the energy scale @xmath172 j and for the characteristic time for vortex creation @xmath173 ms .
this well within the range of current experiments with bose - einstein condensates .
we mention that we have also made a numerical study of this creation process for the numbers given here .
the result is in excellent agreement with our analytical calculations but will not be presented here since the analytical results already perfectly describe the physical situation for a non - interacting bose gas . for all details of the numerical procedure
we refer to the next section where it is applied to the correponding case of an interacting bose gas .
to study the influence of the atomic interaction analytically we assume that the nonlinearity in eq .
( [ gpe ] ) is weak enough so that we still can use the variational ansatz ( [ vari ] ) .
this is not allowed for the strong interaction regime where this ansatz becomes meaningless , but it includes the case that the interaction energy is of the order of the kinetic and potential energy and thus overcomes the restriction of perturbation theory .
we further assume that radial excitations of the condensate are unimportant because the energy levels @xmath174 ( @xmath175 ) are considerably larger than the corresponding lowest value @xmath176 for the same angular momentum @xmath177 .
we then only need to take into account the states @xmath178 .
the numerical results presented below further justify this assumption .
writing the wavefunction as @xmath179 \psi_{n,0}(\vec { x})$ ] and projecting eq .
( [ gpe ] ) onto @xmath178 we arrive at @xmath180 here we have introduced the energies @xmath181 for scattering between atoms in state @xmath61 and @xmath182 , and @xmath183 for anomalous scattering of atoms .
to solve eq .
( [ ff ] ) analytically we first make the assumption that only the two states @xmath125 and @xmath126 are populated . though this is not justified for a quantitative analysis it provides very useful physical insight into the effectiveness of the process of vortex creation .
( [ ff ] ) then reduces to @xmath184 this form of the nonlinear schrdinger equation is particularly suited to study the effect of a time - dependent variation of the laser frequencies @xmath185 .
it was originally proposed in ref .
@xcite to use such a variation to compensate for the nonlinear interaction energy between the atoms . here
we will go into more detail and examine the question of which variation is the most effective one to create a vortex state .
let us first assume that @xmath186 is constant and chosen so that initially ( @xmath187 ) the transition from @xmath125 to @xmath126 is on resonance , i.e. , @xmath188 after some time the population of the two states will be altered ( @xmath189 ) so that the magnitude of the terms in curly brackets in eqs .
( [ 2levnl1 ] , [ 2levnl2 ] ) is changed .
thus , the 2-level system is driven out of resonance by the nonlinear atomic interaction and a complete transfer of the condensate from @xmath125 to @xmath126 becomes impossible .
it is not hard to see that a time - dependent variation of @xmath186 can compensate for this interaction - induced shift of the resonance frequency . in ref .
@xcite we proposed to vary @xmath70 linearly in time . while this works well if the rate of change is suitably chosen , it has the disadvantage that one needs a relatively long time to create a vortex state .
this is similar to a landau - zener transition @xcite which is destroyed by nonadiabatic transitions if the frequency s rate of change is too high ( see ref .
@xcite , for instance ) . to improve this situation
we assume that @xmath186 is implicitely chosen so that the transition between @xmath125 and @xmath126 is _ always _ on resonance , @xmath190 under this condition the solution of eqs .
( [ 2levnl1 ] , [ 2levnl2 ] ) is given by @xmath191 where the phase factor @xmath192 is easy to calculate but not needed here .
this solution corresponds to a resonant rabi oscillation between the two states .
hence the two - level model predicts that the creation of a vortex state can be as effective as in the linear case provided the nonlinear interaction energy is compensated for .
the corresponding variation of the frequency difference @xmath186 can be found by reinserting eq .
( [ 2levlsg ] ) into eq .
( [ rescond ] ) and turns out to be a simple sinusodial variation with frequency @xmath193 .
such a sinusodial variation of the frequency difference @xmath186 can be realized by simple frequency modulation techniques .
the above analysis was made under the oversimplified assumption that we only need two levels to take into account .
the influence of other modes can be studied in several ways .
for instance , they can be treated as an elementary excitation around the solution ( [ 2levlsg ] ) . for the sake of brevity
we will not discuss the analytical treatment here and present instead numerical solutions of the full nonlinear schrdinger equation ( [ gpe ] ) .
_ numerical solution : _ our numerical procedure started from the ground state of the interacting condensate in the quartic anharmonic potential .
this state was found by propagating a gaussian trial wavefunction in imaginary time and normalizing it between each step .
the formation of a vortex state was studied by using the modified split - step method presented in ref .
this consists in approximating the time evolution operator @xmath194 $ ] for small @xmath195 as @xmath196\exp [ -i \delta t t\hbar ] \exp [ -i \delta t v/(2 \hbar)]$ ] , where @xmath197 and @xmath198 denote the kinetic and potential part of the hamiltonian , respectively . the terms containing @xmath198 are diagonal in position space and can easily be applied to the wavefunction .
the term involving @xmath197 is computed by transforming the wavefunction to momentum space by using the fast - fourier - transformation algorithm ( fft ) , applying the operator , and performing the inverse fft .
the physical parameters for the trap and the rb atoms are the same as given at the end of the previous section ( trap strength @xmath199 j / m@xmath168 , trap size @xmath171 m , mass @xmath170 kg ) .
in addition , the two - dimensional nonlinearity parameter @xmath137 was assumed to be @xmath200 j m@xmath201 . for rb atoms the scattering length is about 5 nm .
therefore , if we relate an effective 3-dimensional peak density @xmath202 to the peak of the 2-dimensional wavefunction @xmath203 by @xmath204 we can infer that this choice of @xmath137 corresponds to a peak density @xmath202 of about @xmath205 @xmath206 .
the strength of the optical potential ( [ vlvort ] ) was determined by setting @xmath207 j / m .
it turned out that for this system a grid size of @xmath208 points was accurate enough for the numerical study of the creation process .
the most laborious part of the numerical simulation was to find the proper variation of the frequency difference @xmath186 .
it is not difficult to find an approximation of the physical parameters in eq .
( [ rescond ] ) by use of the variational wavefunctions ( [ vari ] ) .
however , in order to get satisfactory results , these parameters must be determined more precise than the variationally calculated values which deviate from the exact values by up to 10% .
we finally found that the best results are produced if the sinusodial variation is delayed for a time @xmath209 ms to compensate for effects of other modes than @xmath125 and @xmath126 .
thus the frequency difference was given by @xmath210 with @xmath211 s@xmath212 , @xmath213 s@xmath212 , and @xmath214 s@xmath212 .
the numerical results for this choice of parameters are shown in fig . 4 and 5 .
4a - d show the modulus squared of the wavefunction at different times of the evolution ( 0 ms , 87 ms , 142 ms , 247 ms ) .
it shows that for a moderately interacting bose gas in an anharmonic trap an almost pure vortex state can be created .
5 displays the phase of the wavefunction at the same times ( the coarse - grained structure at the border of the figures represents numerical noise ) . fig .
5b demonstrates that during the evolution two vortex lines are created , one of which finally becomes the vortex state in fig.4d and the other slips out of the trap .
this situation can be described reasonably well by assuming that the wavefunction is a superposition of the form @xmath215 , where @xmath216 describes a vortex state with angular momentum @xmath217 .
4b shows the corresponding deformation of the wavefunction .
the modulus of the expansion coefficients @xmath218 can be approximately determined by looking at fig . 5 since the closer one ( two ) vortex lines are to the origin the larger the value of @xmath219 ( @xmath220 ) is .
we also mention that the vortex lines rotate around the origin during the creation process .
this can be seen in an animation of the creation process @xcite .
we have presented a scheme based on four laser beams which induces a spatially homogeneous force to transfer angular momentum to a trapped condensate . for a harmonic trap
this scheme is equivalent to a rotation of the trapping potential around its original position . in this case
the condensate will be prepared into a superposition of pure vortex states , corresponding to a circular motion of the condensate as a whole around the original trap center .
the reason for the creation of a superposition rather than a pure vortex state is that the energy eigenvalues of the trap are equidistant for a harmonic trap .
thus , transitions to neighboring states are simultaneously resonant . in an anharmonic trap
this phenomenon does not occur .
if the intensity of the laser beam is sufficiently low one can tune the transition from the ground state to the first excited vortex state into resonance while leaving the transition to the second excited vortex state far off - resonant .
the interaction - induced energy shift appearing in a non - ideal bec can be compensated by a sinusodial variation of the rotation frequency of the force .
it then is possible to create a pure vortex state in an anharmonic trap . beside employing an anharmonic trap one alternative method to create a vortex state
is to use raman transitions induced by laguerre - gaussian laser beams , as it was propsed in refs .
@xcite . in this case
a single laguerre - gaussian laser beam already carries an orbital angular momentum which can be transferred to the condensate .
it is of interest to understand why the proposal with raman transitions also works for a harmonic trapping potential and is not restricted to anharmonic potentials . in the notation used in the present paper
the basic mechanism of the scheme based on raman - transitions is that laser beams of the laguerre - gaussian type couple the condensate , which is initially in the ground state @xmath221 of the trap and the internal state @xmath222 , to a vortex state @xmath223 and the internal state @xmath224 . here
the state vectors are defined by @xmath225 .
the interaction hamiltonian is then roughly of the form @xmath226 .
this combination of internal transition and excitation of the center - of - mass motion guarantees that the condensate never can occupy higher vortex states like @xmath227 since the condensate can only oscillate between the states @xmath228 and @xmath229pmarzlin / index.html _ | we study the creation of vortex states in a trapped bose - einstein condensate by a rotating force . for a harmonic trapping potential
the rotating force induces only a circular motion of the whole condensate around the trap center which does not depend on the interatomic interaction . for the creation of a pure vortex state
it is necessary to confine the atoms in an anharmonic trapping potential .
the efficiency of the creation can be greatly enhanced by a sinusodial variation of the force s angular velocity .
we present analytical and numerical calculations for the case of a quartic trapping potential .
the physical mechanism behind the requirement of an anharmonic trapping potential for the creation of pure vortex states is explained .
15.5 cm 0.2 cm @xmath0 + |
a new generation of astronomical instrumentations is now making it possible to image the thermal emission from the disks of dust that surround some stars .
the disks that have been imaged around main sequence stars ( i.e. , the vega - type , or debris , disks ) have aroused considerable interest because it is hoped that interpretation of their structure will provide valuable information about the evolution of planetary systems , possibly even leading to the indirect detection of planets hiding in the disks .
one such disk is that around the a0v star hr 4796a .
mid - ir ( 10 - 20 @xmath1 m ) images of the hr 4796 disk ( @xcite ; @xcite ; @xcite ) show that its emission is concentrated in two lobes , one on either side of the star , indicating that the disk is being observed nearly edge - on , and that its inner region is almost completely devoid of dust . the same double - lobed feature is seen in nicmos ( 1.1 @xmath1 m ) images of the disk s scattered light ( @xcite ) .
the size of the disk s central cavity , which was previously inferred from the star s spectral energy distribution ( @xcite ) , is approximately 40 au in radius from the star , about the same size as the solar planetary system . since the age of the hr 4796 system , @xmath2 myr ( @xcite ; @xcite ) places it at a stage in its evolution when the formation of any planets is expected to be almost complete ( e.g. , @xcite ) ,
many authors have speculated that the central cavity could be indicative of planetary formation in this inner region ( @xcite ; @xcite ; @xcite ; @xcite ) .
a similar double - lobed feature has been observed in the disk around the star fomalhaut , with a similar explanation proposed for its origin ( @xcite ) .
some authors , however , remain skeptical about the existence of planets in these systems ( e.g. , @xcite ) .
observations of the hr 4796 disk show a further interesting feature : the disk s lobes appear to be of unequal brightness , although this observed asymmetry is of low statistical significance ( @xmath3 ) .
mid - ir observations of the disk are described in a companion paper by telesco et al .
( 1999 , hereafter t99 ) , and their ihw18 ( 18.2 @xmath1 m ) waveband observation ( their fig .
4b ) is reproduced in this paper in fig .
their observation suggests that the ne lobe ( on the left of the image in fig .
[ fig8]a ) is @xmath0% brighter than the sw lobe .
this lobe asymmetry may also be apparent in the nicmos ( 1.1 @xmath1 m ) images of the disk ( @xcite ) , and in the @xmath4 @xmath1 m images of koerner et al .
the brightnesses of the fomalhaut disk s lobes also appear to be asymmetric ( @xcite ) .
there are many possible explanations for the hr 4796 disk s lobe asymmetry .
this paper describes a model of the t99 observation that provides one possible dynamical explanation for the asymmetry : that it is the long - term consequence of the gravitational perturbations of one or more massive bodies on the disk ( i.e. , the consequence of the system s `` secular perturbations '' ) . quite apart from any speculation about a nascent planetary system in the inner region of the hr 4796 disk , we know that the disk must have been gravitationally perturbed , since hr
4796 is a visual binary system .
the m dwarf star hr 4796b , located at a projected distance of @xmath5 au ( @xcite ) , is the common proper motion companion of hr 4796a ( @xcite ) .
an understanding of how asymmetries in the observed structure of the zodiacal cloud , the tenuous disk of dust in the solar system , are linked to the solar system s planets ( @xcite ) , shows that if there is at least one massive perturber in the hr 4796 system that is on an eccentric orbit , then the system s secular perturbations would have caused the disk s center of symmetry to be offset from the star ( @xcite ) .
this offset would mean that the material in one of the disk s observed lobes is closer to the star than that in the other lobe ; consequently this lobe would be hotter and brighter .
the aim of this paper is to ascertain how large the perturbations would have to be to cause the observed 5% asymmetry , and to discuss whether perturbations of this magnitude are physically realistic , or even to be expected , in this system .
since this paper is based on the interpretation of a circumstellar disk observation , it starts in [ sec - ocd ] with a consideration of how information about a disk is stored in an observation .
[ sec - ddc ] gives a comprehensive discussion of the physical processes that govern the dynamical evolution of a disk s particles , and sets up a broad theoretical framework ( the `` dynamic disk '' ) with which to investigate a disk s structure .
[ sec - spd ] then shows how secular perturbations cause a disk s structure to have offset and warp asymmetries , and provides evidence that these asymmetries have been observed in the structure of the zodiacal cloud .
[ sec - hr4796 ] uses the theoretical discussions of [ sec - ocd]-[sec - spd ] to create an offset model for the hr 4796 disk that matches the 18.2 @xmath1 m brightness distribution observed by t99 .
the interpretation of this model is discussed in
[ sec - disc ] .
the observed brightness of a circumstellar disk comes from two sources : starlight that has been absorbed by the disk particles and re - emitted as thermal radiation ( primarily at mid - ir , far - ir , and submillimeter wavelengths at @xmath6 @xmath1 m ) , and starlight that has been scattered by the disk particles ( primarily at optical and near - ir wavelengths at @xmath7 @xmath1 m ) .
this paper only discusses a disk s thermal emission .
a particle of diameter @xmath8 , that is at a distance @xmath9 from a star , is heated by the stellar radiation to a temperature @xmath10 that can be calculated from the equilibrium between the energy that the particle absorbs and that which it re - emits as thermal radiation .
this temperature depends on the particle s optical properties ( @xcite ) : @xmath11^{1/4}*t_{bb } , \label{eq : tdr}\ ] ] where this equation must be solved iteratively , since the particle s temperature appears on both sides of the equation , @xmath12 and @xmath13 are the particle s absorption efficiency averaged over the stellar spectrum ( which can be approximated as that of a black body radiating at the star s effective temperature @xmath14 ) and the spectrum of a black body radiating at a temperature @xmath10 , and @xmath15 is the equilibrium temperature of the particle if it were a black body : @xmath16 where @xmath15 is given in k , @xmath17 is the ratio of the particle s cross - sectional area to its surface area ( e.g. , spherical particles have @xmath18 and @xmath19 , giving @xmath20 ) , and @xmath21 and @xmath22 are the luminosities of the star and the sun .
if this particle is at a distance @xmath23 from the earth , the contribution of its thermal emission to the flux density at a wavelength , @xmath24 , received at the earth is given by : @xmath25\omega(d ) , \label{eq : flux}\ ] ] where @xmath26 is the planck function , and @xmath27 is the solid angle subtended at the earth by the cross - sectional area of the particle .
a circumstellar disk consists of particles with a range of sizes , compositions and morphologies . throughout this paper , however , disk particles are assumed to have the same composition and morphology , and a particle s size is characterized by its diameter , @xmath8 .
disk particles span a range of sizes from @xmath28 , the smallest ( probably submicron - sized ) particles sustainable for a given disk , up to @xmath29 , the largest ( probably kilometer - sized ) members of the disk that were formed from the proto - planetary disk .
the spatial distribution of these particles can be defined by @xmath30 , where @xmath31 is the volume density ( number per unit volume ) of particles in the size range @xmath32 at a location in the disk defined by @xmath9 , the radial distance from the star , @xmath33 , the longitude relative to an arbitrary direction , and @xmath34 , the latitude relative to an arbitrary reference plane . however , since it is a particle s cross - sectional area that is apparent in an observation ( eq . [ [ eq : flux ] ] ) , a disk s observable structure is better defined in terms of @xmath35 , the cross - sectional area per unit volume per unit diameter .
the definition of a disk s structure can be simplified by assuming the size distribution of its particles to be independent of @xmath33 and @xmath34 : @xmath36 where @xmath37 is the proportion of the total cross - sectional area of the disk at @xmath9 that is in particles in the size range @xmath32 , and @xmath38 is the spatial distribution of cross - sectional area of particles of all sizes in the disk . to determine the observed brightness of a circumstellar disk in , for example , one pixel of an image of the disk , two components of the observation
must be defined : the vector , @xmath39 , which extends from the observer to the disk , and which describes how the line of sight intersects the disk in terms of @xmath9 , @xmath33 , and @xmath34 ; and the solid angle of the observation , @xmath40 , where @xmath41 for pixels of width @xmath42 radians .
consider a volume element along this line of sight that is at a location in the disk defined by @xmath43 , and that has a length @xmath44 ; the element volume is @xmath45 .
the contribution of the thermal emission of the particles in this element to the disk s brightness in the observation is given by : @xmath46 \sigma(d , r,\theta,\phi)dd \mathbf{dr } , \label{eq : brightness } \\ & = & p(\lambda , r)\sigma(r,\theta,\phi)\mathbf{dr } , \label{eq : fnu}\end{aligned}\ ] ] where equation ( [ eq : fnu ] ) uses the simplification for the disk structure given by equation ( [ eq : sig2 ] ) , and @xmath47\bar{\sigma}(d , r)dd .
\label{eq : p}\ ] ] thus , the brightness of this element is not affected by the solid angle of the observation , neither is it affected by the distance of the element from the earth . equation ( [ eq : p ] ) can also be written as : @xmath48 \rangle_{\sigma(d , r ) } ; \label{eq : p2}\ ] ] i.e. , @xmath49 is the combination of the particles optical properties given by @xmath50 $ ] averaged over the disk s cross - sectional area distribution , @xmath51 .
thus , @xmath49 can also be given by : @xmath52 , \label{eq : pdtyp}\ ] ] where @xmath53 is the size of disk particle that characterizes ( and hence dominates ) the disk s emission .
this characteristic particle size could be different in different wavebands , as well as at different distances from the star , but always lies in the range @xmath54 ; its value can be found by considering the relative contribution of particles of different sizes to @xmath49 . however , unless the optical properties of the particles prevent it , the particles that dominate a disk s emission are also those that contribute most to its cross - sectional area , i.e. , they are those that dominate the disk s structure , @xmath55 .
this is usually the case for mid - ir ( @xmath56 @xmath1 m ) observations , such as those of t99 ( see e.g. , [ ssec - mpp ] ) .
the total brightness of the disk in this observation is the integral of equation ( [ eq : brightness ] ) over @xmath39 .
thus , a disk observation is composed of three parts : the disk s structure , the optical properties of the disk particles , and the orientation of the disk to the line of sight .
an image of the disk is made up of many pixels , each of which has a different line of sight vector , @xmath39 , and corresponding brightness .
the image at the detector has been convolved with the observational point spread function ( psf ) , a combination of the seeing conditions and the telescope and instrumental optics , which , in the diffraction - limited case , for an ideal instrument , can be approximated as gaussian smoothing with @xmath57 , where this @xmath8 is the diameter of the telescope .
the observed image also contains photospheric emission from the point - like star , as well as random noise fluctuations .
a useful image of the disk can be recovered by subtracting the image of the star , but the accuracy of this subtraction depends on how well both the psf and the stellar flux density are known .
further smoothing of the image to increase the signal - to - noise may also prove useful .
this section outlines the theoretical framework upon which later discussion of circumstellar disk structure ( such as how secular perturbations affect this structure ) is based .
a disk is a dynamic entity , the constituent particles of which are undergoing constant dynamical and physical evolution ; [ ssec - pp ] gives an extensive discussion of the physical processes acting on disk particles .
[ ssec - ddpc ] then summarizes our understanding of the dynamic disk by showing how disk particles can be categorized according to the dominant physical processes affecting their evolution .
if we are to make generalizations about the physical processes relevant to circumstellar disk evolution , then the zodiacal cloud is the best example of a circumstellar disk on which to base this understanding , since its properties have been determined with a certain degree of confidence .
this confidence stems from the wealth of observational , theoretical , and physical evidence describing its present state , its evolutionary history , and the physical environment of the system it is in .
the physical processes that are described in this section are those thought to have dominated the evolution of the zodiacal cloud , supposedly since the sun reached the main sequence ( e.g. , @xcite ; @xcite ) .
these processes should serve as an adequate basis for an understanding of the evolution of the circumstellar disks around other main sequence stars .
there is , however , one obvious distinction between the zodiacal cloud and exosolar dust disks .
the emission observed from the zodiacal cloud is dominated by that from dust in the inner solar system ( within 5 au of the sun ) , which has its origins in the asteroid belt ( @xcite ; @xcite ) and the short period comets ( @xcite ) .
in contrast , the emission observed from exosolar dust disks is dominated by that originating from dust in regions analogous to the kuiper belt in the solar system , i.e. , @xmath58 au from the star ( e.g. , @xcite ) . relatively little
is known about the kuiper belt , but , like the inner solar system , it appears to be populated with many asteroid- or comet - like objects that are probably the remnants of the solar system s planetary formation phase ( e.g. , @xcite ) .
the dominant force acting on all but the smallest disk particles is the gravitational attraction of the star : @xmath59 where @xmath60 is the gravitational constant , @xmath61 is the mass of the star , @xmath62 is the mass of the particle , and @xmath9 is the distance of the particle from the star .
all material is assumed to orbit the star on keplerian ( elliptical ) orbits , with other forces acting as perturbations to these orbits .
the orbit of a particle is defined by the following orbital elements : the semimajor axis , @xmath63 , and eccentricity , @xmath64 , that define the radial extent and shape of the orbit ; the inclination , @xmath65 , and longitude of ascending node , @xmath66 , that define the plane of the orbit ( relative to an arbitrary reference plane ) ; and the longitude of pericenter , @xmath67 , that defines the orientation of the orbit within the orbital plane ( relative to an arbitrary reference direction ) .
the orbital period of the particle is given by : @xmath68 where @xmath69 is given in years , @xmath70 au is the semimajor axis of the earth s orbit , and @xmath71 is the mass of the sun . at any instant ,
the location of the particle in its orbit is defined by the true anomaly , @xmath72 , where @xmath73 and @xmath74 at the pericenter and apocenter , respectively , and its distance from the star , @xmath9 , and its velocity , @xmath75 , are defined by : @xmath76 since at any given time a particle could be at any point along its orbit , its contribution to the distribution of material in a disk can be described by the elliptical ring that contains the mass of the particle spread out along its orbit , the line density of which varies inversely with the particle s velocity ( eq . [ [ eq : vgrav ] ] ) .
each disk particle has an orbit defined by a different set of orbital elements , with a contribution to the spatial distribution of material in the disk that can be described by a corresponding elliptical ring .
thus , a disk s structure can be defined by the distribution of orbital elements of its constituent particles , @xmath77 , where @xmath78 is the number of disk particles with sizes in the range @xmath32 , and orbital elements in the range @xmath79 , @xmath80 .
a disk s orbital element distribution can be quantified in terms of the evolutionary history of the system and the physical processes acting on the disk s particles . using techniques such as those described in
[ sssec - simul ] , the resulting structure can then be compared with the disk s observable structure , @xmath81 , to link a disk observation with the physics of the particles in that disk .
it is often convenient to discuss the dependence of the orbital element distribution on the different parameters separately ; e.g. , the distribution of semimajor axes , @xmath82 , is defined such that @xmath83 is the total number of particles with orbital semimajor axes in the range @xmath84 .
a typical disk particle is created by the break - up of a larger `` parent '' body , either as the result of a collision with another body , or simply by its disintegration .
this parent body could have been created by the break - up of an even larger body , and the particle itself will most likely end up as a parent body for particles smaller than itself .
this `` collisional cascade '' spans the complete size range of disk material , and the particles that share a common ancestor are said to constitute a `` family '' of particles .
the size distribution that results from this collisional cascade can be found from theoretical arguments ( @xcite ) : @xmath85 where @xmath86 ; this distribution is expected to hold for disk particles that are large enough not to be affected by radiation forces ( [ sssec - rf ] ) .
a disk with this distribution has its mass , @xmath87 , concentrated in its largest particles , while its cross - sectional area , @xmath88 , is concentrated in its smallest particles .
collisions in such a disk are mostly non - catastrophic ( see eqs .
[ [ eq : dimp]]-[[eq : fcc ] ] ) , and a particle in this disk is most likely to be broken up by a particle that has just enough mass ( and hence energy ) to do so .
this in turn means that collisional fragments have velocities , and hence orbital elements , that are almost identical to those of the original particle ; i.e. , in the absence of other forces , all members of the same family have identical orbits .
due to the interaction of the competing physical processes , the size distribution of disk particles that are affected by radiation forces is only really understood qualitatively ( [ sssec - rf ] and [ sssec - pcsig ] ) ; their distribution is particularly important , since , in general , a disk s cross - sectional area ( and hence its observable structure ) is concentrated in these smaller particles . the importance of collisions in determining a particle s evolution depends on its collisional lifetime , which is discussed in appendix [ app - tcoll ] .
the collisional lifetime of the particles that constitute most of a disk s cross - sectional area ( i.e. , those that are expected to characterize the disk s mid - ir emission , [ ssec - los ] ) , can be approximated by ( eqs .
[ [ eq : tcol5 ] ] and [ [ eq : fccdtyp ] ] ) : @xmath89 where @xmath90 is the average orbital period of particles at @xmath9 ( eq . [ [ eq : tper ] ] with @xmath63 replaced by @xmath9 ) , and @xmath91 is the disk s effective face - on optical depth ( eq . [ [ eq : taueffdefn ] ] ) , which would be equal to the disk s true optical depth if its particles had unity extinction efficiency . the collisional lifetime of particles with @xmath92 can be considerably longer than that of equation ( [ eq : tcol6 ] ) ( see e.g. , eq .
[ [ eq : tcol7 ] ] ) , and a disk s largest particles , those for which @xmath93 , may not have suffered any catastrophic collisions since they were first created ; such particles are primordial particles .
the cascades of very young disks may still contain a significant proportion of primordial particles ; i.e. , their cascades may not be fully evolved .
the collisional cascade theory is well - supported by evidence from the zodiacal cloud .
the size distribution of the largest ( @xmath94 km ) members of the zodiacal cloud s collisional cascade , the observable asteroids , is well - approximated by equation ( [ eq : nd ] ) ( @xcite ; durda , greenberg , & jedicke 1998 ) ; the distribution of the very largest ( @xmath95 km ) asteroids deviates from this distribution , however , because of the transition from strength - scaling to gravity - scaling for asteroids larger than @xmath96 m ( @xcite ) .
the size distribution of the zodiacal cloud s medium - sized ( 1 mm @xmath97 km ) members is also expected to follow equation ( [ eq : nd ] ) ( @xcite ) , but there is no observational proof of this , since these members are too faint to be seen individually , and too few to be studied collectively ( @xcite ) .
there is , however , proof that the zodiacal cloud s collisional cascade extends from its largest members down to its smallest dust particles : the shapes of the `` dust band '' thermal emission features ( @xcite ) correspond to those expected from the small ( @xmath98 @xmath1 m ) particles resulting from the break - up , some time ago , of a few very large asteroids , the largest fragments of which are still observable as the asteroids in the themis , koronis , and , possibly , eos families ( @xcite ; @xcite ) .
the size distribution of the zodiacal cloud s smallest ( @xmath99 mm ) dust particles ( e.g. , @xcite ; @xcite ) can be explained qualitatively ( e.g. , @xcite ; see also [ sssec - rf ] ) .
analysis of the collision rates of objects in the kuiper belt ( @xcite ) shows that a collisional cascade should exist here too ; there is also evidence to suggest that the kuiper belt was once more massive than it is today ( @xcite ) , meaning that in the past collisions would have played a much larger role in determining its structure than they do today , maybe even causing the supposed mass loss ( @xcite ) .
the size distribution of the observed kuiper belt objects appears to be slightly steeper than that in the inner solar system ( @xmath100 , @xcite ) , while observations have been unable , as yet , to determine its dust distribution ( @xcite ; @xcite ) , for most of the collisional cascade , gravity can be considered to be the only significant force acting on disk particles .
the smallest particles , however , are significantly affected by their interaction with the photons from the star .
[ [ ssssec - rp ] ] radiation pressure + + + + + + + + + + + + + + + + + + radiation pressure is the component of the radiation force that points radially away from the star .
it is inversely proportional to the square of a particle s distance from the star , and is defined for different particles by its ratio to the gravitational force of equation ( [ eq : fgrav ] ) ( @xcite ) : @xmath102 where @xmath103 kg / m@xmath104 , @xmath105 is the ratio of the particle s cross - sectional area to its mass ( e.g. , @xmath106 for spherical particles of density @xmath107 ) , and @xmath108 is the particle s radiation pressure efficiency , where @xmath109 accounts for the asymmetry of the scattered radiation . ] averaged over the stellar spectrum , @xmath110 .
an approximation for large particles is that @xmath111 ; thus , large spherical particles have : @xmath112 where @xmath107 is measured in kg / m@xmath113 , and @xmath8 in @xmath1 m .
this approximation is valid for particles in the solar system with @xmath114 @xmath1 m ( @xcite ) . for particles in exosolar systems ,
this limit scales with the wavelength at which the star emits most of its energy , @xmath115 ; i.e. , equation ( [ eq : beta2 ] ) is valid for spherical particles with @xmath116 @xmath1 m , where @xmath117 k is the effective temperature of the sun . since @xmath118 , this means that the smaller a particle , the larger its @xmath101 ; this holds down to micron - sized particles , smaller than which @xmath101 decreases to a level that is independent of the particle s size ( @xcite ) .
the effect of radiation pressure is equivalent to reducing the mass of the star by a factor @xmath119 .
this means that a particle for which @xmath120 moves slower around the same orbit by a factor of @xmath121 than one for which @xmath122 ( eq . [ [ eq : vgrav ] ] ) .
it also means that daughter fragments created by the break - up of a parent body move on orbits that can differ substantially from that of the parent .
the reason for this , is that while the positions and velocities of a parent and its daughter fragments are the same at the moment of break - up ( apart from a small velocity dispersion ) , their @xmath101 are different , and so the daughter fragments move in effective potentials that are different from that the parent moved in .
daughter fragments created in the break - up of a parent particle that had @xmath122 , and for which orbital elements at the time of the collision were @xmath123 , and @xmath72 , move in the same orbital plane as the parent , @xmath124 and @xmath125 , but on orbits with semimajor axes , @xmath126 , eccentricities , @xmath127 , and pericenter orientations , @xmath128 , that are given by ( burns , lamy , & soter 1979 ) : @xmath129 , \label{eq : aprime } \\ e^ { ' } & = & ( 1-\beta)^{-1}\sqrt{e^2 + 2\beta e\cos{f } + \beta^2 } , \label{eq : eprime } \\ \tilde{\omega}^ { ' } - \tilde{\omega } & = & f - f^ { ' } = \arctan{\left [ \beta \sin{f}/(\beta\cos{f } + e ) \right]}. \label{eq : wprime}\end{aligned}\ ] ] analysis of equations ( [ eq : aprime])-([eq : wprime ] ) shows that the orbits of the largest fragments , those for which @xmath130 , are similar to that of the parent . on the other hand , the smallest fragments , those for which @xmath131 , have hyperbolic orbits ( @xmath132 )
; these particles are known as `` @xmath101 meteoroids '' are @xmath101 meteoroids even if they were not created collisionally , since they `` see '' a negative mass star . ] . since @xmath101 meteoroids are lost from the system on the timescale of the orbital period of the parent ( eq .
[ [ eq : tper ] ] ) , the diameter of particle for which @xmath133 essentially defines the lower end of the collisional cascade .
however , there may also be a population of submicron particles that have @xmath134 ( @xcite ) .
the intermediate - sized fragments , those for which @xmath135 , that we call `` @xmath101 critical '' particles , have orbits that differ substantially from that of the parent . however , the point of closest approach to the star of the orbits of all daughter fragments , irrespective of their size , is the same as that of the parent : combining equations ( [ eq : aprime ] ) and ( [ eq : eprime ] ) gives the pericenter distance of daughter fragments , @xmath136 , as @xmath137 the orbits of collisional fragments with different @xmath101 from a parent particle that was on a circular orbit are shown in fig .
[ fig1 ] . [ [ ssssec - pr ] ] poynting - robertson ( p - r ) light drag + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the component of the radiation force tangential to a particle s orbit
is called the p - r drag force .
this force is also proportional to @xmath101 .
it results in an evolutionary decrease in both the semimajor axis and eccentricity of the particle s orbit ( @xcite ) : @xmath138 where @xmath139 au@xmath104/year .
p - r drag does not change the plane of the particle s orbit , @xmath140 ; neither does it affect the orientation of the particle s pericenter , @xmath141 . for a particle with zero eccentricity , equation ( [ eq : adot ] ) can be solved to find the time it takes for the particle to spiral in from a radial distance of @xmath142 to @xmath143 : @xmath144/\beta \label{eq : tpr},\ ] ] where @xmath145 is given in years .
consider the daughter fragments created in the break - up of a parent body that was on an orbit at a distance @xmath9 from the star .
the largest fragments are broken up by collisions before their orbits have suffered any significant p - r drag evolution , while the smaller fragments , for which the p - r drag evolution is faster , can reach the star without having encountered another particle ( at which point they evaporate ) .
particles for which p - r drag significantly affects their orbits in their lifetime can be estimated as those for which their collisional lifetime ( eq . [
[ eq : tcol6 ] ] ; the use of this equation is justified in the next paragraph ) is longer than their p - r drag lifetime ( eq . [ [ eq : tpr ] ] with @xmath146 ) , i.e. , those for which @xmath147 , where @xmath148 for large spherical particles , this is also those for which @xmath149 , where @xmath150(l_\star / l_\odot)\sqrt{(m_\odot / m_\star ) ( a_\oplus / r ) } , \label{eq : dtr}\ ] ] @xmath107 is measured in kg / m@xmath113 , and @xmath151 in @xmath1 m .
consider the daughter fragments created in the break - up of an endless supply of parent bodies that are on orbits with the same semimajor axis , @xmath152 .
ignoring collisional processes , the fragments with orbits that are affected by p - r drag , those with @xmath147 , have their semimajor axes distributed from @xmath153 to @xmath154 according to ( eq . [
[ eq : adot ] ] ) : @xmath155 this corresponds to a volume density distribution that is roughly inversely proportional to distance from the star ) means a spherical shell of width @xmath156 , the volume of which is @xmath157 , would contain a number of particles that is @xmath158 ( see e.g. , @xcite ) . ] . if the collisional processes leading to the size distribution of the parent bodies , @xmath159 , still holds for the production of the p - r drag affected particles , then their size distribution is given by : @xmath160 if @xmath159 can be given by equation ( [ eq : nd ] ) with @xmath161 , the cross - sectional area of a disk s p - r drag affected particles is concentrated in the largest of these particles , while that of its unaffected particles is concentrated in the smallest of these particles ; i.e. , most of a disk s cross - sectional area is expected to be concentrated in particles with @xmath162 , justifying the use of equation ( [ eq : tcol6 ] ) for the collisional lifetime of these particles .
observations of the zodiacal cloud at 1 au show that its effective optical depth here is @xmath163 . since these particles originated in the asteroid belt at @xmath164 au , arriving at 1 au due to the p - r drag evolution of their orbits ,
the zodiacal cloud s volume density should vary @xmath165 , and its effective optical depth at 3 au should be similar to that at 1 au . assuming zodiacal cloud particles to have a density @xmath166 kg / m@xmath113 ( @xcite ) , the cross - sectional area of material in the asteroid belt
should be concentrated in particles with @xmath167(500 @xmath1 m ) ( eq .
[ [ eq : dtr ] ] ) , for which both the collisional lifetime , and the p - r drag lifetime , is @xmath168 myr .
the cross - sectional area of material at 1 au is expected to be concentrated in particles smaller than that in the asteroid belt , since many of the larger particles should have been broken up by collisions before they reach the inner solar system ; this is in agreement with observations that show the cross - sectional area distribution at 1 au to peak for particles with @xmath169 @xmath1 m ( @xcite ; @xcite ) .
also , equation ( [ eq : tcol7 ] ) with @xmath170 @xmath1 m , @xmath171 , and @xmath161 , predicts that the collisional lifetime of large bodies in the asteroid belt should be : @xmath172 where @xmath173 is given in years , and @xmath8 in km .
since the solar system is @xmath174 years old , this implies that asteroids larger than @xmath4 km should be primordial asteroids ; this is in agreement with more accurate models of the observed size distribution of these asteroids ( @xcite ) .
disk particles of different sizes can be categorized according to the dominant physical processes affecting their evolution .
particles in the different categories have different lives ; i.e. , the way they are created , their dynamical evolution , and the way they are eventually destroyed , are all different .
each of a disk s categories has a different spatial distribution ; which of these categories dominates the disk s observable structure depends on the relative contribution of each to the disk s cross - sectional area ( see [ ssec - los ] ) .
a disk s largest particles , those with @xmath130 , have orbital elements that are initially the same as , or at least very similar to , those of their parents . of these large particles , only those with @xmath175
suffer no significant p - r drag evolution to their orbits in their lifetime .
these truly large particles continue on the same orbits as those of their ancestors until they collide with a particle large enough to cause a catastrophic collision ; the resulting collisional fragments populate the collisional cascade .
the spatial distribution of these `` large '' particles in a disk is its base distribution .
the spatial distributions of a disk s smaller particles can only be understood in terms of how they differ from the disk s base distribution .
disk particles with @xmath176 , spiral in from their parent s orbits due to p - r drag , so that they are closer to the star than their parents by the time of their demise ( which could be caused either by collisions or by evaporation close to the star ) ; the spatial distribution of these `` p - r drag affected '' particles in a disk differs from the disk s base distribution in that it extends closer in to the star .
the orbits of particles with @xmath135 also undergo significant p - r drag evolution before their demise , but their original orbits are already different from those of their parents ; the spatial distribution of these `` @xmath101 critical '' particles in a disk extends both further out , and further in , from the disk s base distribution .
particles with @xmath133 leave their parents on hyperbolic orbits , and so are quickly lost from the system ; the spatial distribution of these `` @xmath101 meteoroids '' in a disk extends further out , but not further in , from the disk s base distribution .
thus , a disk comprises four particle categories , each of which has a different spatial distribution , although all are inextricably linked to that of the large particles through the collisional cascade . for disks that have @xmath177 , however , i.e. , those with @xmath178 there is no significant p - r drag evolution of any of its constituent particles .
such disks comprise just three categories , since the p - r drag affected category is empty .
the size distributions of a disk s large particles , and its p - r drag affected particles , were discussed in [ sssec - coll ] and [ sssec - rf ] .
these discussions imply that the cross - sectional area of a disk in which there is a population of p - r drag affected particles ( see eq .
[ [ eq : taueffnopr ] ] ) is dominated by particles with @xmath179 ( eq . [ [ eq : dtr ] ] ) . as a first - cut approximation , the size distribution of a disk
in which there are no p - r drag affected particles follows equation ( [ eq : nd ] ) from @xmath29 down to @xmath180 ; however , since a particle s catastrophic collision rate is affected by the size distribution of particles smaller than itself ( @xcite ; eq . [ [ eq : rcoll ] ] ) , this distribution can not be expected to hold all the way down to @xmath181 .
this means that the cross - sectional area of such a disk is concentrated in its smallest particles , and the contribution of @xmath101 critical particles to the disk s total cross - sectional area is given by @xmath182_{d(\beta=0.1)}^{d(\beta=0.5 ) } / [ d^{5 - 3q}]_{d_{max}}^{d(\beta=0.5)};\ ] ] e.g. , if @xmath86 , and @xmath118 , this means that half of the disk s cross - sectional area comes from its @xmath101 critical particles .
since @xmath101 meteoroids have hyperbolic orbits , they are expected to contribute little to a disk s cross - sectional area unless they are produced at a high enough rate to replenish their rapid loss from the system .
this could be the case if the disk was very dense , since material would pass quickly through the cascade ; such a disk would undergo considerable mass loss .
an estimate of how dense the disk would have to be for this to be the case depends on the assumptions made about the physics of collisions between small particles . for heuristic purposes ,
it is assumed here that the total cross - sectional area of @xmath101 meteoroids created by the collisional break - up of a parent body is comparable to that of the parent itself ; this is probably an underestimate if the collision is destructive , but an overestimate if the collision is erosive .
if this were the case , then the disk s @xmath101 meteoroids would dominate a disk s cross - sectional area only if their lifetime , which is of the order of the orbital period of their parents , is longer than the lifetime of these parents , which can be approximated by equation ( [ eq : tcol6 ] ) , i.e. , only if @xmath183 in conclusion , from a theoretical stand - point , there are few solid assumptions that can be made about a disk s size distribution , other than that the denser a disk is , the smaller the diameter of particles that its cross - sectional area is concentrated in .
in addition to the physical processes described in [ ssec - pp ] ,
the particles of the dynamic disk described in this section are affected by a number of perturbing processes ; these produce subtle , but perhaps observable , changes in a disk s structure .
the dominant perturbing processes in the zodiacal cloud are the secular and resonant gravitational perturbations of the planets .
secular perturbations are discussed in [ sec - spd ] .
resonant perturbations give rise to planetary resonant rings : a planet s resonant ring is an asymmetric circumstellar ring of material that co - orbits with the planet as a result of the resonant trapping of particles into the planet s exterior mean motion resonances ( not to be confused with circumplanetary dust rings ) .
there is both observational and theoretical evidence for the existence of the earth s resonant ring ( @xcite ; @xcite ) ; many of the observed kuiper belt objects are trapped in 2:3 resonance with neptune ( @xcite ) , thus forming neptune s resonant ring ( @xcite ) .
other possible perturbing processes include : stellar wind forces , that , at least for dust in the solar system , effectively increase the value of @xmath101 for p - r drag ( e.g. , @xcite ) ; lorentz forces acting on charged particles , that , while negligible for particles in the inner solar system , are increasingly important for particles at distances further from the sun ( e.g. , @xcite ) ; interactions with dust from the interstellar medium ( e.g. , @xcite ) ; the sublimation of icy dust grains , which is one of the mechanisms that has been suggested as the cause of the inner hole in the hr 4796 disk ( @xcite ) ; and the self gravity of a massive disk , which could have played an important role in determining the evolution of the primordial kuiper belt ( @xcite ) .
the orbit of a particle in a circumstellar disk that is in a system in which there are one or more massive perturbers is inevitably affected by the gravitational perturbations of these bodies . the consequent evolution of the particle s orbit can be used to obtain a quantitative understanding of the effect of these perturbations on the structure of the disk .
the gravitational forces from a planetary system that act to perturb the orbit of a particle in the system can be decomposed into the sum of many terms that are described by the particle s disturbing function , @xmath184 .
the long - term average of these forces are the system s secular perturbations , and the terms of the disturbing function that contribute to these secular perturbations , @xmath185 , can be identified as those that do not depend on the mean longitudes of either the planets or the particle ( the other forces having periodic variations ) .
consider a particle that is orbiting a star of mass @xmath61 , that also has @xmath186 massive , perturbing , bodies orbiting it .
this particle has a radiation pressure force acting on it represented by @xmath101 , and its orbit is described by the elements @xmath63 , @xmath64 , @xmath65 , @xmath66 and @xmath67 . to second order in eccentricities and inclinations , the secular terms in the particle s disturbing function are given by ( @xcite ; @xcite ; @xcite ; @xcite ) : @xmath187 \right ] , \label{eq : rsec}\ ] ] where @xmath188 is the mean motion of the particle in rad / s , @xmath189 s is one year measured in seconds , and @xmath190 where @xmath191 and @xmath192 for @xmath193 , and @xmath194 for @xmath195 , and @xmath196 are the laplace coefficients ( @xmath197 ) .
@xmath198 , @xmath199 and @xmath200 are in units of rad / s , and @xmath185 is in units of m@xmath104/s@xmath104
. the effect of these perturbations on the orbital elements of the particle can be found using lagrange s planetary equations ( @xcite ; @xcite ) .
the semimajor axis of the particle remains constant , @xmath201 , while the variations of its eccentricity and inclination are best described when coupled with the variations of its longitude of pericenter and ascending node using the variables defined by its complex eccentricity , @xmath202 , and complex inclination , @xmath203 : @xmath204 where @xmath205 . using these variables
lagrange s planetary equations give the orbital element variations due to secular perturbations as : @xmath206 where @xmath207 and @xmath208 are the complex eccentricities and inclinations of the perturbers , which have a slow temporal variation due to the secular perturbations of the perturbers on each other ( @xcite ; @xcite ) : @xmath209 where @xmath210 and @xmath211 are the eigenfrequencies of the perturber system , the coefficients @xmath212 and @xmath213 are the corresponding eigenvectors , and @xmath214 and @xmath215 are constants found from the initial conditions of the perturber system . ignoring the evolution of a particle s orbital elements due to p - r drag ,
equations ( [ eq : zdot ] ) and ( [ eq : ydot ] ) can be solved to give the secular evolution of the particle s instantaneous complex eccentricity and inclination ( a.k.a . the particle s osculating elements ) .
this secular evolution is decomposed into two distinct time - varying elements the `` forced '' , subscript @xmath72 , and `` proper '' , subscript @xmath216 , elements that are added vectorially in the complex planes ( see e.g. , fig .
[ fig2]a ) : @xmath217 \ast\exp{i(g_kt + \beta_k ) } + e_p\ast\exp{i(+at+\beta_0 ) } , \label{eq : z } \\ y(t ) = y_f(t ) + y_p(t ) & = & \sum_{k=1}^{n_{pl } } \left [ \frac { \sum_{j=1}^{n_{pl}}b_j i_{jk}}{f_k+a } \right ] \ast\exp{i(f_kt + \gamma_k ) } + i_p\ast\exp{i(-at+\gamma_0 ) } , \label{eq : y } \end{aligned}\ ] ] where @xmath218 , @xmath219 , and @xmath220 , @xmath221 are determined by the particle s initial conditions .
these equations have simple physical and geometrical interpretations .
a particle s forced elements , @xmath222 and @xmath223 , depend only on the orbits of the perturbers in the system ( that have a slow secular evolution , eqs .
[ [ eq : zj ] ] and [ [ eq : yj ] ] ) , as well as on the particle s semimajor axis ( which has no secular evolution ) .
thus , at a time @xmath224 , a particle that is on an orbit with a semimajor axis @xmath63 , has forced elements imposed on its orbit by the perturbers in the system that are defined by @xmath225 and @xmath226 .
the contribution of the particle s proper elements to its osculating elements , @xmath227 and @xmath228 , is then given by : @xmath229 and @xmath230 ; thus defining the particle s proper eccentricity , @xmath218 , and proper inclination , @xmath220 , which are its fundamental orbital elements ( i.e. , those that the particle would have if there were no perturbers in the system ) , as well as the orientation parameters @xmath219 and @xmath221 .
since both the forced elements , and the osculating elements , of collisional fragments are the same as those of their parent ( apart from fragments with @xmath231 ) , particles from the same family have the same proper elements , @xmath218 and @xmath220 .
the evolution of a particle s proper elements is straight - forward they precess around circles of fixed radius , @xmath218 and @xmath220 , at a constant rate , @xmath198 , counterclockwise for @xmath232 , clockwise for @xmath233 .
the secular precession timescale depends only on the semimajor axis of the particle s orbit : @xmath234 where @xmath235 is given in years , and @xmath198 is given in equation ( [ eq : a ] ) ; secular perturbations produce long period variations in a particle s orbital elements ( e.g. , @xmath236(0.1 myrs ) in the asteroid belt ) .
the centers of the circles that the proper elements precess around are the forced elements ( see e.g. , fig .
[ fig2]a ) .
actually the forced elements vary on timescales that are comparable to the precession timescale ( eq.[[eq : tsec ] ] ) ; thus , it might appear ambitious to talk of the precession of a particle s osculating elements around circles when its real evolutionary track in the complex eccentricity and complex inclination planes may not be circular at all .
the reason it is presented as such is that at any given time , all of the particles at the same semimajor axis precess ( at the same rate ) around the same forced elements on circles of different radii , and this has consequences for the global distribution of orbital elements ( see [ sssec - opsfam ] ) .
there are two things that are worth mentioning now about a particle s forced elements . if there is just one perturber in the system , @xmath237 , its complex eccentricity and complex inclination do not undergo any secular evolution , and the forced elements imposed on a particle in the system are not only constant in time , but also independent of the mass of the perturber : @xmath238 e_j\ast\exp{i\tilde{\omega}_j } , \label{eq : zf1 } \\
y_f & = & i_j\ast\exp{i\omega_j}. \label{eq : yf1 } \end{aligned}\ ] ] this implies that a body of low mass , such as an asteroid , has as much impact on a particle s orbit as a body of high mass , such as a jupiter mass planet .
the perturbations from a smaller perturber , however , produce longer secular precession timescales ( eq . [ [ eq : tsec ] ] ) : @xmath239^{-1 } ; \label{eq : tsec2}\ ] ] they would also be similar in magnitude to those of the disk s self - gravity , which in that case could no longer be ignored . if there is more than one perturber in the system , @xmath240 , then particles on orbits for which their precession rate equals one of the system s eigenfrequencies ( @xmath241 , or @xmath242 ) have infinite forced elements imposed on their orbits , and so are quickly ejected from such a `` secular resonance '' region .
the solution given by equations ( [ eq : z ] ) and ( [ eq : y ] ) accounts for the fact that small particles see a less massive star due to the action of radiation pressure , but not for the p - r drag evolution of their orbits ; the solution for these particles is discussed in appendix [ app - pr ] . also ,
the perturbation theory of [ sssec - pe ] is only valid for particles with small eccentricities ; i.e. , it is not valid for the evolution of a disk s @xmath101 critical particles , or its @xmath101 meteoroids .
however , if the evolution of a disk s @xmath101 critical particles is affected by secular perturbations ( i.e. , if their lifetime is longer than the secular timescale ) , then it is probably also affected by p - r drag ( i.e. , their lifetime is probably also longer than the p - r drag timescale ) , in which case the disk s @xmath101 critical particles do not contribute much to its observable structure ( [ sssec - rf ] and [ sssec - pcsig ] ) .
there is no secular evolution to the orbits of @xmath101 meteoroids because of their short lifetimes .
the effect of secular perturbations on the structure of a disk can be understood by considering the effect of the secular evolution of the constituent particles orbits on the distribution of their orbital elements .
the perturbation equations of [ sssec - pe ] show that secular pertubations affect only the distribution of disk particles complex eccentricities , @xmath243 , and complex inclinations , @xmath244 , while having no effect on their size distribution ( and hence the division of the disk into its particle categories ) , or on their semimajor axis distribution ( and hence the disk s large - scale radial distribution ) .
consider the family of collisional fragments originating from a primordial body , the orbital elements of which were described by @xmath63 , @xmath218 , and @xmath220 .
here we consider only fragments that are unaffected by p - r drag ( those that are affected by p - r drag are discussed in appendix [ app - pr ] ) .
[ [ ssssec - lpfam ] ] large ( @xmath130 ) fragments + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the orbital elements of the largest fragments , those with @xmath245 , created in the break - up of the primordial body are initially very close to those of the primordial body ; they do not have identical orbits due to the velocity dispersion imparted to the fragments in the collision .
the forced elements imposed on the orbits of all of these collisional fragments are the same as those imposed on the primordial body .
the secular evolution of their osculating complex eccentricities ( eq . [ [ eq : z ] ] ) and complex inclinations ( eq . [ [ eq : y ] ] ) , is to precess about the forced elements ( which are also varying with time ) , but at slightly different rates ( due to their slightly different semimajor axes ) .
a similar argument applies for all particles created by the collisional break - up of these fragments .
thus , after a few precession timescales , the complex eccentricities and complex inclinations of the collisional fragments of this family lie evenly distributed around circles that are centered on @xmath246 and @xmath247 , and that have radii of @xmath218 and @xmath220 ( e.g. , their complex eccentricities lie on the circle shown in fig .
[ fig2]a ) , while their semimajor axes are all still close to @xmath63 .
this is seen to be the case in the asteroid belt : there are families of asteroids that have similar @xmath63 , @xmath218 , and @xmath220 , that are the collisional fragments resulting from the break - up of a much larger asteroid ( @xcite ) .
the distribution of the complex eccentricities , @xmath243 , of these particles , has a distribution of pericenters that is biased towards the orientation in the disk that is defined by @xmath248 .
the consequence of this biased orbital element distribution on the spatial distribution of this family material is best described with the help of fig .
this shows a face - on view ( i.e. , perpendicular to the plane of symmetry ) , of the family material in orbit around a star @xmath249 .
the resulting disk is made up of particles on orbits that have the same @xmath63 , @xmath250 , @xmath248 , and @xmath218 , but random @xmath251 .
the contribution of each particle to the spatial distribution of material in the disk can be described by the elliptical ring of material coincident with the particle s orbit ( see [ sssec - grav ] ) . in fig .
[ fig2]b , these elliptical rings have been represented by uniform circles of radius @xmath63 , with centers that are offset by @xmath252 in a direction opposite to the pericenter direction , @xmath67 ( this is a valid approximation to first order in the particles eccentricities ) ; a heavy line is used to highlight the orbital ring with a pericenter located at @xmath253 , and a displaced circle center located at @xmath8 , where @xmath254 .
the vector @xmath255 can be decomposed into its forced and proper components ; this is shown by the triangle @xmath256 , where @xmath257 , @xmath258 , and @xmath259 ( there is a similar triangle in fig .
[ fig2]a ) . given that the distribution of @xmath251 is random , it follows that the distribution of the rings centers , @xmath8 , for the family disk are distributed on a circle of radius @xmath260 and center @xmath261 .
thus , the family forms a uniform torus of inner radius @xmath262 and outer radius @xmath263 centered on a point @xmath261 displaced from the star @xmath249 by a distance @xmath264 in a direction away from the forced pericenter , @xmath248 ( @xcite ; @xcite ) .
the distribution of the complex inclinations , @xmath244 , of these particles , is also the distribution of their orbital planes .
changing the reference plane relative to which the particles orbital inclinations are defined to that described by @xmath223 , shows that the secular complex inclination distribution of this family material leads to a disk that is symmetrical about the @xmath223 plane ; the flaring of this disk is described by @xmath220 .
[ [ ssssec - spfam ] ] small ( @xmath231 ) fragments + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + since the small particles in this family originate from the larger particles , the orbital elements of the parents of the small particles have the same @xmath63 , @xmath218 and @xmath220 ( they also have the same @xmath222 and @xmath223 ) , but random @xmath251 and @xmath265 . consider the @xmath101 critical particles that are produced at the same time from the population of family particles that have the same @xmath251 and @xmath265 at that time ; this parent population is spread out along the orbit defined by the elements @xmath63 , @xmath64 , and @xmath67 , which could be one of the rings shown in fig .
the average pericenter orientation of these @xmath101 critical particles , @xmath266 , is the same as that of the orbit of the larger particles , @xmath67 ( obvious because of the symmetry of eq .
[ [ eq : wprime ] ] with respect to @xmath72 ) ; their pericenter locations are also the same ( eq . [ [ eq : rp ] ] ) .
thus , the ring shown in fig .
[ fig2]b defines the inner edge of the disk of @xmath101 critical particles created in the break - up of large particles on this ring ; i.e. , their disk is offset by an amount @xmath252 in the @xmath67 direction .
consequently , the inner edge of the disk of @xmath101 critical particles created in the break - up of all large particles in this family is offset by an amount @xmath264 in the @xmath248 direction .
this disk has the same plane of symmetry as the families large particles , since all particle categories from the same family have the same distribution of orbital planes , @xmath244 .
similar arguments apply for the families @xmath101 meteoroids . the disks of material from all of the families that have the same semimajor axis , or equivalently that are at the same distance from the star , have the same offset inner edge , and the same plane of symmetry .
this is because their large particles have the same forced elements imposed on their orbits .
the complex eccentricities and complex inclinations of the large particles of all of these families lie evenly distributed around circles with the same centers , @xmath222 and @xmath223 , but with a distribution of radii , @xmath267 and @xmath268 , that are the distributions of the proper elements of these families ( defining the width and flaring of the torus consisting of these families material ) .
the whole disk is made up of families with a range of semimajor axes ; the families that are at different semimajor axes can have different forced elements imposed on their orbits , and in view of the proper element distributions in the asteroid belt and in the kuiper belt , they can also have different distributions of proper elements . if the forced eccentricity imposed on the disk is non - zero , which it is if there is at least one perturber in the system that is on a non - circular orbit ( eqs .
[ [ eq : z ] ] and [ [ eq : zf1 ] ] ) , then the disk s center of symmetry is offset from the star . if the forced inclination imposed on the disk is different for families at different semimajor axes , which it is if there are two or more perturbers in the system that are moving on orbits that are not co - planar ( eqs .
[ [ eq : y ] ] and [ [ eq : yf1 ] ] ) , then the disk s plane of symmetry varies with distance from the star ; i.e. , the disk is warped .
there is also a physical explanation for the secular perturbation asymmetries .
the secular perturbations of a massive body are equivalent to the gravitational perturbations of the elliptical ring that contains the mass of the perturber spread out along its orbit , the line density of which varies inversely with the speed of the perturber in its orbit ( gauss averaging method , @xcite ; @xcite ) .
the ring s elliptical shape , as well as its higher line density at the perturber s apocenter , mean that the center of mass of the `` star - ring '' system is shifted from the star towards the perturber s apocenter .
the focus of the orbits of particles in such a system is offset from the star ; i.e. , the center of symmetry of a disk in this system is offset from the star .
the gravitational perturbations of the ring also point to the plane coincident with the perturber s orbital plane . in systems with two or more perturbers , the system s plane of symmetry ( that in which the perturbing forces out of this plane cancel ) varies with distance from the star ; i.e. , a disk in such a system is warped .
mid - ir geocentric satellite observations ( such as the iras , cobe , and iso observations ) are dominated by the thermal emission of the zodiacal cloud s p - r drag affected particles in all directions except that of the galactic plane ( @xcite ) .
such observations contain detailed information about the spatial structure of the zodiacal cloud , especially since their observing geometry changes throughout the year as the earth moves around its orbit .
since there are 9 massive perturbers in the solar system , the resulting secular perturbation asymmetries should be apparent in the iras , cobe , and iso data - sets ( see appendix [ app - pr ] for a discussion of why the distribution of a disk s p - r drag affected particles should also contain the signatures of the system s secular perturbations ) .
[ fig3]a shows cobe observations of the sum of the brightnesses in the 25 @xmath1 m waveband at the north and south ecliptic poles , @xmath269 ( @xcite ) , where there is no contamination from the galactic plane .
if the zodiacal cloud was rotationally symmetric with the sun at the center , then the cross - sectional area density of particles in the near earth region would vary according to @xmath270 , where @xmath271 is a constant . because the earth s orbit is eccentric
, geocentric observations sample the zodiacal cloud at different radial distances from the sun .
thus , the minimum of the @xmath269 observation is expected to occur either at the earth s aphelion , @xmath272 , or perihelion , @xmath273 , depending on whether @xmath274 or @xmath275 , which is determined by the collisional evolution of particles in the near - earth region ( e.g. , @xcite discuss the observational evidence and conclude that @xmath276 as found by the helios zodiacal light experiment )
. however , the minimum in the 25 @xmath1 m waveband observations occurs at @xmath277 , and a similar result is found in the 12 @xmath1 m waveband .
this is expected only if the sun is not at the center of symmetry of the zodiacal cloud .
parametric models of the zodiacal cloud have also shown the need for an offset to explain the observations ( e.g. , @xcite ) .
[ fig3]b shows the variation of the brightnesses of the ecliptic poles with ecliptic longitude of the earth ( @xcite ) .
the north and south polar brightnesses are equal when the earth is at either the ascending or descending node of the local ( at 1 au ) plane of symmetry of the cloud , giving an ascending node of @xmath278
. however , cobe observations of the latitudes of the peak brightnesses of the zodiacal cloud measured in the directions leading and trailing the earth s orbital motion give @xmath279 ( @xcite ) .
since such observations sample the cloud external to 1 au , this implies that the plane of symmetry of the zodiacal cloud varies with heliocentric distance , i.e. , that the zodiacal cloud is warped . to observe the zodiacal cloud s offset and warp asymmetries , an observer outside the solar system , would ,
at the very least , need an observational resolution greater than the magnitude of these asymmetries ; e.g. , to observe the offset asymmetry , the observer needs a resolution of @xmath280 arcseconds , where the distance from the observer to the sun , @xmath281 , is measured in pc .
an offset would be more observable in a disk such as hr 4796 due to its central cavity , since this causes a brightness asymmetry in the emission from the inner edge of the disk .
this section describes our model of the hr 4796 disk that accounts for the brightness distribution seen in the ihw18 waveband observation of t99 ( fig .
[ fig8]a ) .
there are three components of the observation that had to be included in the model ( [ sec - ocd ] ) : the disk s structure , @xmath55 , defined in [ ssec - dds ] ; the combination of the optical properties and the size distribution of the disk s particles given by @xmath49 , defined in [ ssec - los ] ; and the disk s orientation .
while the modeling techniques used here are new to the study of circumstellar disks , they have already been widely used to study the observed structure of the zodiacal cloud ( see , e.g. , @xcite ; @xcite ) , and so can be used with a certain degree of confidence .
the disk s structure was modeled as that of a secularly perturbed dynamic disk . a combination of the theory of [ sec - ddc ] and [ sec - spd ] , and inferences from the observation ( fig . [ fig8]a ) ,
was used to parameterize the distribution of the orbital elements of the disk s large particles ( [ sssec - doe ] ) .
this was used to create parameterized models of the spatial distribution of these large particles ( [ sssec - simul ] ) , that could then be compared with the observed spatial distribution .
since the observed spatial distribution is that of the `` emitting '' particles , the underlying assumption is that these emitting particles either are the disk s large particles , or that their spatial distribution is the same as that of the disk s large particles ( whether they are large or not ) .
the implications of this assumption are discussed in the interpretation of the model ( [ sec - disc ] ) .
the quintessentially secular part of the distribution of the orbital elements of the large particles in a secularly perturbed disk is the distributions of their complex eccentricities , @xmath243 , and complex inclinations , @xmath244 . for a particle with an orbit of a given semimajor axis , @xmath63 , its complex eccentricity ,
@xmath202 , and complex inclination , @xmath203 , are the addition of forced elements , @xmath283 and @xmath284 , to proper elements that have @xmath251 and @xmath265 chosen at random , while @xmath218 and @xmath220 are chosen from the distributions @xmath267 and @xmath268 .
since there is insufficient information available to determine the variation of the forced and proper elements with semimajor axis in this disk , they were assumed to be constant across the disk .
the forced elements were left as model variables : the forced eccentricity , @xmath250 , defines the magnitude of the offset asymmetry in the disk model ; the forced pericenter orientation , @xmath248 , defines the orientation of this asymmetry ; and the forced inclination , @xmath223 , defines the plane of symmetry of the disk model . when creating a disk model , both @xmath248 and @xmath223
were set to zero ; these were incorporated later into the description of the disk s orientation to our line of sight ( see [ ssec - do ] ) .
the distributions of the proper eccentricities , @xmath267 , and proper inclinations , @xmath268 , of particles in the disk model were taken to be like those of the main - belt asteroids with absolute magnitudes @xmath285 ( @xcite ) .
these large asteroids constitute a bias - free set ( @xcite ) and have mean proper eccentricities and proper inclinations of @xmath286 and @xmath287 .
not enough kuiper belt objects have been discovered yet to infer a bias - free distribution for their orbital elements .
the distribution of the semimajor axes , @xmath82 , of particles in the disk defines its radial distribution .
there is no way of guessing this distribution from theoretical considerations , since it depends on the outcome of the system s planetary formation process , which varies from system to system ( compare the distribution of the solar system s planets , and its disk material , with those found in exosolar systems , e.g. , @xcite ; @xcite ) .
thus , it had to be deduced purely observationally .
[ fig8]a shows that the disk has an inner edge , inside of which there is a negligible amount of dust ; this was modeled as a sharp cut - off in the distribution of semimajor axes at @xmath288 , a model variable .
the observation also shows that the disk has an outer edge at @xmath289 au ; this was modeled as a sharp cut - off in the distribution of semimajor axes at @xmath290 au ( this is a non - critical parameter , since particles near the outer edge of the disk contribute little to the observation , see [ sssec - pg ] ) .
the distribution between @xmath288 and 130 au was taken as @xmath291 , where @xmath83 is the number of particles on orbits with semimajor axes in the range @xmath84 , and @xmath292 is a model variable . to get an idea of the radial distribution resulting from this semimajor axis distribution , consider that if the particles had zero eccentricity , this distribution would result in a volume density ( number of particles per unit volume ) distribution that is @xmath293 ( since the number of particles in a spherical shell of width @xmath156 , the volume of which is @xmath157 , would contain a number of particles that is @xmath294 ) .
disk models were created from the orbital element distribution of [ sssec - doe ] using the `` simul '' program ; simul was developed by the solar system dynamics group at the university of florida ( @xcite ) .
a disk model is a large three - dimensional array , @xmath55 , that describes the spatial distribution of the cross - sectional area of material in the disk model per unit volume binned in : @xmath9 , the radial distance from the star ; @xmath33 , the longitude relative to an arbitrary direction ( set here as the forced pericenter direction , @xmath248 ) ; and @xmath34 , the latitude relative to an arbitrary plane ( set here as the forced inclination , @xmath223 , or symmetry , plane ) .
simul creates a disk model by taking the total cross - sectional area of material in the disk ( specified by the model variable @xmath295 ) , and dividing it equally among a large number of orbits ( 5 million in this case ) , the elements of each of which are chosen randomly from the specified distribution ( [ sssec - doe ] ) .
the disk model is populated by considering the contribution of each orbit to the cross - sectional area density in each of the cells it crosses .
the spatial distribution of material in one of our models of the hr 4796 disk can be described by the three variables @xmath288 , @xmath292 , and @xmath250 ; @xmath295 simply scales the amount of material in the model , and @xmath248 and @xmath223 describe the orientation of the disk to our line of sight .
[ fig4 ] is a plot of the surface density of material in a disk model with @xmath296 au , @xmath297 , and @xmath298 ( this is our final model of [ ssec - mpr ] ) .
this illustrates how the specified distribution of orbital elements affects the spatial distribution of material in the disk model : the sharp cut - off in semimajor axes at @xmath288 determines the radial location of the inner hole , which has a sloping cut - off in @xmath9 due to the particles eccentricities ; as predicted in
[ sssec - opsfam ] , particles at the inner edge of the disk in the forced pericenter direction are closer to the star than those in the forced apocenter direction by @xmath299 ; the distribution of semimajor axes has produced a surface density distribution that is @xmath300 , but only exterior to 70 au .
the optical properties of the disk particles were found assuming the particles to be made of astronomical silicate ( @xcite ; @xcite ) , a common component of interplanetary dust found in both the zodiacal cloud ( @xcite ) and exosolar systems ( e.g. , @xcite ; @xcite ; @xcite ) ; such an assumption can be tested at a later date using spectroscopy to look for silicate features in the hr 4796 disk emission .
furthermore , the particles were assumed to be solid , spherical , and have a density of @xmath301 .
their optical properties were calculated using mie theory , assuming that hr 4796a has a luminosity and temperature of @xmath302 , @xmath303k ( @xcite ) , and using for its spectrum , that of the a0v star vega ( @xcite ) . in all calculations ,
the mass of hr 4796a was assumed to be @xmath304 ( @xcite ) . the properties of particles of different sizes , and at different distances from hr 4796a ,
are shown in fig .
the temperatures of the particles are plotted in figs .
[ fig5]a and [ fig5]b .
the form of fig .
[ fig5]a can be understood by consideration of equation ( [ eq : tdr ] ) , and the wavelengths at which the star and a particle , if it was a black body , emit most of their energy : @xmath305 @xmath1 m , and @xmath306 @xmath1 m . as a crude approximation
, a particle with diameter @xmath8 has @xmath307 for @xmath308 and @xmath309 for @xmath310 .
thus , in terms of their thermal properties , disk particles can be divided into four categories : the largest particles , @xmath311 , are efficient absorbers and emitters at all relevant wavelengths and so achieve nearly black body temperatures , @xmath15 ; particles with @xmath312 , are inefficient emitters at their black body temperature , and so need temperatures higher than @xmath15 to re - radiate all of the incident energy ; the smallest particles , @xmath313 , are also inefficient absorbers at the stellar temperature , and so do not need as high temperatures as slightly larger particles to re - radiate the absorbed energy ; and particles with @xmath314 @xmath1 m have temperatures below that of a black body this is because these particles are super - efficient emitters at their black body temperatures ( due to silicate resonances , @xmath315 can go up as high as 2 ) , and so need lower temperatures to re - radiate the incident energy .
the form of fig .
[ fig5]b can be understood in the same way .
the fall - off of a large ( e.g. , @xmath316 @xmath1 m ) particle s temperature with distance from hr 4796a is like that of a black body , i.e. , @xmath317 , while the fall - off for smaller particles is not that steep because these particles emit less efficiently the further they are from the star ( due to their lower temperatures , and consequently higher @xmath318 ) ; e.g. , the fall - off for @xmath319 @xmath1 m particles is @xmath320 , which is close to the @xmath321 fall - off expected for particles with an emission efficiency that decreases @xmath322 ( e.g. , @xcite ) .
more important observationally is the variation of the particles @xmath323 $ ] , since this determines the contribution of a particle s thermal emission to the flux density received at the earth ( eqs .
[ [ eq : flux]],[[eq : fnu ] ] and [ [ eq : p ] ] ) .
this is plotted for @xmath324 @xmath1 m in figs .
[ fig5]c and [ fig5]d , and for @xmath325 @xmath1 m in figs . [ fig5]e and [ fig5]f . the form of figs .
[ fig5]c and [ fig5]e can be explained in the same way that fig .
[ fig5]a was explained : all three figures have similar forms , which is to be expected since the particles temperature also appears in @xmath26 ; figs .
[ fig5]c and [ fig5]e are , however , attenuated for @xmath326 , since these particles are inefficient emitters at that wavelength .
the fall - off with distance of the different particles shown in figs .
[ fig5]d and [ fig5]f is due solely to their different temperature fall - offs ( fig . [
fig5]b ) ; e.g. , for @xmath324 @xmath1 m , the approximate fall - off for @xmath316 @xmath1 m particles is @xmath327 , while that for @xmath319 @xmath1 m particles is @xmath328 .
in addition to the ihw18 waveband , t99 observed hr 4796 in the n ( @xmath333 @xmath1 m ) waveband .
the n band observations also show the double - lobed feature , but the inaccuracy of the subtraction of the image of hr 4796a from the observations means that they can not be used to constrain the disk s structure .
the n band observations can , however , be used to constrain the disk s brightness in this waveband ; the brightness of a disk in different wavebands differs only in the factor @xmath49 ( eq . [ [ eq : fnu ] ] ) , and so the n band observation can be used to obtain information about the disk s size distribution . since figs . [ fig5]c and [ fig5]e have similar forms , the two observations should be dominated by the emission of similarly sized particles ; i.e. , they should have the same @xmath53 , and show similar structures .
the same characteristic particle size , @xmath53 , was used to calculate the brightness of a disk model in both the ihw18 and n wavebands , using @xmath49 from the appropriate lines on the plots of figs .
[ fig5]d and [ fig5]f .
there is just one @xmath53 that simultaneously matches the disk s observed brightnesses in both wavebands .
we made an initial estimate for @xmath53 based on the flux densities observed in the two wavebands : the flux densities of the disk in the ihw18 and n wavebands are 857 and 40 mjy , respectively ( t99 ) ; i.e. , the observed flux density ratio ( n / ihw18 ) is @xmath334 .
the expected flux density ratio of the two wavebands ( the ratio of @xmath49 for different @xmath53 ) is plotted in fig .
[ fig6]a . assuming the emission to arise mostly from particles near the inner edge of the disk , @xmath335 60 - 80 au , fig .
[ fig6]a shows that the observed emission can be fairly well - constrained to come from particles with @xmath336 @xmath1 m .
[ fig7]a shows a contour plot of an unsmoothed ihw18 waveband observation of the disk model of fig .
[ fig4 ] viewed face - on ( i.e. , perpendicular to the disk s plane of symmetry , @xmath223 ) .
this shows the observational consequence of the offset center of symmetry of the disk model . because the particles at the inner edge of the disk ( those that contribute most to the disk s brightness ) are closer to the star in the forced pericenter direction , @xmath248 , than those in the forced apocenter direction ( fig .
[ fig4 ] ) , they are hotter and so contribute more to the disk s thermal emission than those at the forced apocenter .
this is the `` pericenter glow '' phenomenon , which leads to the horseshoe - shaped highest contour line ( the filled - in 1.02 mjy / pixel line ) , which is pointed in the @xmath248 direction .
this asymmetry is a consequence of @xmath288 and @xmath250 only , and its magnitude is determined by @xmath250 only .
in particular , if there is a gradient of @xmath250 across the disk , then it is @xmath250 at the inner edge of the disk that controls the magnitude of the asymmetry .
the outermost contour plotted on fig .
[ fig7]a , which is an offset circle with a radius of 95 au , is that corresponding to 0.17 mjy / pixel .
thus , there is little emission from the outer edge of the disk , justifying the arbitrary use of @xmath290 au in the modeling .
the two variables that define the orientation of the hr 4796 disk to our line of sight are @xmath248 and @xmath337 ; how they define this orientation is best explained using fig .
imagine that the disk starts face - on with the forced pericenter direction pointing to the left .
it is then rotated clockwise by @xmath248 ( this is shown in fig .
[ fig7]a , where @xmath338 ) , and then tilted by @xmath339 about the dotted line on fig .
the direction of this tilt , whether the top or bottom of the disk ends up closer to the observer , is not constrained in the modeling , since no account was made for either the extinction of the disk s emission by the disk itself , or for the disk s scattered light ( e.g. , an observer would see forward - scattered starlight from the closest part of the disk and back - scattered starlight from the farthest part , a phenomenon that could produce an apparent asymmetry in a symmetric disk , @xcite ) .
if the resulting inclination of the disk s symmetry plane to our line of sight , @xmath337 , is small , then the resulting nearly edge - on observation shows two lobes , one either side of the star . since the hotter , brighter , pericenter glow material is predominantly in one of the lobes ( unless @xmath340 ) , the lobes have asymmetric brightnesses .
pseudo - observations of disk models in the ihw18 and n wavebands were produced that mimicked the real oscir ( the university of florida mid - ir imager ) observations in both pixel size , at 67 pc ( @xcite ) , and smoothing , using the observed psfs ( which are asymmetric and slightly fatter than diffraction limited , t99 ) , and including the post - observational gaussian smoothing of fwhm = 3 pixels . the model variables : @xmath288 , @xmath292 , @xmath337 , @xmath250 , @xmath248 , and @xmath295 ,
were optimized so that the modeled ihw18 observation correctly predicts the observed ihw18 brightness distribution ; at the same time , the variable , @xmath53 , was optimized so that the modeled n band observation correctly predicts the observed n band brightness .
the model observations were compared with the real observations using the following diagnostics : the lobe brightnesses , @xmath341 , and their projected radial offsets from hr 4796a , @xmath342 , that were found by fitting a quintic polynomial surface to a @xmath343 pixel region around each of the lobes with 0.1 pixel resolution ( the location of hr 4796a in the real observations was also found in this way ) ; and line - cuts through the disk both parallel ( e.g. , figs .
[ fig8]d and [ fig9 ] ) , and perpendicular ( e.g. , figs .
[ fig8]c and [ fig8]e ) to the line joining the two lobes in the ihw18 observations .
an understanding of how the different model variables affect the different diagnostics allowed the modeling process to be decoupled into solving for : the disk s symmetrical structure , defined by @xmath337 , @xmath288 , and @xmath292 ; the particle size @xmath53 ; and the disk s asymmetrical structure , defined by @xmath250 and @xmath248 . throughout the modeling ,
the amount of material in a model , @xmath295 , was scaled so that the model observation predicted the correct observed mean brightness of the lobes in the ihw18 waveband , @xmath344 mjy / pixel ; its final value , @xmath345 m@xmath104 , was calculated once the other variables had been constrained .
since @xmath250 and @xmath248 only pertain to the disk s asymmetrical structure , then for a given @xmath53 , the variables pertaining to the disk s symmetrical structure , @xmath337 , @xmath288 , and @xmath292 , could be solved using a model with @xmath346 ; such a model is axisymmetric , and so the variable @xmath248 is redundant .
the inclination of the disk s plane of symmetry to the line of sight , @xmath347 , was constrained to give the best fit to the line - cuts perpendicular to the lobes ( figs .
[ fig8]c and [ fig8]e ) ; this is in agreement with that found by previous models of hr 4796 disk observations ( @xcite ; @xcite ) .
since our model is that of a `` fat '' disk , a different proper inclination distribution would lead to a different @xmath337 ; e.g. , if the disk is actually thinner than modeled here , @xmath348 , then the inferred @xmath337 is an underestimate , and vice versa .
the inner edge of the disk , , was constrained such that the model observation reproduces the observed mean radial offset of the lobes from hr 4796a , .
the semimajor axis distribution , @xmath349 , was constrained to give the best fit to the cut along the line joining the two lobes ( fig .
[ fig8]d ) . since
@xmath53 was already estimated to be about 2 - 3 @xmath1 m ( [ sssec - dtyp ] ) , the modeling of the disk s symmetrical structure was repeated for @xmath350 2 , 2.5 , and 3 @xmath1 m .
adjusting @xmath53 by such a small amount did not affect the inferred symmetrical structure parameters .
this was expected , since the @xmath253(18.2 @xmath1m@xmath351 for each of these @xmath53 are very similar ( fig .
[ fig5]d ) . remembering that the model is always normalized to predict the observed mean ihw18 lobe brightnesses , the predicted n
band lobe brightnesses are compared for the three values of @xmath53 in fig .
this shows that the particle size can be constrained to be @xmath352 @xmath1 m ; i.e. , the crude method of calculating the particle size of [ sssec - dtyp ] gives a very good estimate of this size .
this particle size means that the total mass of emitting particles in the disk model is @xmath353 , where @xmath354 is the mass of the earth .
however , this is not a useful constraint on the disk s mass , since the disk s mass is expected to be concentrated in its largest particles .
the best estimate of the mass of the hr 4796 disk , from submillimeter observations , is that it is between @xmath355 and @xmath356 ( @xcite ; @xcite ) , which , as expected , is well above the mass of our disk model .
the disk s observed asymmetries are defined by : the lobe brightness asymmetry , @xmath357 , and the radial offset asymmetry , @xmath358 .
these asymmetries are also apparent in the disk model , and their magnitudes are determined by both @xmath250 and @xmath248 .
the lobe brightness asymmetry was used to constrain @xmath250 and @xmath248 , and it was found that the @xmath250 necessary to cause the @xmath359 asymmetry depends on the geometry of the observation according to the relation shown in fig .
thus , for the majority of the geometries , a forced eccentricity of between 0.02 and 0.03 is sufficient to cause the observed brightness asymmetry . in the context of this modeling
, the observed brightness asymmetry implies a radial offset asymmetry of @xmath360 , which is within the limits of the observation . our final model shown in figs .
[ fig4 ] and [ fig7]-[fig9 ] assumes a modest value of @xmath297 , which corresponds to an @xmath248 orientation of @xmath361 ( fig .
[ fig7]b ) . whether this rotation puts the pericenter glow material above or below the horizontal , @xmath362 or @xmath363 , is not constrained here , since it has a minimal effect on the observation : in the model observation of fig . [ fig8 ] , the top of the disk is brighter than the bottom of the disk by a fraction that would be undetectable in the observation due to noise and the disk s unknown residual structure .
the line - cuts of fig .
[ fig8 ] show how well the model fits all aspects of the observation the vertical structure , the horizontal structure , and the lobe location and asymmetry .
the standard deviations of the oscir lobe observations quoted in this section were found using model observations that mimicked the noise present in the oscir observations .
the background sky noise in the ihw18 observation was found to be approximately gaussian with zero mean and a @xmath364 noise per pixel of @xmath365 mjy ( t99 ) ; this was included after the psf smoothing , but before the post - observational smoothing .
observations of the model were repeated for 50,000 different noise fields to obtain the quoted standard deviations . since the observed psf was asymmetric ( t99 ) , this introduces an apparent lobe asymmetry of @xmath366 in an observation of a symmetric disk ( one with @xmath367 ) , and so the observed lobe asymmetry is @xmath368 from the mean , and its statistical significance is @xmath369 . while this is small
, it does show that the pericenter glow phenomenon is observable with current technology : hr 4796 was observed with the infrared imager oscir for only one hour on keck ii ( t99 ) ; in the background - limited regime , the significance level of any asymmetry increases at a rate @xmath370 ; thus , one good night on a 10 meter telescope should be enough to get a definitive observation of the hr 4796 lobe asymmetry .
the real significance of the hr 4796 asymmetry may also be higher than quoted above , since it also seems to be apparent in other observations of this disk ( @xcite ; @xcite ) .
even if subsequent observations happen to disprove the existence of the asymmetry , this is still significant , since , as we show in [ ssec - ias ] , an asymmetry is to be expected if the companion star , hr 4796b , is on an eccentric orbit .
the interpretation of our hr 4796 disk model is broken down into sections that cover discussions of : the dynamics of the disk particles , [ ssec - ddch ] ; the lobe asymmetry , [ ssec - ias ] ; the emitting particle category , [ ssec - deps ] ; the origin of the inner hole , [ ssec - lsd ] ; and the residual structure of the observation once the model has been subtracted , [ ssec - rs ] .
our interpretation of the observed structure of the hr 4796 disk starts with a discussion of the dynamics of the particles in the disk , and where the emitting particles fit into our understanding of the dynamic disk ( [ sec - ddc ] ) .
the radiation forces , defined by @xmath101 ( eq . [
[ eq : beta ] ] ) , acting on particles in the hr 4796 disk can be found from their optical properties .
[ fig6]b shows the @xmath101 of particles with the optical properties assumed in our model ( [ sssec - pop ] ) .
thus , particles in the disk with @xmath371 @xmath1 m are @xmath101 meteoroids ( even the submicron particles have @xmath372 ) , and a good approximation for the non-@xmath101 meteoroids is : @xmath373 where @xmath8 is measured in @xmath1 m ( in agreement with eq . [
[ eq : beta2 ] ] ) .
[ fig9 ] shows that the emitting particles have @xmath374 @xmath1 m ; these particles have @xmath375 .
so , the modeling implies that the emitting particles are @xmath101 meteoroids , i.e. , that they are blown out of the system on hyperbolic orbits .
since the lifetime of @xmath101 meteoroids is @xmath332(370 years ) for those created at @xmath376 au , which is much shorter than the age of the hr 4796 system , any @xmath101 meteoroids that are currently in the disk can not be primordial particles , rather they must be continuously created from a reservoir of larger particles ; i.e. , the existence of @xmath101 meteoroids implies the existence of a dynamically stable population of larger particles .
if the emitting particles are on hyperbolic orbits , we can use the mass of the disk s emitting particles , @xmath377 , and the emitting lifetime of these particles , @xmath378 years , to estimate the mass loss rate of the disk to be @xmath379year .
if this mass loss rate has been sustained over the age of the system , the original disk must have been @xmath380 more massive than it is today .
in fact , a more massive disk would have had a higher mass loss rate , since this rate increases proportionally with the total cross - sectional area in the disk ( i.e. , @xmath381 ) .
this means that if the current disk has a mass @xmath356 , the original disk must have had a mass @xmath382 ; i.e. , the hr 4796 disk may provide evidence for the type of collisional mass loss that may have happened in the early kuiper belt ( @xcite ) .
the collisional lifetime of the disk s emitting particles can be calculated directly from the disk model using equations ( [ eq : tcol3 ] ) and ( [ eq : fccdtyp ] ) : @xmath383 years across most of the disk ( 55 - 85 au ) , with a minimum at @xmath376 au of @xmath384 years .
this collisional lifetime is much less than the age of the hr 4796 system .
thus , the emitting particles can not be primordial particles ( irrespective of whether they are @xmath101 meteoroids ) .
the collisional lifetime of the emitting particles can be corroborated using equation ( [ eq : tcol6 ] ) .
the model s effective optical depth ( eq . [ [ eq : taueffdefn ] ] ) is its surface density , plotted in fig .
[ fig4 ] , multiplied by the cross - sectional area of a particle in the model , @xmath385 .
this peaks at 70 au , where @xmath386(70 au ) = @xmath387 , and @xmath388(370 years ) , giving a collisional lifetime that has a minimum of @xmath389 years .
the disk s effective optical depth at 70 au , @xmath386(70 au ) , can also be corroborated directly from the observation using equation ( [ eq : taueff ] ) .
the observed edge - on , smoothed lobe brightness is @xmath390 mjy / pixel ; this can be scaled to the unsmoothed face - on brightness by the factor 1.10/1.40 ( see figs . [ fig7]a and [ fig8]b ) , which , since each pixel subtends @xmath391 sr , gives a brightness of @xmath392(18.2 @xmath1m,70 au)@xmath393 jy / sr . for 2.5 @xmath1 m particles , figs .
[ fig5]c and [ fig5]d give @xmath253(18.2 @xmath1m,70 au ) @xmath394 jy / sr , thus confirming that @xmath386(70 au ) @xmath395 ( this is also in agreement with @xmath396 found by @xcite ) .
in fact , assuming that the disk s total ihw18 flux density , 857 mjy ( t99 ) , comes from particles between @xmath397 au , the unsmoothed face - on brightness of the disk can also be corroborated ; equation ( [ eq : fnuobs ] ) with @xmath398 pc gives @xmath392(18.2
@xmath1 m , 70 au)@xmath399 jy / sr ) . since particles are only broken up by collisions with particles that have diameters more than a tenth of their own ( eq . [ [ eq : dimp ] ] ) , the collisional lifetime of the disk s large particles must be longer than that of the smaller emitting particles . assuming the cross - sectional area distribution to follow equation ( [ eq : nd ] ) with @xmath161 down to particles of size @xmath53 , the collisional lifetime of particles with @xmath400 can be estimated as ( eq . [ [ eq : tcol7 ] ] ) : @xmath401 where @xmath173 is measured in years , and d in km .
since particles for which @xmath402 can not be primordial , this implies that particles currently in the hr 4796 disk that are smaller than 60 m can not be original , rather they must have been created in the break - up of larger particles ; i.e. , these disk particles form a collisional cascade , through which the spatial distributions of the smaller particles are related to that of the larger particles .
the disk s particles that are larger than 60 m are its primordial particles .
the disk s high effective optical depth means that @xmath403 at 70 au ( eq . [ [ eq : btr ] ] ) , which in turn means that there is not expected to be significant p - r drag evolution for any of the disk s particles .
analysis of the p - r drag evolution of disk particles ( eq . [ [ eq : tpr ] ] ) shows that even the most affected particles , those with @xmath404 , have a p - r drag lifetime of @xmath405(1.6 myr ) at 70 au , and that these particles would only make it to @xmath406 au before they are broken up by collisions and blown out of the system by radiation pressure . appendix [ app - pr ] , which gives a discussion of lobe asymmetries in disks for which p - r drag is important , gives an analysis that shows that the cumulative effect of p - r drag on the orbits of all of a disk particle s ancestors is also insignificant .
since p - r drag is not an important process in the hr 4796 disk s evolution , there are just three categories of particles in the disk : large particles , @xmath101 critical particles , and @xmath101 meteoroids . if we are to believe the modeled emitting particle size ( discussed further in [ ssec - deps ] ) , the particles that are seen in both the ihw18 and the n band observations are the disk s @xmath101 meteoroids .
since we modeled the disk as if the emitting particles are large particles , this inconsistency needs to be borne in mind in our interpretation of the model .
while the modeling used a distribution of orbital elements that is only appropriate for the disk s large particles , it was the spatial distribution of the disk s emitting material , @xmath55 , that was constrained by the modeling , not the distribution of orbital elements , @xmath282 .
therefore , the inferred distribution , @xmath55 , which is that shown in fig . [ fig4 ] , is an accurate description of the spatial distribution of the disk s emitting particles , whatever their size .
indeed , it is in excellent agreement with that inferred from other observations of the hr 4796 disk ( @xcite ; @xcite ; @xcite ) .
if the emitting particles are the disk s large particles , then the inferred model variables have physical interpretations for the distribution of the orbits of the disk s large particles . if , as appears to be the case , the emitting particles are @xmath101 meteoroids , then further modeling of @xmath55 needs to be done to infer the distribution of the orbits of these particles .
however , since the disk s @xmath101 critical and @xmath101 meteoroid particles are created from its large particles , the spatial distributions of all of these particles share a great deal in common ( see [ ssec - ddpc ] and [ ssec - ow ] ) : they all have the same plane of symmetry , the same flaring , and the same offset and warp asymmetries , but the radial distributions of the smaller particles are more extended than that of the large particles .
this means that if we are seeing the disk s @xmath101 meteoroids , then the spatial distribution of its large particles has a plane of symmetry that is defined by @xmath337 , and an inner edge that is at the same radial location , and that is offset by the same amount and in the same direction , as that shown in fig .
[ fig4 ] ; their radial distribution , however , would not be as extended as that of fig .
thus , the model parameters @xmath288 , @xmath250 , and @xmath248 have physical interpretations for the distribution of the orbits of the disk s large particles , irrespective of the size of the emitting particles . the lobe asymmetry in the model of [ sec - hr4796 ] is due solely to the offset inner edge of the disk .
the model shows that secular perturbations amounting to a comparatively small forced eccentricity , @xmath297 , imposed on the orbits of large particles at the inner edge of the disk , @xmath407 au , would cause the disk s inner edge to be offset by a sufficient amount to cause the observed 5% lobe asymmetry .
this section considers what kind of a perturber system would impose such a forced eccentricity on the inner edge of the disk , and whether such a system is physically realistic .
a discussion of the system s secular perturbations also allows interpretation of the disk s inferred orientation , defined by the parameters @xmath248 and @xmath337 .
if hr 4796a s binary companion , hr 4796b , is on an eccentric orbit , it would have imposed a forced eccentricity on the disk particles . however , a forced eccentricity could also have been imposed on the disk by an unseen planet close to the inner edge of the disk , a planet which could be responsible for clearing the inner region ( e.g. , @xcite ) .
the secular perturbations imposed on the hr 4796 disk by a perturber system that includes hr 4796b and a putative planet located at the inner edge of the disk are shown in fig .
[ fig10 ] for the four cases : @xmath408 , @xmath409 , @xmath410 , and just planet .
the parameters of the two perturbers are assumed to be : * hr 4796b * @xmath411(@xcite ) ; the orbit of hr 4796b is unknown at present ( @xcite ) , so the semimajor axis of its orbit is arbitrarily taken as its projected distance , @xmath412 au ( @xcite ; note that we are not assuming that this is the semimajor axis of the ellipse that the star s orbit traces on the sky ) this gives an orbital period of @xmath413 years ( eq . [ [ eq : tper ] ] ) , and a timescale for secular perturbations from hr 4796b to have built up at 62 au of @xmath332(1 myr ) ( eq . [ [ eq : tsec2 ] ] ) ; @xmath414 , the eccentricity necessary to cause @xmath297 at @xmath407 au if there were no unseen perturbers ( eq . [ [ eq : zf1 ] ] ) ; @xmath415 , defining the reference plane for the analysis .
* planet * @xmath416 is a variable measured in jupiter masses , where @xmath417 ( current observations have limited the size of a planet in the system to @xmath418 , @xcite ) ; @xmath419 au ( see [ ssec - lsd ] ) ; @xmath420 , the eccentricity necessary to cause @xmath297 at @xmath407 au if the planet was the only perturber ( eq .
[ [ eq : zf1 ] ] ) ; @xmath421 , an arbitrary choice that represents the fact that the orbital plane of the planet is not necessarily be aligned with that of hr 4796b . for the cases when
there is just one perturber in the system , the forced elements in the system can be found from equations ( [ eq : zf1 ] ) and ( [ eq : yf1 ] ) : the forced eccentricity , @xmath422 , is determined by the ratio of the semimajor axes of the perturber and the particle , and by the eccentricity of the perturber s orbit , but is independent of the perturber s mass ; the forced pericenter , @xmath248 , is aligned with the pericenter of the perturber ; and the plane of symmetry of the disk , @xmath223 , is constant across the disk , and is the orbital plane of the perturber .
so , if the only perturber is hr 4796b , then to impose @xmath297 at @xmath407 au , the eccentricity of its orbit would have to be @xmath414 ; the consequent forced eccentricity imposed on the disk is plotted in fig .
[ fig10]a .
this also means that if @xmath423 , then a brightness asymmetry in this disk of @xmath424 would be expected unless adverse geometrical conditions prevented it .
the position angle from north of hr 4796b relative to hr 4796a is @xmath425 ( @xcite ) , while that of the sw lobe ( i.e. , the least bright lobe ) is @xmath426 ( t99 ) . for the lobe asymmetry to be the consequence of perturbations from hr 4796b
only , hr 4796b must currently be close to its apastron , and its orbital plane must be the plane of symmetry of the disk , i.e. , inclined at @xmath427 to the line of sight .
all of these conclusions are consistent with our initial estimate that the semimajor axis of the star s orbit is equal to its observed projected distance , 517 au .
consider the effect of adding a planet at the inner edge of the disk into the hr 4796 system .
if there are two perturbers in the system , then the forced element variation with semimajor axis depends both on the masses of the perturbers and on the orientations of their orbits . in the plot of fig .
[ fig10]b , @xmath428 was chosen so that the forced eccentricity ( and hence the lobe asymmetry ) is aligned with the planet s
pericenter for @xmath429 , and aligned with hr 4796b s pericenter for @xmath430 , where @xmath431 is the semimajor axis for which @xmath346 . since the two perturbers were also chosen to have different orbital planes , a similar change in the disk s alignment is seen in the plot of @xmath432 ( fig .
[ fig10]d ) : the disk s plane of symmetry is aligned with the planet s orbital plane at its inner edge , and with the orbital plane of hr 4796b at its outer edge ; this could cause an image of the disk to appear warped .
such a warp could be modeled using the same modeling techniques that are described in this paper , and would provide further constraints on the perturbers in the system ( even if no warp was observed ) .
so , it is possible that the brightness asymmetry , and the symmetry plane of the lobes , are determined by a planet close to the edge of the disk that has @xmath433 , rather than by hr 4796b .
using such an analysis , the pericenter glow phenomenon could be used to test for the existence of a planet in the hr 4796 system , but only after the orbit of hr 4796b has been determined ; e.g. , if @xmath434 , @xmath435 or the plane of hr 4796b s orbit contradicted the observed asymmetry orientation , brightness asymmetry magnitude , or the plane of symmetry of the lobes , then the existence of a planet at the inner edge of the disk with @xmath436 could be inferred . a double - lobed disk structure could also be observed in a system with no observable companion .
the only possible perturber in such a system is an unseen planet , the secular perturbations of which warrant the same kind of discussion as for the case when hr 4796b was the only perturber ( [ sssec - jb ] ) . depending on the planet s mass , radial location and eccentricity , it too could give rise to a detectable pericenter glow . the only constraint on the planet s mass is that the disk must be old enough for its secular perturbations to have affected the distribution of orbital elements of the disk particles over the age of the system .
since it takes of the order of one precession timescale to distribute the complex eccentricities of collisional fragments around the circle centered on the forced eccentricity , the constraint on the planet s mass can be approximated as that for which is that the age of the system , @xmath437 , is greater than the secular timescale , @xmath438 ( eq . [
[ eq : tsec2 ] ] ) ; for the hr 4796 system this limit is @xmath439 .
the constraint on the planet s eccentricity is even less stringent than for the binary companion because the planet is closer to the edge of the disk : a planet in the hr 4796 system would only need an eccentricity of @xmath440 to produce the observed @xmath441 lobe asymmetry , and the forced eccentricity imposed on the disk by a planet with @xmath420 is plotted in fig .
[ fig10]c .
so , the signature of even a low - mass planet would not escape detection and symmetrical double - lobed features are unlikely to be observed in systems that contain planets .
if the disk itself is massive enough to cause significant gravitational perturbations to the orbits of the disk particles , then the mass of the disk should be incorporated into the analysis of the secular perturbations in the system .
a massive disk could dampen the eccentricity of a planet at the inner edge of the disk ( @xcite ) , thus reducing the offset asymmetry .
the fomalhaut disk lobes may have asymmetric brightnesses ( @xcite ) , but the statistical significance of this asymmetry is low .
fomalhaut is a wide visual binary system .
gliese 879 is fomalhaut s common proper motion companion ( @xcite ) ; the two stars are separated by @xmath442 , which corresponds to a projected separation of @xmath332(55,000 au ) at @xmath443 pc . at such a distance ,
the forced eccentricity imposed on the disk by the binary star is insignificant ( eq .
[ [ eq : zf1 ] ] ) . a secular perturbation offset asymmetry in this disk
would be expected only if there is a planet in the disk that has a non - circular orbit .
the emitting lifetime of the disk s @xmath101 meteoroids , @xmath332(370 years ) , is less than the emitting lifetime of their parents , @xmath444 years ) ; equivalently , @xmath445 .
thus , our understanding of the dynamic disk implies that the disk s cross - sectional area distribution should not contain a significant amount of @xmath101 meteoroids .
rather , since there are no p - r drag affected particles in the disk , the disk s emission is expected to come from its @xmath101 critical particles , and its smallest large particles ( [ sssec - pcsig ] ) .
not all disk particles have the same composition and morphology ; even if this were a close approximation , there is , as yet , no evidence to suggest whether the particle properties chosen in our modeling ( [ sssec - pop ] ) are correct .
are we indeed seeing the disk s @xmath101 meteoroid particles , or did the assumptions of the modeling lead us to this conclusion ? consider the initial crude estimate of the particle size ( [ sssec - dtyp ] ) ; this proved to be an accurate method for estimating the particle size ( [ sssec - tps ] ) . if different assumptions had been made about the particles properties ( e.g. , if the particles had been assumed to be made of ice ) or morphologies ( e.g. , if the particles had been assumed to be like the `` bird s nest '' structures of @xcite ) , both figs
. [ fig6]a and [ fig6]b would be different , and different conclusions might have been drawn about the @xmath101 of the emitting particles .
if a size distribution had been included in the modeling , this would also have affected our conclusions .
these are considerations that should be modeled before any firm conclusions about the dynamics of the emitting particles can be reached . however , since irrespective of their assumed properties large particles have black body temperatures and brightness ratios similar to those of the @xmath446 @xmath1 m particles in fig .
[ fig6]b , a flexible interpretation of the observed brightness ratio is that the emitting particles must have temperatures that are hotter than black body .
thus , the emitting particles are either small ( e.g. , the simple analysis of [ sssec - pop ] implies that @xmath447 @xmath1 m ) , in which case they are likely to be @xmath101 meteoroids ( e.g. , 10 @xmath1 m particles have @xmath133 unless they have densities @xmath448 kg / m@xmath113 , eq .
[ [ eq : beta2 ] ] ) , or they are large particles that are made up of smaller hotter particles . the assumptions about the particle properties in the model could also have affected our conclusions about the collisional lifetime of the emitting particles .
consider the estimate of @xmath386(70 au ) derived from the ihw18 lobe brightness ( [ sssec - collh ] ) .
changing the properties of particles in the model would change the estimate of @xmath386 because of the resultant changes in @xmath449 @xmath1m@xmath351 ; e.g. , if we had modeled the disk using 30 - 50 @xmath1 m particles , we would have had to put more cross - sectional area in the model for it to give the observed lobe brightness .
[ fig5]c shows that for astronomical silicate mie spheres that are larger than 0.01 @xmath1 m in diameter , @xmath449 @xmath1m,70 au ) must lie in the range @xmath450 jy / sr , depending on whether the disk s cross - sectional area is concentrated in its 30 - 50 @xmath1 m particles that emit at cool temperatures , or in its 2 - 3 @xmath1 m particles that are small and hot , but large enough to emit efficiently at 18.2 @xmath1 m . for particles with different optical properties , @xmath449 @xmath1m,70
au ) could be below @xmath451 jy / sr if the particles have temperatures well below black body ( or if they have low emission efficiencies ) ; equally , it could be higher than @xmath452 jy / sr if the particles are hotter than the 2 - 3 @xmath1 m astronomical silicate mie spheres . taking @xmath451 jy / sr as a lower limit for @xmath449 @xmath1m,70 au )
implies that @xmath386(70 au ) @xmath453 , giving a collisional lifetime for the emitting particles that could be as low as 85 years
. however , given the temperature of the emitting particles inferred from the observed brightness ratio , @xmath444 years ) remains the best estimate for their collisional lifetime . a further note of caution is necessary about the inferred collisional lifetime : appendix [ app - tcoll ] assumes collisions between disk particles to be either catastrophic or irrelevant . while this may be appropriate for the disk s larger particles ,
since these are likely to be solid bodies , collisions between its smaller particles , which may have fluffy `` bird s nest '' structures ( @xcite ) , could be more erosive than destructive , and could lead to significant grain growth . in conclusion , neither observational , nor theoretical considerations
can provide a definitive answer as to the dynamics of the emitting particles .
however , the confirmation of the emitting particles collisional lifetimes means that we can be sure that these particles are not primordial , and that there are no p - r drag affected particles in the disk .
@xmath101 meteoroids remain the most likely candidate for the emitting particles .
mid - ir emitting particles that are on hyperbolic orbits have been inferred from observations of the disks around both @xmath101 pictoris ( @xcite ) and hd 141569 ( @xcite ) .
whatever the size of the emitting particles , analysis of the optical depth of the disk s inner region ( t99 ) shows that it is a few hundred times less than that of the outer disk , and so there are very few emitting particles in this region . because this central hole is necessary for the secular perturbation offset asymmetry to be observed ( without the hole only the radial offset
could be observed ) , its physical origin requires attention .
since the existence of small emitting particles in the disk requires the existence of large particles , the question to answer is why there are so many large particles in the outer disk , but so few in the inner disk ? either the physical conditions were such that they were able to form in the outer region , but not in the inner region , or they formed across the whole disk , but those formed in the inner region have since been removed . rather than discussing the planetary formation process and the stage of the system s evolution ( although these are of utmost importance in determining the physics of the disk ) ,
this section offers possible dynamical explanations for the removal of the large particles .
if a planet ( or planets ) exists interior to the inner edge of the disk , then resonance overlap removes all material from a region of radial width @xmath454 about the planet s orbit within about 1000 orbital periods ( @xcite ; @xcite ) .
material is also removed from the secular resonance regions ( @xcite ) the origin of these regions , which cover the range of semimajor axes on the plots of figs .
[ fig10]b and [ fig10]d where @xmath455 , was discussed in [ sssec - lp ] . the radial distribution of material
would also be affected by planetary radial migration ( @xcite ; @xcite ) .
however , such mechanisms can only explain the total lack of large particles in the inner region by invoking a system with either many planets , or just one planet that is either very large , or on a very eccentric orbit .
since all of these mechanisms take longer than a particle s orbital period to take effect , they would cause an inner cut - off in the disk particles semimajor axes ( as opposed to a cut - off in radial distance from the star ) . to estimate the orbit of a putative planet at the inner edge of the hr 4796 disk that is causing the cut - off , consider the inner edge of the kuiper belt .
there is almost no kuiper belt material on orbits with semimajor axes interior to that of pluto ( @xcite ) , which is in * 2:3 * resonance with neptune .
this is supposed to be the result of resonance sweeping that occurred as neptune s orbit expanded early in the history of the solar system due to the clearing of planetesimal debris from the inner solar system and the formation of the oort cloud ( @xcite ) . by analogy , assuming that the inner cut - off of the disk s large particles occurs at the planet s * 2:3 * resonance location , and that this cut - off can be described by @xmath296 au , we can estimate the orbit of the planet to have a semimajor axis of @xmath456^{2/3 } = 47 $ ] au , giving an orbital period of @xmath457 years ( eq . [ [ eq : tper ] ] ) .
so far , no explanation has been offered for the structure of the residuals ( what is left after subtracting the model from the observation , see t99 ) .
analogy with the zodiacal cloud implies that there could be a population of warmer dust in the inner region that may be unrelated to the dust in the outer disk . depending on the perturbers in the inner region ,
such dust could contain considerable structure .
analysis of emission from such regions would reveal a great deal about the system s perturbers , however , this would not be easy , since the resolution required to map such small - scale structure is at the limit of current technological capabilities .
in addition , such emission is masked by that of both the stellar photosphere and the outer disk , the accurate subtractions of which are difficult .
there may also be residual structure associated with the outer regions of the disk .
if there is a planet orbiting hr 4796a close to the inner edge of the disk , then the distribution of large particles in the outer disk would contain structure associated with the planet s gravitational perturbations in addition to the secular perturbations already discussed ( @xcite ; @xcite ; @xcite ; @xcite ) .
some of the emitting particles might be trapped in its resonant ring .
such a ring could be responsible for some of the observed lobe asymmetry .
the existence of such a ring would give the inner edge of the disk structure that co - orbits with the planet ; i.e. , observations of this structure would vary on timescales of @xmath457 years , offering a method of distinguishing between this structure and the large - scale background structure , which would vary on secular timescales of @xmath332(1 myr ) .
the asymmetric structure of the earth s resonant ring includes a @xmath458 au@xmath113 cloud of dust located permanently in the earth s wake with a number density @xmath459 above the background ( @xcite ) .
observations of such an asymmetric structure in an exosolar disk could be modeled using the same techniques that were used to model the earth s resonant ring ( @xcite ; @xcite ) , possibly allowing us to determine the presence , location , and even the mass of the perturbing planet ( @xcite ) .
however , the evidence suggests that such observations may not be possible with currently available technology .
in fact , calculations show that when viewed from a distant point in space normal to the ecliptic plane , the earth s `` wake '' would only have an ir signal @xmath460 times that of the earth ( @xcite ) .
so , regardless of the resolution requirements , if one were to observe the solar system from outside , it would be easier to detect the earth directly than to infer its existence from the structure of the zodiacal cloud .
the earth s resonant ring is a result of the trapping of particles that are evolving into the inner solar system due to p - r drag .
another method of forming a resonant ring is for a planet to undergo radial migration of its orbit , trapping all particles exterior to its orbit into its strongest resonances ; such an interaction is supposed to have happened between neptune and the kuiper belt ( @xcite ) .
the amount of disk material that is trapped in a kuiper belt ring depends on how much radial migration has taken place to the planets orbit .
a ring with all of its particles trapped in the 1:2 and 2:3 resonances with the planet ( these are the strongest resonances ) would have three lobes , with the planet residing in an `` empty '' fourth lobe .
could this be the cause of the tri - lobed structure observed in @xmath461 eridani ( @xcite ) ?
the primary intent of this paper was to show how the long - term effect of the gravitational perturbations , i.e. , the secular perturbations , of a massive perturber could be the cause of the @xmath0% brightness asymmetry of the double - lobed feature seen in observations of the hr 4796 disk ( t99 ) : 1 .
we showed how the secular perturbations of a massive perturber in a disk impose a forced eccentricity on the orbits of particles in the disk , thus causing the disk s center of symmetry to be offset from the star towards the perturber s apastron .
we also showed how the same perturbations impose a forced inclination on the particles orbits , which , if there is more than one perturber in the disk , could cause the disk to be warped .
we produced a model of the hr 4796 disk that accurately maps the 18.2 @xmath1 m brightness distribution observed by t99 ; this model is based on a consideration of the dynamics of the particles in the disk . the model shows how the brightness of a disk that has an inner clear region , that also has an offset center of symmetry caused by a forced eccentricity imposed on the disk particles orbits , would be asymmetric , since the inner edge of one side of the disk is closer to the star , and so is hotter and brighter , than the other side .
we showed that a forced eccentricity as small as 0.02 is all that is necessary to cause a 5% lobe brightness asymmetry in the hr 4796 disk .
if the eccentricity of orbit of the companion star , hr 4796b , is larger than 0.13 , then a forced eccentricity of 0.02 is to be expected .
however , if there is a planet of mass @xmath462 located close to the inner edge of the disk , then the forced eccentricity , and hence the asymmetry , imposed on material in the disk s lobes is controlled by the planet rather than the binary companion ; this could also cause the disk to be warped .
if a forced eccentricity is indeed the cause of the observed lobe asymmetry then observations that constrain the orbit of hr 4796b would help to clarify whether such a planet exists .
if the hr 4796 system had no binary companion , a forced eccentricity of 0.02 could have been imposed on the disk by a lone planet with a mass of @xmath463 , and an eccentricity of @xmath464 .
the statistical significance of the hr 4796 disk s lobe asymmetry in the observations of t99 is only at the @xmath465 level , however , it is also apparent in the observations of other authors ( @xcite ; @xcite ) .
it would take one good night on a 10 meter telescope to get a clear observation of the hr 4796 asymmetry .
thus , the indirect detection of planets , even small planets , hiding in circumstellar disks is clearly within reach using these dynamic modeling techniques .
this is particularly important , since the direct detection of planets around even nearby stars is well beyond current capabilities ( @xcite ) , and indirect detection techniques such as radial velocity ( @xcite ) or astrometric ( @xcite ) techniques , permit detection only of very massive planets that are close to the star .
5 . if there is a planet close to the inner edge of the disk , many of the disk s particles could be trapped in resonance with that planet , thus forming a resonant ring .
such a ring would give the inner edge of the disk structure that rotates on the timescale of the orbital period of the planet , @xmath457 years . this structure
could be contributing to the observed lobe asymmetry , and may also be present in the residuals of the observation .
this possibility could be explored with further observations .
resonant rings may be the predominantly observable structures of some exosolar systems , such as that recently observed around @xmath461 eridani ( @xcite ) .
the hr 4796 disk modeling also revealed important information about the large - scale symmetrical structure of the disk , as well as about the dynamic properties of its emitting particles : 1 .
the spatial distribution of material in the disk inferred from our modeling matches that inferred by other authors ( @xcite ; @xcite ; @xcite ) .
the surface density of cross - sectional area in the disk peaks at @xmath376 au from hr 4796a , falling off @xmath466 outside this radius , dropping by a factor of @xmath467 between 70 and 60 au , and falling to zero by 45 au .
this soft inner edge to the disk is to be expected if the disk particles orbits are eccentric .
2 . assuming the particles to be astronomical silicate mie spheres , the diameter of the emitting particles was estimated to be @xmath336 @xmath1 m .
particles this small have radiation forces that are characterized by @xmath468 , and so are blown out of the system on hyperbolic orbits on timescales of @xmath378 years .
the hr 4796 disk is very dense ; the collisional lifetime of its emitting particles is @xmath469 years .
thus , the emitting particles can not be primordial particles , rather they must be continuously created from a reservoir of larger particles .
the collisional lifetimes of all of the disk s particles are shorter than their p - r drag lifetimes ; i.e. , none of the disk s particles are affected by p - r drag .
further investigation of the particles properties needs to be done before any firm conclusions can be reached about whether the disk is dense enough to support a population of particles on hyperbolic orbits that is large enough to dominate the disk s emission . if the emitting particles are on hyperbolic orbits , the modeling implies a mass loss rate for the disk @xmath379year , and further modeling would have to be done to ascertain the spatial distribution of the dynamically stable population of large particles from which these emitting particles originated .
consider a collision between two disk particles , the larger of which is denoted by the subscript 1 , and the smaller by the subscript 2 .
for this collision to be `` catastrophic '' , that is , for it to result in the break - up of the larger particle , the impact energy of the collision must be large enough both to overcome the tensile strength of the larger particle , and to impart enough energy to the collisional fragments to overcome its gravitational binding energy . in the asteroid belt
this limit means that a collision is only catastrophic if @xmath470 ( @xcite ) .
since the impact energy of a collision is @xmath471 , assuming that exosolar disk particles have similar tensile strengths to the solar system s asteroids , this limit can be scaled to exosolar disks by the square of the ratio of the mean relative velocity of collisions in the asteroid belt ( at @xmath164 au ) , @xmath472 km / s ( @xcite ) , to that of collisions in the exosolar disk .
the mean relative velocity of collisions in exosolar disks can be described by : @xmath473 where @xmath474 is some function of the disk particles eccentricities and inclinations , and @xmath475 km / s is the average velocity of particles at @xmath9 ( eq . [
[ eq : vgrav ] ] with @xmath63 replaced by @xmath9 ) .
thus , assuming @xmath476 as for collisions in the asteroid belt , an exosolar disk particle of diameter @xmath477 , would only suffer a catastrophic collision if the other particle in the collision had a diameter @xmath478 , where @xmath479^{1/3}d .
\label{eq : dimp}\ ] ] the `` collisional lifetime '' , i.e. , the mean time between catastrophic collisions , of a particle of diameter @xmath8 , at a location in a disk denoted by @xmath480 and @xmath34 , is the inverse of its catastrophic collision rate ( @xcite ) : @xmath481^{-1 } , \label{eq : tcol1}\ ] ] where @xmath482 @xmath483 is the catastrophic collision cross - section seen by the particle , and @xmath484 is the mean encounter velocity of disk particles at @xmath9 ( eq . [ [ eq : vrel ] ] ) . using the definition of a disk s structure given by equation ( [ eq : sig2 ] ) ,
this catastrophic collision cross - section is : @xmath485 where @xmath486 and @xmath487 is the smallest particle with which a catastrophic collision could occur ( eq . [ [ eq : dimp ] ] ) . however , unless @xmath488 , the particle s orbit takes it through a range of @xmath33 and @xmath34 before a collision occurs ( there is also a variation of @xmath9 along the particle s orbit due to the eccentricity of its orbit ) .
thus , it is more appropriate to calculate the particle s collisional lifetime using the mean catastrophic collision rate of the particles in the size range @xmath32 that are in the spherical shell of radius , @xmath9 , and width @xmath156 .
consider an element of this shell that has a volume , @xmath489 .
the number of particles in the diameter range @xmath32 in this element is given by @xmath490 , and each of these particles has a catastrophic collision rate given by equation ( [ eq : rcoll ] ) .
integrating over the whole shell gives : @xmath491 ^ 2 d\theta \cos{\phi } d\phi \ast f_{cc}(d , r ) v_{rel}(r ) } , \label{eq : tcol3}\ ] ] where @xmath492 is the maximum inclination of the disk particles orbits to the reference plane . equation ( [ eq : tcol3 ] ) can be simplified by considering a cylindrical shell , defined by @xmath480 and @xmath202 , rather than a spherical one .
an element of the cylindrical shell has a volume @xmath493 , and the corresponding collisional lifetime of a particle in the shell is given by equation ( [ eq : tcol3 ] ) , but with @xmath34 , @xmath494 and @xmath495 , replaced by @xmath202 , @xmath496 , and @xmath497 , where @xmath498 .
here we introduce the parameter @xmath386 , the disk s face - on effective optical depth : @xmath499 where the dependence on @xmath33 has been dropped since orbits sample the full range of @xmath33 .
this is not a true optical depth , since that would include a consideration of the particles extinction coefficients ( @xmath500 ) ; rather , it is the disk s face - on surface density of cross - sectional area , which is equal to its true optical depth if its particles had @xmath501 . assuming that @xmath502 is independent of @xmath202 , so that @xmath503 ^ 2 dz = 0.5\tau_{eff}(r)^2/h$ ] , and that the encounter speed is determined by the vertical motion of particles in the disk , so that @xmath504 , equation ( [ eq : tcol3 ] ) can be simplified to : @xmath505 where @xmath90 is the average orbital period of particles at @xmath9 ( eq . [ [ eq : tper ] ] with @xmath63 replaced by @xmath9 ) .
a disk s effective optical depth , @xmath386 , can be estimated observationally from equation ( [ eq : brightness ] ) : @xmath506 where @xmath507 is the disk s face - on unsmoothed brightness .
the disk s face - on unsmoothed brightness can be calculated either from the observed brightness , making corrections to account for both the disk s orientation , as well as for the psf smoothing , or from the disk s total flux density , @xmath508 , and assuming the disk material to be evenly distributed between @xmath509 : @xmath510 where @xmath511 au@xmath104/pc@xmath104/sr , and @xmath512 is the distance of the star from the observer .
consider the particles in a disk that make up most of the disk s cross - sectional area , i.e. , those with diameters close to @xmath53 . by definition , these particles are most likely to collide with each other ( a collision that would definitely be catastrophic ) , and so their collisional lifetime can be found from equation ( [ eq : tcol3 ] ) using the approximation : @xmath513 applying this approximation to equation ( [ eq : tcol5 ] ) gives equation ( [ eq : tcol6 ] ) , which is also the collisional lifetime determined by artymowicz ( 1997 ) for particles in @xmath101 pictoris .
the collisional lifetime of particles of different sizes in a disk differ only by the factor @xmath514 .
this factor can be ascertained by making assumptions about the disk particles size distribution .
assuming that the size distribution of equation ( [ eq : nd ] ) holds for disk particles between @xmath28 and @xmath29 , the normalized cross - sectional area distribution is given by : @xmath515 substituting into equation ( [ eq : fcc ] ) gives : @xmath516 , \label{eq : fcc2}\ ] ] where @xmath517 for @xmath518 , and @xmath519 for @xmath520 ; the collisional lifetime of particles in a disk with this distribution is a minimum for particles for which @xmath521 .
the size distribution of particles in a real disk is more complicated than equation ( [ eq : sigd ] ) , however , equation ( [ eq : fcc2 ] ) can be used to give a crude approximation for the collisional lifetime of a disk s large particles : @xmath522 if the p - r drag lifetime , @xmath145 ( eq .
[ [ eq : tpr ] ] ) , is comparable to , or shorter than @xmath173 , then the effect of p - r drag on the collisional lifetime must be accounted for ; e.g. , on average , particles from a parent at @xmath523 , survive until they reach @xmath524 where @xmath525 .
also , if there is a significant change in @xmath9 along a particle s orbit due to the eccentricity of its orbit , a similar analysis can be done to take this into account when calculating its collisional lifetime . by considering the washer - like disk of particles on orbits with @xmath63 , @xmath64 , and random @xmath67 ( @xcite ) :
@xmath526dr]^{-1}$ ] , where @xmath527 is this disk s inner edge , and @xmath528 is its outer edge .
to find the secular evolution of the orbital elements of a particle that is affected by p - r drag , i.e. , one with @xmath176 , the equations governing the evolution of its complex eccentricity , @xmath529 ( eqs .
[ [ eq : edot ] ] and [ [ eq : zdot ] ] ) , and its complex inclination , @xmath530 ( eq . [ [ eq : ydot ] ] ) , must both be solved in conjunction with the p - r drag evolution of its semimajor axis ( eq .
[ [ eq : adot ] ] ) . while the solution given by equations ( [ eq : z ] ) and ( [ eq : y ] ) is no longer applicable , the decomposition of the particle s complex eccentricity and complex inclination into forced and proper elements , and the physical meaning of these elements , is still valid ; however , each of these elements now depends on the particle s dynamical history .
consider the p - r drag affected particles created by the break - up of the asteroid family group described in
[ sssec - opsfam ] . immediately after they are created ,
the orbital elements of these particles are the same as those of the rest of the family ; i.e. , they have semimajor axes @xmath63 , and complex eccentricities and complex inclinations that are uniformly distributed in these planes around circles of radii equal to the proper elements of the family , @xmath218 and @xmath220 . the dynamical evolution of a wave of these particles , i.e. , those that were created at the same time , can be followed by numerical integration to ascertain how the orbital elements of the particles in the wave vary as their semimajor axes decrease due to p - r drag ; this is the `` particles in a circle '' method ( @xcite ) .
it was found that the complex eccentricities and complex inclinations of a wave of particles originating in the asteroid belt remain on circles , and that as the wave s semimajor axis , @xmath531 , decreases : its effective proper eccentricity ( the radius of the wave s circle in the complex eccentricity plane ) decreases @xmath532 ; its effective proper inclination ( the radius of the wave s circle in the complex inclination plane ) remains constant at @xmath220 ; the distributions of the particles @xmath251 and @xmath265 remain random ; while its effective forced elements ( the centers of the circles in the complex eccentricity and complex inclination planes ) have a more complicated variation ( @xcite ; @xcite ) .
thus , the orbital element distributions , @xmath243 and @xmath244 , of p - r drag affected particles are like that of the large particles , in that they are the addition of @xmath222 and @xmath223 to symmetrical proper element distributions ; however , their forced and proper elements are different for particles from different families , as well as being different for particles of different sizes and with different orbital semimajor axes .
this means that their spatial distribution is subject to offset and warp secular perturbation asymmetries . for disks in which p - r drag
is a significant physical process , the existence of an inner hole implies not only that there must be no large particles in the inner region , but also that there must be some mechanism that prevents the particles that are in the outer disk from evolving into the inner disk by p - r drag . in hr 4796 , and other disks in which p - r drag
is insignificant , this mechanism is collisions ; i.e. , particles created in the outer disk are broken up by collisions before they reach the inner region .
this section presents dynamical explanations of how a planet located at the inner edge of the disk would help to prevent the emitting particles from reaching the inner disk .
this is not just important for disks in which its individual particles are not affected significantly by p - r drag in their lifetime , since while the inner edge of a disk s largest particles may be at @xmath288 , the cumulative effect of p - r drag on all of the stages of the collisional cascade that a primordial particle goes through before the fragments are small enough to be blown out of the system by radiation pressure could mean that the inner edge of these blow - out particles is further in than @xmath288 . to assess whether the cumulative effect of p - r drag has an impact on the inner edge of the emitting particles in the hr 4796 disk ,
consider its size distribution to follow that assumed in equation ( [ eq : sigd ] ) , where @xmath533 @xmath1 m , and @xmath161 .
the lifetime of an intermediate particle can be found using equations ( [ eq : tcol5 ] ) and ( [ eq : fcc2 ] ) , and the amount of p - r drag evolution in its lifetime is ( eqs .
[ [ eq : adot ] ] and [ [ eq : betah ] ] ) : @xmath534 where @xmath535 and @xmath63 are measured in au , @xmath8 in @xmath1 m , and @xmath536 in years .
assuming that the largest fragment created in a collision has a diameter half that of the original particle , a @xmath101 meteoroid particle ( e.g. , a 2.5 @xmath1 m particle ) at 62 au , assuming it to have originated from a gravitationally bound particle ( i.e. , one with @xmath134 ) , can at the very most be removed by 27 generations from its primordial ancestor , which would at the very most have been originally at 62.1 au .
one consequence of a planet at the inner edge of the disk is that it accretes some of the particles that pass it on their way into the inner region .
a simple estimate for the proportion of particles lost in such a way can be obtained by considering the p - r drag evolution of a torus of particles with orbital elements @xmath63 ,
@xmath64 , @xmath65 , and random @xmath66 , @xmath67 and @xmath24 ; the volume of this torus is @xmath537 ( @xcite ) . in the time it takes for the torus to pass the planet , , the planet accretes a volume of dust given by ( @xcite ) : @xmath538 , where @xmath539 is the capture cross - sectional area , @xmath540 is the radius of the planet , @xmath541 is the escape velocity of the planet , and @xmath542 is the mean relative velocity of encounter between the planet and the particles .
thus , the proportion of dust accreted onto the planet , @xmath543 , is given by : @xmath544 where @xmath545 kg / m@xmath113 is the mean density of jupiter , @xmath546^{-1}$ ] , and @xmath75 is the velocity of the particle ; e.g. , if @xmath547 , then a jupiter - like planet at the inner edge of the disk would need @xmath548 to accrete all of the particles passing it ( i.e. , @xmath549 for @xmath134 ) .
another consequence of a planet at the inner edge of the disk is that it traps some of the disk particles into its exterior mean motion resonances .
the resulting resonant ring has three consequences that may aid with the formation of a clear inner region .
firstly , the calculation of the probability of accretion onto the planet given in equation ( [ eq : paccrn ] ) does not apply to the particles that are trapped in the ring .
it is thought that resonance trapping helps the accretion process , since trapped particles may leave the resonance upon a close encounter with the planet ( @xcite ) .
this means that the constraint on the mass of the planet derived in the last paragraph is an upper limit for the accretion process to be the sole removal mechanism .
secondly , the ring increases the number of particles that are lost by collisional break - up , since it both decreases the collisional lifetime of disk particles ( by increasing the number density of particles in the ring region ) and increases the amount of time it takes for particles that are trapped in the ring to reach the inner region .
finally , if the trapping timescales , @xmath550 , are longer than the age of the system , @xmath437 , then the trapped particles can not have reached the inner disk yet ; i.e. , the ring causes a bottle - neck in the flow of particles to the inner disk .
artymowicz , p. 1997 , annu .
earth planet .
, 25 , 175 artymowicz , p. , & clampin , m. 1997 , , 490 , 863 backman , d. e. , & paresce , f. 1993 , in protostars and planets iii , 1253 backman , d. e. , dasgupta , a. , & stencel , r. e. 1995 , , 450 , l35 backman , d. e. 1998 , in exozodiacal dust workshop conference proceedings , eds .
d. e. backman et al .
( nasa / cp-1998 - 10155 ) , 13 barrado y navascus , d. , stauffer , j. r. , hartmann , l. , & balachandran , s. c. 1997 , , 475 , 313 bowell , e. 1996 , asteroid orbital elements database ( lowell observatory ) brouwer , d. , & clemence , g. m. 1961 , methods of celestial mechanics ( new york : academic press ) burns , j. a. , lamy , p. l. , & soter , s. 1979 , icarus , 40 , 1 cohen , m. 1999 , private communication dermott , s. f. , nicholson , p. d. , burns , j. a. , & houck , j. r. 1984 , , 312 , 505 dermott , s. f. , nicholson , p. d. , burns , j. a. , & houck , j. r. 1985 , in properties and interactions of interplanetary dust , eds .
r. h. giese , & p. lamy ( dordrecht : d. reidel ) , 395 dermott , s. f. , & nicholson , p. d. 1986 , , 319 , 115 dermott , s. f. , gomes , r. s. , durda , d. d. , gustafson , b. . s. , jayaraman , s. , xu , y. l. , & nicholson , p. d. 1992 , in chaos , resonance and collective dynamical phenomena in the solar system , ed . s. ferraz - mello ( iau ) , 333 dermott , s. f. , jayaraman , s. , xu , y. l. , gustafson , b. . s. , & liou , j. c. 1994 , , 369 , 719 dermott , s. f. , jayaraman , s. , xu , y. l. , grogan , k. , & gustafson , b. .
s. 1996 , in unveiling the cosmic infrared background , ed .
e. dwek ( new york : aip ) , 25 dermott , s. f. , grogan , k. , holmes , e. k. , & wyatt , m. c. 1998 , in exozodiacal dust workshop conference proceedings , eds .
d. e. backman et al .
( nasa / cp-1998 - 10155 ) , 59 dermott , s. f. , grogan , k. , holmes , e. k. , & kortenkamp , s. 1999 , in formation and evolution of solids in space , ed
. j. m. greenberg ( dordrecht : kluwer ) , in press dohnanyi , j. s. 1969 , j. geophys .
res . , 74 ,
2531 draine , b. t. , & lee , h. m. 1984 , , 285 , 89 duncan , m. , quinn , t. , & tremaine , s. 1989 , icarus , 82 , 402 durda , d. d. , & dermott , s. f. 1997 , icarus , 130 , 140 durda , d. d. , greenberg , r. , & jedicke , r. 1998 , icarus , 135 , 431 fajardo - acosta , s. b. , telesco , c. m. , & knacke , r. f. 1993 , , 417 , l33 fisher , r. s. , telesco , c. m. , pia , r. k. , knacke , r. f. , & wyatt , m. c. 1999 , , submitted gatewood , g. 1996 , , 188 , 4011 gorkavyi , n. n. , ozernoy , l. m. , mather , j. c. , & taidakova , t. 1997 , , 488 , 268 greaves , j. s. , et al .
1998 , , 506 , l133 grogan , k. , dermott , s. f. , jayaraman , s. , & xu , y. l. 1997 , , 45 , 1657 grn , e. , zook , h. a. , fechtig , h. , & giese , r. h. 1985 , icarus , 62 , 244 gurnett , d. a. , ansher , j. a. , kurth , w. s. , & granroth , l. j. 1997 , geo .
lett . , 24 , 3125 gustafson , b. .
s. 1994 , annu .
earth planet .
, 22 , 553 hirayama , k. 1918 , , 31 , 185 holland , w. s. , et al .
1998 , , 392 , 788 jayaraman , s. , & dermott , s. f. 1999 , icarus , submitted jayawardhana , r. , fisher , r. s. , hartmann , l. , telesco , c. m. , pia , r. k. , & fazio , g. 1998 , , 503 , l79 jewitt , d. c. 1999 , areps , in press jura , m. , zuckerman , b. , becklin , e. e. , & smith , r. c. 1993 , , 418 , l37 jura , m. , ghez , a. m. , white , r. j. , mccarthy , d. w. , smith , r. c. , & martin , p. g. 1995 , , 445 , 451 jura , m. , malkan , m. , white , r. , telesco , c. m. , fisher , r. s. , & pia , r. k. 1998 , , 505 , 897 kalas , p. , & jewitt , d. c. 1995 , , 110 , 794 kalas , p. 1998 , science , 281 , 182 kelsall , t. , et al .
1998 , , 508 , 44 kessler , d. j. 1981 , icarus , 48 , 39 kimura , h. , & mann , i. 1998 , , 499 , 454 koerner , d. w. , ressler , m. e. , werner , m. w. , & backman , d. e. 1998 , , 503 , l83 kortenkamp , s. j. , & dermott , s. f. 1998 , icarus , 135 , 469 laor , a. , & draine , b. t. 1993 , , 402 , 441 lecar , m. , & franklin , f. 1997 , icarus , 129 , 134 leinert , c. , & grn , e. 1990 , in space and solar physics , vol .
20 , physics and chemistry in space : physics of the inner heliosphere i , eds . r. schween , & e. marsch ( berlin : springer ) , 207 liou , j. c. 1993 , ph.d .
thesis , univ . of florida lissauer , j. j. 1993 , , 31 , 129 love , s. g. , & brownlee , d. e. 1993 , science , 262 , 550 low , f. j. , et al .
1984 , , 278 , l19 malhotra , r. 1995 , , 110 , 420 malhotra , r. 1996 , , 111 , 504 marcy , g. w. , & butler , r. p. 1998
, , 36 , 57 murray , c. d. , & dermott , s. f. 1999 , solar system dynamics , ( cambridge : cambridge university press ) reach , w. t. , et al .
1995 , nature , 374 , 521 roques , f. , scholl , h. , sicardy , b. , & smith , b. a. 1994 , icarus , 108 , 37 schneider , g. , et al . 1999 , , 513 , l127 sitko , m. l. , grady , c. a. , lynch , d. k. , russell , r. w. , & hanner , m. s. 1999 , , 510 , 408 stauffer , j. r. , hartmann , l. w. , & barrado y navascus , b. 1995 , , 454 , 910 stern , s. a. 1995 , , 110 , 856 stern , s. a. , & colwell , j. e. 1997 , , 490 , 879 sykes , m. v. , lebofsky , l. a. , hunten , d. m. , & low , f. j. 1986 , science , 232 , 1115 sykes , m. v. 1990 , icarus , 84 , 267 telesco , c. m. , decher , r. , becklin , e. e. , & wolstencroft , r. d. 1988 , , 335 , 51 telesco , c. m. , & knacke , r. f. 1991 , , 372 , l29 telesco , c. m. , et al . 1999 , , submitted
trilling , d. e. , benz , w. , guillot , t. , lunin , j. i. , hubbard , w. b. , & burrows , a. 1998 , , 500 , 428 vedder , j. d. 1998 , icarus , 131 , 283 ward , w. r. , & hahn , j. m. 1998 , , 116 , 489 wisdom , j. 1980 , , 85 , 1122 | recent images of the disks of dust around the young stars hr 4796a ( @xcite ; @xcite ; @xcite ; @xcite ) and fomalhaut ( @xcite ) show , in each case , a double - lobed feature that may be asymmetric ( one lobe may be brighter than the other ) .
a symmetric double - lobed structure is that expected from a disk of dust with a central hole that is observed nearly edge - on ( i.e. , close to the plane of the disk ) .
this paper shows how the gravitational influence of a second body in the system with an eccentric orbit would cause a brightness asymmetry in such a disk by imposing a `` forced eccentricity '' on the orbits of the constituent dust particles , thus shifting the center of symmetry of the disk away from the star and causing the dust near the forced pericenter of the perturbed disk to glow .
dynamic modeling of the hr 4796 disk shows that its @xmath0% brightness asymmetry could be the result of a forced eccentricity as small as 0.02 imposed on the disk by either the binary companion hr 4796b , or by an unseen planet close to the inner edge of the disk .
since it is likely that a forced eccentricity of 0.01 or higher would be imposed on a disk in a system in which there are planets , but no binary companion , the corresponding asymmetry in the disk s structure could serve as a sensitive indicator of these planets that might otherwise remain undetected . |
observations of solar , atmospheric , reactor and long - baseline neutrinos have provided compelling evidence for neutrino oscillations @xcite .
in fact , the latest analysis of the data from the super - kamiokande ( super - k ) collaboration @xcite presents a clear and unambiguous evidence for the oscillation hypothesis @xcite by establishing the first _
oscillation minimum _ and hence for non - zero ( and different ) neutrino masses and mixing . in the meanwhile ,
the sudbury neutrino observatory ( sno ) @xcite has provided incontrovertible evidence for neutrino oscillation from solar neutrino data .
clearly neutrino physics is now entering an era of precision measurements .
the focus from now on will be to precisely determine the oscillation parameters apart from the observation of the oscillation pattern beyond the first minimum . given the reach of present and proposed experiments , the goal is to understand in detail the sensitivity of the oscillation probabilities to fundamental neutrino parameters , which are the mass - squared differences and mixing angles and phases .
it is therefore crucial to choose processes or observables that maximise the sensitivity to the parameter(s ) of interest .
it turns out that the density profile of matter that the neutrinos pass through before detection naturally accentuates the sensitivity to the mixing angles @xmath2 , @xmath0 and the mass squared difference @xmath1 . these _
matter effects _ have been studied before , with atmospheric neutrinos , in various contexts , in refs .
in particular , substantial matter dependences occur in the energy range of interest only if @xmath5 is different from zero ; a non - zero value of this parameter also opens the window to cp violation in the lepton sector . since only an upper bound exists on this parameter so far ,
the variation in sensitivity to other parameters can be large or small , depending on the true value of @xmath2 .
this paper is devoted to the detailed study of oscillation probabilities of energetic ( gev energy ) neutrinos and anti - neutrinos propagating through the earth in a three flavour framework using accurate numerical methods .
the paper is divided into two parts : the first part is devoted to the broad theme of sensitivity of the neutrino ( and anti - neutrino ) oscillation probabilities to the fundamental neutrino parameters in the gev region .
the results of this analysis are equally applicable to atmospheric neutrinos as well as to neutrino beams from future neutrino factories .
while some of this information is already known , we emphasise a few surprising variations . having determined the various sensitivities , we then use the results of the first part to maximise matter effects in a study of atmospheric neutrinos ( and anti - neutrinos ) using detectors having charge identification capability and study how best these parameters can be constrained / determined by such an analysis .
the paper is organised as follows . in section 2 ,
we outline the framework for the calculation of neutrino probabilities . though numerical procedures have been developed to calculate any desired survival or conversion probability , in this paper we focus on the probabilities relevant for the accurate calculation of muon - neutrino ( and anti - neutrino ) events at a detector .
we also list here the current limits on the oscillation parameters . in section 3 ,
we consider , in turn , the variation of the muon neutrino survival and conversion probabilities as the oscillation parameters are varied within the allowed limits .
the compendium of results listed here applies generically to any combination of neutrino source and detector where gev energy neutrinos traverse the earth . in section 4 ,
we present results for event rates calculated with atmospheric neutrinos .
we also identify in passing interesting regions in energy nadir angle space for long - baseline neutrinos .
the results are calculated for up / down ratios of event rates binned as a function of the zenith angle or equivalently distance @xmath6 to bring out the maximal sensitivity to parameters , especially @xmath2 , @xmath0 and the sign of @xmath1 .
the question of hierarchy from the sign of @xmath1 has been discussed in detail before @xcite ; we focus here on the sensitivity of event rates and event ratios to the magnitude of this parameter .
an important open question is the deviation from maximal of the mixing angle @xmath0 , that is , its octant sensitivity .
this was analysed recently by choubey and roy @xcite ; we present results here which are complementary to their analysis .
the issue of both the mass hierarchy and the octant of @xmath0 have relevance to the structure of neutrino mass models .
in section 5 we present a summary of our results and conclusions ; we also highlight the improvement over the earlier results .
some remarks on the numerical computation are presented in the appendix .
we also discuss the possible points of departure from earlier calculations , especially those that have been used in neutrino event generators such as nuance @xcite .
for completeness we review the basic framework used in evolving the neutrino states and the calculation of oscillation probabilities . in a three neutrino framework ,
the neutrino flavour states @xmath7 , @xmath8 , are defined as linear superpositions of the neutrino mass eigenstates @xmath9 , @xmath10 , with masses @xmath11 : @xmath12 the @xmath13 are the states produced in association with the charged leptons
. the @xmath14 unitary matrix @xmath15 may be parametrised @xcite ( ignoring majorana phases , which may be included by multiplying @xmath15 by a diagonal phase matrix involving two more phases in addition to @xmath16 ) as : @xmath17 where @xmath18 , @xmath19 and @xmath16 denotes the cp violating ( dirac ) phase .
the 3 @xmath20 3 neutrino mass matrix @xmath21 in the charged - lepton mass basis is diagonalised by @xmath15 : @xmath22 the fundamental neutrino parameters therefore are the mixing angles @xmath23 , phase @xmath16 , and the mass - squared differences @xmath24 .
the time evolution of the mass eigenstates is governed by the schroedinger equation , @xmath25 consequently the time evolution of the flavour states is given by the equation , @xmath26 = \frac{1}{2e } u m_\nu^2 u^\dagger \left[\nu_\alpha\right]~.\ ] ] where @xmath27 $ ] is the vector of flavour eigenstates , @xmath28^t = \left[\vert\nu_e\rangle,\vert\nu_{\mu}\rangle,\vert\nu_{\tau}\rangle \right]$ ] . the evolution equation in the presence of matter is given by @xmath29 = \frac{1}{2e } \left[u m_{\nu}^2 u^{\dagger } + \left ( \begin{array}{ccc } a & 0 & 0 \\ 0 & 0 & 0 \\ 0 & 0 & 0 \end{array } \right)\right ] \left
[ \nu_\alpha \right]~ , \label{eq : ham}\end{aligned}\ ] ] where the matter term @xmath30 ( ignoring the diagonal neutral current contribution ) is given by @xmath31 here @xmath32 and @xmath33 are the fermi constant and electron number density in matter and @xmath34 is the matter density .
the evolution equation for anti - neutrinos has the sign of @xmath30 and the phase @xmath16 reversed . in general ,
the 3-flavour probabilities , in particular , @xmath35 and @xmath36 of interest here , depend on all the oscillation parameters : @xmath37 , @xmath0 , @xmath2 , @xmath38 , @xmath1 , and the cp phase @xmath16 .
we will focus everywhere on effects due to propagation in ( earth s ) matter .
it is useful to compute these probabilities in the constant density approximation for comparison with the exact numerical results below .
we have @xcite , @xmath39~ , \\
p_{e\mu } & \approx & \sin^2 \theta_{23 } \sin^2 2\theta_{13}^{m}\ , \sin^2 \delta_{31}^m~,\nonumber ~ \label{eq : p3num}\end{aligned}\ ] ] where terms involving @xmath38 have been ignored @xcite and the superscript @xmath40 refers to mixing angles and mass square differences in matter .
the relevant @xmath41-dependent terms in matter are given by , @xmath42~ , \nonumber \\ \delta_{32}^m & = & \frac{1.27 \delta_{32 } l}{e}\frac{1}{2}\ , \left[\frac{\sin 2\theta_{13}}{\sin 2\theta_{13,m } } + 1 + \frac{a}{\delta_{32 } } \right]~ , \nonumber \\ \delta_{31}^m & = & \frac{1.27 \delta_{32 } l}{e}\ , \left[\frac{\sin 2\theta_{13}}{\sin 2\theta_{13,m } } \right]~. \end{aligned}\ ] ] note that for sufficiently large values of @xmath43 , all the three scales are of the same order of magnitude including @xmath44 which can not therefore be neglected . while @xmath2 is small , its value in matter can be enhanced : @xmath45 in particular , @xmath46 at resonance , when @xmath47 it is convenient to introduce the scale @xmath48@xcite instead of @xmath1 : @xmath49 so that a normal or inverted hierarchy is simply indicated by a sign ( and not magnitude ) change in this parameter .
data on neutrino oscillations are available from the following experiments : * a combination of solar neutrino experiments and the kamland reactor@xcite experiment sets limits on the parameters @xmath38 and @xmath37 ; it also establishes the mass ordering @xmath50 . * a combination of atmospheric neutrino experiments and the k2k experiment@xcite constrains the parameters in the ( 23 ) sector : @xmath1 ( or @xmath48 ) and @xmath0 .
note that the sign of this mass - squared difference as well as the deviation of @xmath0 from maximality are not yet determined . *
the reactor neutrino experiment , chooz@xcite , sets an upper bound on the effective ( 13 ) mixing angle using the above parametrisation . *
the cp phase @xmath16 is unknown
. a combined analysis of these data @xcite places limits on the oscillation parameters as summarised in table [ tab : bestfit ] .
.table showing currently accepted @xcite best - fit values of oscillation parameters with 2@xmath51 errors . in the case of the mixing angle @xmath2
, a 2@xmath51 upper bound is shown . [ cols="<,>",options="header " , ]
we now discuss in detail the @xmath52 , @xmath53 survival and conversion probabilities as these are of interest here .
note that , in the absence of oscillations , the number of events in the neutrino sector is always a factor of two or more larger than in the case of anti - neutrinos due to the larger neutrino cross - section .
we will comment further on its implication in section 4 .
the extraction of neutrino parameters to high precision requires the determination of neutrino oscillation probabilities in matter to a high degree of accuracy .
analytically , the effect of non - uniform matter poses a problem , while , in principle , the oscillation probabilities may be computed numerically to the requisite precision .
many , sometimes complicated , analytical formulae exist for neutrino propagation in vacuum and in matter with either constant or variable density @xcite .
we use a runge - kutta solver to calculate the oscillation probabilities for various energies and nadir angles .
some technical details are given in appendix a. all results presented in this and the following sections have been obtained using the density profile of the earth as given by the preliminary reference earth model ( prem ) @xcite and numerically evolving the flavour eigenstates through earth s matter . in particular , the approximate expressions for the probabilities as shown in eq .
[ eq : p3num ] are not used . the matter profile in the prem model is shown in fig .
[ fig : prem ] as a function of the radius @xmath54 from the centre of the earth . the density jumps at inner - outer core and core - mantle transitions are clearly seen .
an up - going neutrino with nadir angle @xmath55 ( shown in the upper @xmath56-axis ) traverses all density zones of radii larger than the corresponding @xmath54 shown in the lower @xmath56-axis . for @xmath57 ,
neutrinos graze the core - mantle boundary , while for smaller @xmath55 , they traverse the core .
we also implicitly assume an iron calorimeter detector with charge identification capability as in the minos@xcite and the proposed monolith @xcite and ical / ino @xcite detectors .
we therefore focus on neutrino energies in the gev range which is of relevance to these detectors .
we will see that the most interesting and sensitive region , purely in terms of the probabilities , is between 210 gev .
we look at the sensitivity of the oscillation probabilities to the fundamental neutrino parameters , namely the mass squared differences @xmath38 , @xmath48 , the mixing angles @xmath37 , @xmath2 , @xmath0 and the cp phase @xmath16 . in order to assess the parameter sensitivity , we consider @xmath58 variations from the current best fit value of each parameter while the others are kept at their best fit values .
we shall see that the probabilities are sensitive mainly to the ( 23 ) and ( 13 ) mixing angles and the larger mass - squared difference @xmath48 .
when matter effects are turned on the probabilities are also sensitive to the sign of this mass - squared difference .
there is also a dependence on the density profile of the earth .
we shall discuss these dependences one by one .
since matter effects are proportional to @xmath59 , which is small , we are looking for sub - leading effects .
the main role of matter is to enhance @xmath59 due to resonance effects .
hence , when we emphasise a large effect , it is understood to be `` large '' with respect to this small parameter ; matter effects , unfortunately , remain rather small with respect to the total events / processes we consider and hence need large exposures for their study .
we shall consider neutrino energies in the range 210 gev , which are relevant for atmospheric neutrino studies .
typically , we shall hold other parameters at their known central values when we discuss the variation with any one parameter .
in particular , there is only a marginal / negligible dependence of these probabilities on @xmath37 and @xmath60 and we shall not discuss them further . in general the normal hierarchy is assumed ; this means that all matter effects are enhanced due to resonance in the neutrino sector , with smaller effects ( and no resonances ) in the anti - neutrino sector . since we are only discussing probabilities
, there is an approximate symmetry between particle probabilities with normal hierarchy and anti - particle probabilities with inverted hierarchy @xcite .
this symmetry is exact in the limit that terms containing @xmath38 can be ignored .
thus we choose @xmath48 to be positive ( normal hierarchy ) while discussing probabilities .
results with the inverted hierarchy are discussed in the next section . in general ,
matter effects are significant provided @xmath2 is not small .
the dependence of matter effects on this parameter has been exhaustively studied elsewhere @xcite and will not be discussed in any detail in this paper .
we merely show a sample dependence on this parameter in figs .
[ fig:13 t ] and [ fig : e13 t ] where @xmath35 and @xmath36 are plotted as a function of the nadir angle @xmath55 for three energies , @xmath61 gev .
the variation in the probabilities as @xmath2 varies over its 2@xmath51 allowed region , @xmath62 , is significant at small nadir angles for low energies ( @xmath63 gev ) , at virtually all nadir angles at larger energies ( around @xmath64 gev ) , and tapers off at larger energies ( by about @xmath65 gev ) when matter effects become irrelevant .
notice the effect of core crossing at nadir angle @xmath66 which is especially visible when @xmath67 gev when @xmath68 . . for completeness
we show the relatively weaker sensitivity to this parameter of the anti - neutrino probability @xmath69 in fig .
[ fig:13bart ] . when the hierarchy is inverted the behaviour of @xmath70 and @xmath71
are interchanged .
we begin with @xmath35 , which is extremely sensitive to this parameter , as can be seen in fig .
[ fig:32 t ] .
the figure shows the variation of @xmath35 with the nadir angle @xmath55 at different neutrino energies , for a 2@xmath51 variation of @xmath48 : @xmath72 ev@xmath73 .
the top three panels are for @xmath74 and the lower three for @xmath75 at three different energies .
notice that the smooth oscillations seen in the lower panels are modified due to matter effects ( turned on when @xmath76 ) in the upper panel .
in general , the modification is significant in the small nadir angle bins for lower energies ( @xmath77 gev ) , and at all nadir angle bins in the intermediate energy region ( represented by the central panel at @xmath67 gev ) when resonance occurs in the mantle .
the interplay between the variation in @xmath48 ( even at the 2@xmath51 level ) and that in the relatively unknown @xmath2 is clear from the @xmath67 gev panels : the same muon survival probability for upward - going neutrinos ( @xmath78 ) can come from ( @xmath79 ) both large : ( @xmath80 ev@xmath81 ) or both small : ( @xmath82 ev@xmath73 , @xmath83 ) .
since matter effects are relatively unimportant for anti - neutrinos ( in the normal hierarchy ) the anti - neutrino survival probability at all allowed @xmath2 is close to the neutrino survival probability at @xmath84 .
this can be seen from fig .
[ fig:32bart ] where @xmath69 is shown .
in fact , the difference between the neutrino and anti - neutrino survival probabilities is a measure of the matter effect and hence that of @xmath2 and the sign of @xmath85 ( the mass hierarchy ) as has been discussed earlier @xcite .
the conversion probability @xmath36 is small unless @xmath2 is substantial .
we show the high sensitivity of this probability to variations in @xmath48 for @xmath86 in fig .
[ fig : e32 t ] .
an interesting feature is the substantially larger probability for small nadir angles ( 0@xmath87 ) and low energy ( @xmath77 gev ) for the @xmath88 ev@xmath73 value due to resonance inside the core .
this feature is visible only at low @xmath48 values : for example , for @xmath77 gev , it occurs roughly for @xmath89 ev@xmath73 , with a small uncertainty in the limit due to the unknown @xmath2 . for larger values of @xmath48 , larger densities than are available inside the earth
are needed to satisfy the resonance condition , eq .
[ eq : resonance ] , at @xmath77 gev ; hence there is no substantial enhancement of these probabilities in the small nadir - angle region .
_ a substantial ( @xmath90% ) @xmath36 in this region at small energies around @xmath91 gev is thus a clear indication that _ @xmath92 ev@xmath73 . while this fact may not have an impact on the overall event rates for atmospheric neutrinos
, it will be important for long baseline neutrinos since one can separately study @xmath36 through _ wrong sign muons_. the anti - neutrino conversion probability , however , remains small even for large @xmath2 and is therefore not shown .
the super - kamiokande data favour a best - fit value of @xmath93 in a 2-flavour analysis .
although a recent 3-flavour global analysis @xcite indicates that this parameter may be slightly smaller , @xmath94 , the 2@xmath51 allowed parameter space spans both octants : @xmath95 .
the dominant ( matter - independent ) term in the survival probability depends on @xmath96 .
it is therefore difficult to substantially decrease errors on @xmath0 via a measurement of the leading contribution .
furthermore , even if a deviation from the maximal value is measured , it is impossible to determine the octant with such a term .
methods to determine the octant of @xmath0 have been discussed in detail elsewhere @xcite .
sensitivity to the octant of this angle is realised through sub - leading terms that depend on @xmath0 , not @xmath97 , as can be seen from eq .
[ eq : p3num ] .
these are matter - dependent and sub - leading because they are also proportional to @xmath59 , which is known to be small but is otherwise unknown .
[ fig:23 t ] shows the sensitivity of @xmath35 to variations in @xmath0 . for clarity of discussion
, we choose a central value of @xmath98 ( that is , choose a maximal value rather than the slightly smaller central best - fit value as used earlier ) and the _ same _ 2@xmath51 variation about it .
when @xmath75 , symmetric 2@xmath51 variations about the central value of @xmath98 lead to the _ same _ survival probability , as can be seen at all neutrino energies in the lower panels of the figure .
when @xmath2 is not zero , the survival probability for @xmath0 less than maximal ( red dotted line ) lies consistently above the central value ( solid line ) for all values of @xmath48 , @xmath99 and nadir angles .
this happens because the terms in @xmath35 involving @xmath100 and @xmath101 come with the same relative ( negative ) sign , as seen from eq .
[ eq : p3num ] . as @xmath0 increases from the first octant towards
maximal , both these terms are increasing so that the probability systematically decreases .
however , as @xmath0 increases from the maximal to a value in the second octant , @xmath102 increases while @xmath101 _ decreases _ , leading to a complicated dependence of the probability on this parameter .
hence no such regular behaviour is seen when @xmath0 is larger than @xmath98 although , in a large portion of phase space at @xmath6710 gev , the probability for @xmath103 @xmath98 is systematically larger ( smaller ) than that at @xmath104 .
it is clear from fig .
[ fig:23 t ] that , for non - zero @xmath2 , the probability curve for @xmath0 in the first octant ( @xmath105 ) is better separated from that for @xmath106 , than the curve for @xmath0 in the second octant ( @xmath107 ) .
evidently , the ability to determine whether @xmath0 is different from maximal will depend on which octant the true value of @xmath0 lies in .
this is also reflected in our numerical results with atmospheric neutrino events , as discussed in the next section .
precisely the opposite effect is seen in the conversion probability @xmath36 as seen in fig .
[ fig : e23 t ] . here , the probability increases with @xmath0 for all @xmath48 values at all energies and nadir angles so that the probability for @xmath103 @xmath98 is _ systematically larger ( smaller ) than the probability with @xmath108 .
_ this can be easily seen from the fact that there is a single term proportional to @xmath100 in eq .
[ eq : p3num ] for @xmath36 , appearing with the opposite sign to the matter terms in @xmath35 .
hence a long base - line experiment , where @xmath109 and @xmath110 beams are separately available , will be able to clearly establish the octant of @xmath0 by studying the @xmath36 conversion probability via the wrong - sign muon events . note that if @xmath84 exactly , then the octant determination can not be made from @xmath35 although @xmath36 still shows the same dependence on @xmath0 as discussed above .
however , @xmath36 is insignificant in this energy range and hence octant discrimination at @xmath75 is possible only through a study of @xmath36 at low energies around @xmath111 gev , when resonance effects in the ( 12 ) sector enhance this probability .
this is outside the scope of the present paper and will be discussed in a separate publication .
this relatively systematic dependence on @xmath0 is in contrast to the rather complex dependence of the survival and conversion probabilities on @xmath48 and @xmath2 .
both these probabilities contribute to the observed muon events in studies of atmospheric neutrinos .
we shall explore this further in a later section to maximise this `` octant - seeking '' effect .
in contrast , the matter dependence is small for the anti - neutrinos ; hence , the anti - neutrino survival probabilities for all @xmath2 values are similar to the neutrino survival probability at @xmath112 and do not distinguish the octant of @xmath0 .
we reiterate that this is a consequence of the chosen hierarchy ; opposite results are obtained with the inverted hierarchy ( matter effects are ultimately larger in the normal hierarchy since the anti - neutrino cross - sections are smaller than the neutrino cross - sections so that event rates are larger and matter effects better differentiated in the neutrino sector ) . in order to compute the effect of earth s matter on the probabilities , we have used the prem model @xcite which gives the density profile of a `` spherical equivalent earth ''
. while the locations of the discontinuities and mantle - core transitions are rather well - known , the absolute values of the densities themselves are not so well established .
hence , in order to study effects of uncertainties in the earth s density , we have retained the locations of the discontinuities while changing the density values .
we have done this in a simple manner by changing all the densities within the core by @xmath113 = 0.6 gm / cc ( a change of roughly 5% , as allowed by the model ) , and suitably adjusting the mantle densities in such a way as to maintain the overall mass of the earth to be constant .
such a procedure results in a much smaller change in the mantle densities of @xmath114 , in the opposite direction .
the consequences of such a density change to the neutrino survival and conversion probabilities are shown in figs .
[ fig : det ] and [ fig : edet ] respectively . here
@xmath115 and @xmath116 correspond to increasing and decreasing the core densities by 5% , with the mantle densities adjusted suitably .
the solid black line is for the original prem density profile .
again , the size of matter effects depends on @xmath48 ; changing the density alters the @xmath48 at which resonance occurs , for a given neutrino energy .
a point to note is the difference between the prem profile and constant density slabs .
the former has a continuously changing density profile , even within slabs .
hence neutrinos with a band of energy can undergo resonance in any given slab .
in contrast , only a neutrino with a single definite energy can undergo resonance in a constant density slab ( apart from having purely adiabatic propagation inside any slab ) . due to this , the probability curve as a function of zenith angle for a given energy can look very different when computed with the prem profile and with a constant density approximation , even when the latter closely follows the prem profile .
however , when an integration over an energy bin is performed , such variations get averaged out . more details on this issue
are given in the appendix . at low energies where the matter effects are significant only for core - crossing neutrinos at small nadir angles , the effects of altering the density profile
are also seen only at these small nadir angles , as can be seen from the upper panels of the figures . at higher energies ,
changes appear at larger nadir angles as well , as can be seen from the lower panels , since resonance ( and enhanced matter effects ) occurs in the mantle itself .
while the variations are small , except at small nadir angles , again , they are complex and depend on both energy and a precise knowledge of @xmath48 . probing the earth density , especially in the core , with neutrino oscillation physics using neutrino factory beams has been discussed earlier @xcite . here , only the dependence on @xmath2 was considered while @xmath48 was kept fixed .
we remark that in all cases there is also a dependence on @xmath2 at small nadir angles since the variations with @xmath2 are substantial here , as can be seen from fig .
[ fig:13 t ] .
the dependence on density variations may be disentangled from that on @xmath2 since substantial variations in the intermediate nadir angle region can only be due to @xmath2 , as seen from fig .
[ fig:13 t ] .
the anti - neutrino probabilities again do not show any significant dependence on these parameters in the normal hierarchy .
we now explore the dependence of the probabilities on the cp phase @xmath16 .
this dependence is shown in fig .
[ fig : cpt ] , again for @xmath74 .
apart from the dashed vertical line at @xmath117 that separates the `` up - going '' probability from the `` down - going '' one , the dotted vertical line at @xmath118 indicates the nadir angle corresponding to the so - called `` magic base - line '' @xcite where the probabilities are independent of the cp phase .
since the earth mantle density is not a constant , this is actually a band around that nadir angle , as can be seen from the figure .
the effect of the cp phase @xmath16 competes with the matter effect which in turn depends on the relative size of the matter term @xmath30 compared to @xmath48 .
the variation of the probabilities ( as a function of energy and nadir angle ) with @xmath16 for the current 2@xmath51 allowed range of @xmath48 is shown in fig .
[ fig : cpt ] .
this variation is so small that it is unlikely to be measured via a study of atmospheric neutrinos and may need long base - line measurements where the spectrum and nadir angle are well - known .
nevertheless , we make a few remarks on the general behaviour . while the survival probability @xmath35 is symmetric for @xmath119 , @xmath36 can distinguish the cases @xmath120 . at low energy , @xmath77 gev , the top panels in fig .
[ fig : cpt ] indicate that the survival probability @xmath35 is typically larger for @xmath121 at all nadir angles , smaller for @xmath122 and smallest for @xmath123 , _ independent _ of @xmath48 .
this is not true at larger energies , as can be seen from the lower panels of the figure .
hence , if these small variations can indeed be measured , information on @xmath16 may be obtained with some confidence , only if @xmath48 is known to much better than the current precision .
the situation is murkier with the conversion probability @xmath36 , as can be seen from fig .
[ fig : ecpt ] . the clear trend that was visible with @xmath35 at low energy is no longer present . at @xmath67 gev
there is a systematic trend between the curves for @xmath124 and @xmath125 and between @xmath126 and @xmath127 ; they behave in the same way relative to each other .
the complex effect of @xmath16 is overlaid on an already small @xmath36 .
signatures of @xmath36 are very clean however with long - base - line detectors ( via wrong - sign muon measurements ) ; this may facilitate a measurement of @xmath16 at such detectors , as has been discussed elsewhere @xcite .
the anti - neutrino probabilities are rather insensitive to matter effects or the cp phase ( in the normal hierarchy ) and we do not discuss them here .
while one can identify clearly the regions of sensitivity to the fundamental neutrino parameters through an analysis of the oscillation probabilities , it is crucial to identify the appropriate observables which will enable a precision measurement of the parameters .
the experimental observables are primarily the event rates in a detector . the fully differential event rate for neutrinos of flavour @xmath128 to be detected is given by the general expression : @xmath129 where @xmath130 and @xmath131 is the total interaction cross - section for the @xmath128 type neutrino to interact with the detector material . here @xmath132 is the conversion probability of a neutrino of flavour @xmath133 to a flavour @xmath128 ( which also includes the survival probability when @xmath134 ) .
the flux - dependent term @xmath135 is related to the doubly - differential neutrino ( or anti - neutrino ) flux of flavour @xmath133 , @xmath136 , which is a function of the energy @xmath99 and nadir angle @xmath55 through a jacobean of transformation . in the case of atmospheric neutrinos
a range of zenith angles are available while for long - baseline neutrinos this angle is kept fixed by the location of the source .
the factor @xmath137 is the detector dependent factor measured in units of kton - years .
while we assume the detector to be mainly made up of magnetised iron with active detector elements , this is not crucial since only the detector mass and exposure times are relevant . in the monolith and ical / ino proposals , where muons from charged - current @xmath110-nucleus interactions are detected ,
the active detector elements are resistive plate chambers ( rpcs , gas - filled glass chambers ) . in either of these proposals
the detector mass is almost entirely ( @xmath138 98% ) due to its iron content
. we will be interested here in _
event ratios _ ; hence the factor @xmath137 and other actual detector details will only determine the errors .
furthermore , we assume that the detector is capable of correctly identifying the charge of the muons in the final state .
this will be better than 98% for energies @xmath139 gev for the ical / ino detector .
this is crucial for the precision determination of the neutrino parameters in any future experiment .
we now focus on neutrino oscillation studies with atmospheric neutrinos .
both @xmath110 and @xmath109 ( and their anti - particles ) are produced , typically in the ratio 2:1 .
the distance of propagation @xmath6 of the neutrino from the point of production to the detector is given by , @xmath140 where @xmath141 is the average height ( taken to be 15 km ) above the surface of the earth at which the atmospheric neutrinos are produced , @xmath142 is the radius of the earth and @xmath143 , @xmath144 being the depth at which the detector is located underground ( chosen to be 1 km ) .
note that @xmath145 corresponds to neutrinos reaching the detector vertically upwards .
the event rate in a given bin of @xmath130 is , @xmath146 henceforth we discuss only the case of muon - neutrinos , @xmath147 ( or @xmath148 ) .
then both @xmath36 and @xmath35 contribute in the expression above .
however , due to the smallness of the ( 13 ) mixing angle , @xmath2 , the contribution of @xmath36 is generally small and is significant only near large @xmath41 where it is about 1020% ( 5% ) for neutrinos ( anti - neutrinos ) .
the event rate is expressed as a function of @xmath56 , averaged over a bin width that will be appropriately chosen to maximise the sensitivity to the sign of @xmath1 or the octant of @xmath0 . the expression given in eq .
[ eventb ] is the best - case scenario since the integration is over the neutrino or anti - neutrino energy .
we discuss the effect of including the detector resolution functions in the next section and proceed now to analyse the ideal case with a perfect detector .
a useful measure of oscillations is the ratio of up - coming to down - going neutrinos with nadir / zenith angles interchanged .
this is clear from fig .
[ fig : updown ] .
the fluxes of atmospheric neutrinos from directions @xmath55 and @xmath149 are expected to be similar in the absence of oscillations , especially for larger energies ( @xmath99 greater than a few gev ) .
since the path - length traversed , @xmath6 , is related to @xmath55 as @xmath150 ( see eq . [ ldef ] ) , the replacement @xmath151 effectively changes the sign of the second term in the equation above , thus taking , for instance , a down - going neutrino to an up - coming one .
the ratio of events in the up - down directions for a given @xmath152 , therefore , reflects the asymmetry of the up - down fluxes , due to oscillations .
we define @xcite , @xmath153 where @xmath154 and the number of up - coming ( u ) or down - going ( d ) events is calculated using eq .
[ eventb ] .
similarly the ratio @xmath155 defines the corresponding up / down ratio for anti - neutrinos .
since the effect of oscillations on the denominator is small , the ratio @xmath156 is effectively the ratio of oscillated to unoscillated events with the same @xmath41 . while reflecting the effect of oscillations
, this ratio also minimises errors due to the uncertainties in the overall flux normalisation ( which can be as large as 30% ) and those in the cross - sections ( about 10% ) .
we now proceed to determine these ratios numerically , and study their sensitivity to the oscillation parameters of interest .
we use the 3d atmospheric neutrino fluxes as calculated by honda et al .
these have been obtained from the nuance neutrino generator @xcite .
the charged current muon - neutrino nucleus cross - sections used have also been extracted from the nuance code . at the low end of the energy spectrum ,
the interaction with the detector material is mainly through quasi - elastic and resonance interactions .
when the energy is a few gev , deep inelastic scattering ( dis ) dominates , ultimately taking over by about 10 gev , as this cross - section is proportional to the neutrino energy .
there are still issues about the transition from the resonance ( mainly single pion dominated ) region to the dis region @xcite ; however , such errors are minimised by _ always analysing ratios of events_. roughly speaking , in the energy range of interest from few to 10 s of gev , half the events are from quasi - elastic and resonance interactions and the other half from dis .
this proportion is somewhat sensitive to finite detector resolutions , since the neutrino energy is smeared over a much larger range than that observed ; however , the events are still dominantly low - energy ones owing to the steep fall ( faster than @xmath157 ) of the incident atmospheric neutrino flux with energy .
we find , as studies of the probabilities suggest , that variations of the event rates due to uncertainties in @xmath38 and @xmath37 are small .
we therefore fix them to their best - fit values .
we also set the cp phase @xmath121 and fix the earth s density to be as given by the prem density profile @xcite .
the remaining parameters , viz . , @xmath48 ( both magnitude and sign ) , @xmath2 and @xmath0 ( in particular , its deviation from maximality and hence also its octant ) , are the parameters to be determined or constrained in the analysis .
a study of atmospheric neutrino events is sensitive to all these parameters ( but the magnitude of @xmath48 ) only through matter effects .
that is , @xmath2 must be non - zero and substantial for there to be any sensitivity to these parameters .
since this parameter is known to be small , it therefore follows that large exposures are required to determine any of the above parameters to better precision than are currently known . in order to proceed further
, we shall assume in the following that @xmath2 is substantial ; the precise threshold will be determined through the analysis .
it is already known that the sign of @xmath48 can be determined via the _ difference asymmetry _ of the up / down ratios in the neutrino and anti - neutrino channels : @xmath158 where @xmath15 and @xmath159 are the relevant differential event rates for up and down events as a function of @xmath41 and @xmath160 and @xmath161 the corresponding rates for anti - neutrinos .
this asymmetry can be maximised by choosing to integrate over @xmath41 bins that include one half - period of the oscillation .
this can be approximated by the matter - independent condition , @xmath162 a detailed analysis of this asymmetry has already been done in ref .
@xcite ; we only remark here that we require @xmath163 ( @xmath164 ) for the sign of @xmath48 to be determined using this technique .
furthermore , this determination requires 7501000 kton - year exposure , depending on the value of @xmath2 , thus setting the scale for our present analysis .
we focus therefore on the precision to which the magnitude of @xmath85 and the octant of @xmath0 can be determined , given such large exposures .
( current data from the super - kamiokande collaboration restrict these to the ranges @xmath165 ev@xmath166 and @xmath167 . )
this again depends on @xmath168 , which is expected to be shortly measured with good precision ( to within a few percent ) by the double - chooz experiment @xcite .
we study both cases : when @xmath2 is known and hence kept fixed in the analysis , as well as the case when it is allowed to vary freely .
we restrict ourselves to the normal hierarchy , with @xmath48 positive , so that matter effects ( and hence sensitivity to the octant of @xmath0 ) , are enhanced in the neutrino channel .
we integrate event rates over an energy range of @xmath16910 gev to maximise statistics while retaining sensitivity to matter effects . in fig .
[ fig : deltam223 ] we show the variation of the up / down events ratio for different @xmath48 values as a function of @xmath170 for two values of @xmath171 , in two octants . here
@xmath2 is fixed to @xmath172 .
two features are immediately noticeable : ( 1 ) the position of the minima and maxima in @xmath41 are not altered by changing @xmath0 ; furthermore , the events ratio at the first minimum is not very sensitive to @xmath0 , and ( 2 ) the effect of changing @xmath0 from a value in the first octant to a corresponding one in the second octant is to systematically _ decrease the event rates in all bins _ for all @xmath48 values .
furthermore , although not shown in the figure due to constraints of clarity , the curve for maximal @xmath106 lies between the two curves for @xmath173 in all @xmath41 bins beyond the first minimum .
such a systematic decrease with increase in @xmath0 was seen only in some zenith angle and energy ranges ( for example for @xmath174@xmath175 for @xmath67 gev and for @xmath174@xmath176 for @xmath177 gev , when @xmath74 ; see the top panels of fig .
[ fig:23 t ] ) . by a judicious choice of the energy and @xmath41 interval
, this effect has been converted to a systematic difference for all bins to the right of the first minimum ; moreover , it is in these bins that the sensitivity to @xmath0 is significant , so that this dependence is robust and easy to observe .
note that the difference asymmetry defined in eq .
[ eq : da ] is a difference between neutrino and anti - neutrino rates ratios . in this case , the systematic decrease with increasing @xmath0 is true in _ all _ @xmath41 bins for any @xmath48 value , by definition .
that is , the result for @xmath106 always lies between that for a value of @xmath0 in the first and second octant .
this is shown in fig .
[ fig : lebin ] .
however , as can be seen from the relative size of the error bars , the task of actually extracting the octant of @xmath0 from measurement of such an asymmetry is severely limited by statistics .
in fact , it is the poor statistics in the anti - neutrino sector that limits the efficacy of this parameter .
the effect of increasing @xmath2 away from zero is seen in fig .
[ fig : deltam213 ] where the events ratio is plotted as a function of @xmath178 for different @xmath48 values .
the two histograms for each @xmath48 value correspond to @xmath179 , where the latter value is simply chosen as being within the range allowed by current data and where matter effects are known to be substantial .
the effect of increasing @xmath2 is to turn on matter effects that move the muon survival probability maxima ( as a function of @xmath178 ) away from 1 and the minima away from zero .
this means that for non - zero @xmath2 the events ratio is smaller at probability maxima and larger at probability minima than for @xmath84 .
( note that the extent of deviation is somewhat suppressed by the sub - dominant contribution of @xmath36 to the dominant @xmath35 terms ; experiments such as those with neutrino factories where such contamination does not exist will be even more sensitive to matter effects ) .
the contrasting behaviour of the events up / down ratio with @xmath2 and @xmath0 for a given @xmath48 is clear : an increase in the latter systematically decreases the ratio , while an increase in the former decreases the ratio at the @xmath178 maxima and increases them at the @xmath178 minima .
such behaviour enables the extraction of the different parameters when the data is fitted to the theory . to demonstrate this ( and to determine the efficacy and precision with which the parameters can be determined ) , we begin by computing up / down neutrino events ratios for the set of input parameters , @xmath180 ev@xmath181 , in different @xmath182 bins from 1.6 to 3.4 of width 0.1 each .
we then fit this `` data '' to these three parameters , and extract the allowed parameter space to establish how well these parameters can be determined . in particular , we focus on the precision of measurement of all parameters as well as the possible determination of the octant of @xmath0 .
since we are considering ratios , we include only statistical errors in our analysis .
note that a zenith angle cut on very horizontal events that may not be easily detected by horizontally aligned detectors such as ical / ino only removes the small @xmath41 bins .
for instance , a cut of @xmath183 leads to a cut on @xmath184 .
for such small values , as can be seen from fig .
[ fig : deltam213 ] , there is negligible dependence of the up / down ratio on the neutrino oscillation parameters .
hence , such cuts will not affect our analysis . fig .
[ fig : m37 ] shows the allowed region at 95% cl and 99% cl in the @xmath185 parameter space , using standard @xmath186 minimisation , for a fixed value of @xmath187 ev@xmath73 . at 95% cl ,
the up / down ratio barely disallows maximal mixing in @xmath0 ; it also disallows the `` wrong '' octant solution of @xmath188 .
however , both these sensitivities go away at the 99% cl level when an island of allowed parameter space opens up around the wrong - octant solution , in a region to the right of it . at this point
we note that @xmath2 is likely to be rather precisely fixed ( to much better than a percent ) by the double - chooz experiment @xcite .
we therefore observe the effect of keeping @xmath2 fixed and varying the other two parameters .
this is shown in fig .
[ fig : mdelta37 ] where the allowed parameter space in @xmath189 space is shown for the same input parameters as before .
again , both the maximal solution and the wrong octant solution for @xmath0 are disallowed at 95% cl level .
while maximality is still disallowed , a small island opens up around the wrong - octant solution at the 99% cl , again to the right of it .
also , @xmath48 is constrained to a precision of better than 6% at 3@xmath51 .
the results are a complicated function of both @xmath2 and @xmath85 . in general , the wrong - octant and maximal solution is disallowed at 95% cl up to an input value of @xmath190 or @xmath191 , i.e. , a deviation of @xmath192 away from maximal in either direction ( or 10% deviation in @xmath193 ) but is allowed at 99% cl .
this result is mildly dependent on the value of @xmath2 with results obviously improving for larger @xmath2 .
for instance , the 99% cl contours for the same input data set are plotted in fig .
[ fig : mdelta39 ] for @xmath194 .
the precision in @xmath48 is marginally worse for the larger @xmath2 ; however , although at both values of @xmath2 maximality is disallowed at 99% cl , the wrong - octant solution is completely disallowed only for the larger ( 13 ) mixing angle .
it is possible to reduce the allowed parameter space by adding the contribution of the anti - neutrinos as well .
that is , we now consider the input `` data set '' to consist of both the neutrino and anti - neutrino events ratios .
since we always consider up / down ratios to minimise errors due to overall normalisation of fluxes and cross - sections , this still means that we need to separate the charged muons with good efficiency ( that is , identify the process as originating from a neutrino or anti - neutrino ) . in principle ,
charge misidentification can lead to systematic errors since , for example , a neutrino event wrongly identified not only is lost from its parent sample , but also adds to the anti - neutrino events .
however , in proposed experiments such as ical / ino , charge identification efficiency in this energy range is better than 98% so we ignore such correlation errors .
we show the allowed parameter space on including both the neutrino and anti - neutrino data sample in fig .
[ fig : mdelta7 ] . here
the complex dependence on the input value of @xmath48 is also shown : as @xmath85 is increased between its current 1@xmath51 allowed values of @xmath195@xmath196 ev@xmath73 , the allowed parameter space in @xmath189 around the wrong - octant solution shrinks , and disappears at the upper value of @xmath48 .
hence , it appears that deviations from maximality as well as determination of the octant can be typically easily established at the 95% cl level , provided @xmath2 is well - known ( and at least of the order @xmath197 ) .
results at the 99% cl level are harder to quantify and reflect the complex nature of this many - parameter problem .
we now ask how sensitive these results are to the finite resolution ( in determining neutrino energy and direction , and hence its @xmath41 ) of the detector .
inclusion of finite detector resolution reduces the sensitivity , especially beyond the first oscillation minimum and maximum in @xmath41 .
the smearing in direction affects the resolution in path - length .
a straightforward way to include such effects is to smear the observed energy and direction of the neutrino with gaussian resolution functions : @xmath198~ ; \\ \nonumber r_2(l ' , l ) & \equiv & \by{1}{\sqrt{{2}{\pi}}\sigma_l } \exp \left [ \by{-(l - l')^2}{2 \sigma_l^2 } \right]~. \label{resfns}\end{aligned}\ ] ]
hence , the event rate now includes the probability that a neutrino of a given @xmath199 and @xmath200 is detected in the detector with path - length @xmath6 and energy @xmath99 .
we have @xmath201 where @xmath202 is the jacobean of transformation .
we re - evaluate the event rates and the up / down ratios using this equation , with @xmath203 and @xmath204 .
these are realistic widths obtained by a geant analysis of atmospheric neutrino events by both the monolith and the ical / ino collaborations .
such a smearing has the effect of accounting for errors in correctly identifying the @xmath41 bin for a given event , and so accounting for bin - to - bin correlations .
we comment in passing on the use of finite widths in @xmath6 rather than in @xmath55 .
the smearing in @xmath6 actually arises because of the uncertainties involved in reconstructing the neutrino direction .
hence , the smearing should be in @xmath55 not @xmath6 . choosing a resolution in @xmath55 rather than in @xmath6 ( or equivalently
@xmath205 ) may give more realistic spreads , especially at large angles where smearing in @xmath205 may underestimate the large variations in @xmath6 for small changes in @xmath55 .
this can be seen in fig .
[ fig : dtheta ] . applying
a constant @xmath206 of 15% corresponds to large angular spreads at small angles where base - lengths are large , while a constant angular width @xmath207 of @xmath208 leads to larger spreads in baselines at large angles . in this paper
, we go with convention @xcite and define widths in the base - length @xmath6 . in fig .
[ fig : deltam223r ] we show the variation of the up / down events ratio for different @xmath48 values as a function of @xmath170 for @xmath171 when the resolution function is included .
the upper panels correspond to smearing in @xmath99 and @xmath55 while the lower panels correspond to smearing in @xmath99 and @xmath6 . here
@xmath2 is fixed to @xmath172 .
a comparison with the similar fig . [ fig : deltam223 ] , where no resolution functions have been included , shows immediately that the effect of changing the octant of @xmath0 remains roughly the same as before ( about 1@xmath51 maximum deviation in each bin ) while the oscillation pattern itself gets smeared away due to finite resolution functions so that the minima and maxima of the oscillations are not as clearly visible , especially beyond the first minimum .
it is seen that the peaks and troughs are better defined when smearing in @xmath6 rather than @xmath55 is used .
however , the differences between smearing in @xmath6 and @xmath55 otherwise seem marginal . hence the effect of detector resolution
is akin to that of increasing @xmath2 , which moves both the minima and maxima of the peaks towards the average value of 1/2 .
this can be seen in fig .
[ fig : deltam213r ] which shows the variation of the up / down events , with resolution , at @xmath209 ev@xmath73 , for different values of @xmath2 .
it is clear , especially when compared with the corresponding results _ without _ resolution function in fig .
[ fig : deltam213 ] , that the distinction between the curves for different @xmath2 is much reduced . in particular , it is likely that other experiments , for example , double - chooz , will be able to resolve @xmath2 with better accuracy .
we now go on to study the effect of finite detector resolutions on the extraction of oscillation parameters .
we repeat the earlier calculation , where contours of allowed parameter space were generated for an input data of up / down neutrino and anti - neutrino event rates in fixed @xmath41 bins for a set of input values for @xmath210 and focus , as before , on the octant resolution .
we find that for resolution widths @xmath211 15% in both @xmath99 and @xmath6 , atmospheric neutrino ( and anti - neutrino ) data accumulated over 1000 kton - years is sufficient to distinguish a non - maximal solution from a maximal solution as well as the octant of @xmath0 for @xmath212 and @xmath213 or @xmath214 .
however , the inclusion of finite detector resolution severely degrades the errors on the parameters .
for instance , the error on @xmath48 is twice as large as that with an ideal detector .
this is shown for two values of @xmath2 , @xmath215 , for @xmath216 ( @xmath217 ) and @xmath218 ev@xmath73 in figs .
[ fig : mdelta7_39 ] and [ fig : mdelta8_39 ] respectively . while maximality in @xmath0 can be excluded and the octant distinguished at 95% cl , only deviation from maximality can be established at 99% cl for the smaller @xmath219 due to the island around the `` wrong octant '' solution .
it may be mentioned here that a similar analysis has been performed in ref .
@xcite . here
a slightly lower neutrino energy was allowed , down to @xmath220 gev , although only dis interactions were considered . also
, the relevant results in this paper focus on discriminating the `` right octant '' solution from the `` wrong octant '' one .
however , one must exercise caution in such an analysis .
it is possible , as is very clearly visible in several of the figures showing allowed parameter space with `` wrong octant '' islands , that while the exact `` wrong octant '' solution may be disallowed , a small neighbourhood of this point may still remain allowed .
in fact , it is typically seen that the allowed island at 99% cl around the `` wrong octant '' solution is typically to the right of the central value . from figs .
[ fig : mdelta7_39 ] and [ fig : mdelta8_39 ] it is seen that a larger @xmath2 is instrumental in suppressing the `` wrong octant '' solutions but does not shrink the allowed parameter space around the correct @xmath0 value .
in fact , the allowed parameter space around @xmath216 is virtually the same in both figures . with larger @xmath86 ,
it is possible to obtain octant discrimination for larger @xmath221 , but the issue of maximality can only be settled at little less than 99% cl in this case , as can be seen from fig .
[ fig : mdeltar9 ] .
( of course , if deviation from maximality can not be established , the octant determination appears to have no meaning .
what we mean here is that the allowed parameter space lies mostly in the first octant .
the octant mirror of the best - fit point is ruled out , but not the maximal value .
we will refer to such solutions as discriminating the octant but not deviations from maximality . )
it turns out that both octant and maximality discrimination can be established at 95% cl for values of @xmath193 15% away from the maximal value .
contrast this with the result ( 10% ) of the previous section when an ideal detector is used .
it is clear that resolution functions worsen the precision with which the magnitude of @xmath48 and @xmath0 can be determined but do not substantially alter the sensitivity to the octant of @xmath0 .
that is , @xmath219 ( @xmath222 ) remains the limit at which matter effects are substantial enough to determine the octant of the ( 23 ) mixing angle ( as also the mass ordering of the ( 23 ) mass eigenstates , as discussed in ref .
@xcite ) , even on including finite resolution functions , while the actual value of the ( 23 ) angle which simultaneously allows maximality and octant discrimination at 99% cl level moves marginally from @xmath221 in the absence of finite resolution effects , to @xmath216 ( @xmath223 ) at the lower limit of @xmath2 .
we note that the situation is not as clean when the true value of @xmath0 lies in the second octant .
the muon survival probability in this case is not as well separated from that for maximal @xmath0 , unlike when @xmath0 is in the first octant ( see fig .
[ fig:23 t ] ) .
hence , while results for octant discrimination will be symmetric with the case when @xmath0 is in the first octant , the issue of maximality may not be so easily settled .
for example , with other parameters as before , and using @xmath224 , which is the octant mirror of @xmath216 , the `` data '' discriminate between the right and wrong octant at 99% cl , but can not discriminate the true value of @xmath0 from maximality . by going farther away from maximality , for example , at @xmath225 ( and @xmath226 values of @xmath227 ev@xmath228 ) , as before ) , we once again have maximality discrimination . the oscillation probabilities and
hence up / down events ratio in anti - neutrinos with inverted hierarchy are very similar to those in neutrinos with the normal hierarchy .
however , the errors are about two and a half times as large due to the smaller anti - neutrino cross - sections leading to correspondingly smaller event rates .
because of this , a 1000 kton - year exposure is inadequate to address the question of maximality or octant determination of @xmath0 . both can be determined at 99% cl for an ideal detector for @xmath229 .
when resolution functions are included , only maximality can be established ; it is not possible to determine the octant of @xmath0 for any value of @xmath0 allowed by the super - kamiokande results ( see table [ tab : bestfit ] ) . in this ,
our results differ from those of ref .
@xcite . by itself , this result is not surprising ; otherwise it will imply that 500 kton - year exposure should be adequate for such determinations with normal hierarchy , which is not the case .
neutrino oscillation studies are now moving to the precision determination of oscillation parameters . the most interesting questions in this field today centre around the issue of whether the ( 13 ) mixing angle is different from zero and its implications .
the impact of this fundamental parameter and others were analysed in the first part of this paper .
in particular we find : * @xmath2 sensitivity : with @xmath2 nonzero , for energies beyond 3 gev , there is a sharp discontinuity in the survival as well as conversion probabilities , @xmath230 and @xmath36 , at the mantle - core boundary at nadir angle @xmath231 .
( see fig .
[ fig:13 t ] and [ fig : e13 t ] ) . for small path lengths ( within the mantle )
increasing @xmath2 simply decreases ( increases ) the amplitude of @xmath35 at the maxima ( minima ) without significantly altering the oscillation period .
+ however , as the paths cross the mantle - core boundary and propagate within the core , the matter effects significantly modify the period of oscillations as well , for both the survival and conversion probabilities . in particular , the jump in @xmath36 at the mantle - core boundary becomes larger and more visible with increasing @xmath2 .
( see fig .
[ fig : e13 t ] ) .
this important feature can be used with long baseline neutrinos for earth tomography studies . * @xmath48 sensitivity : a variation in this parameter changes the period of oscillations , even in the absence of matter effects . for non - zero @xmath2 , with the matter effects turned on , there are dramatic changes in the probabilities , as again both the periods and amplitudes are dramatically altered , especially at around 5 gev .
( see fig .
[ fig:32 t ] ) . at around 2 gev
, a substantial @xmath232 at small nadir angles indicates a rather small @xmath48 , @xmath233 ev@xmath73 .
( see fig .
[ fig : e32 t ] ) . * @xmath0 sensitivity : for @xmath2 nonzero , we note that as @xmath0 increases from the first octant through to the second octant , the survival probability systematically decreases for most nadir angles in the energy range 510 gev .
( see fig .
[ fig:23 t ] ) .
however , the conversion probability , @xmath232 systematically increases as @xmath0 increases at all nadir angles and all energies .
this is not very significant for atmospheric neutrinos , but can be studied separately with long baseline neutrinos .
finally , the probabilities are sensitive to variations in the prem profile ( within allowed limits ) as also the cp phase . however , these dependences are rather small and insignificant and are unlikely to be measured with atmospheric neutrino studies of the type discussed here . the results discussed for cp phase in this paper can therefore be seen only with long baseline neutrinos when all other parameters are presumably known to good precision .
the probabilities @xmath234 are then used to analyse the atmospheric muon neutrino events to determine the deviation from maximality and/or the octant of the ( 23 ) mixing angle @xmath0 .
such dependence arises purely from matter effects as the neutrinos propagate through earth .
since matter effects are different in the case of neutrinos and anti - neutrinos , we assume a detector with charge discrimination capability , such as the proposed ical / ino detector , to heighten the sensitivity to such matter - dependent effects .
moreover , all matter terms are proportional to @xmath5 so we assume that this ( 13 ) mixing angle is different from zero .
we study the up / down events ratios , where by up ( down)-neutrinos , we mean neutrinos arriving at the detector from below ( above ) , so their nadir angles range from 0@xmath235 ( 90@xmath236 ) .
analysis of ratios of events rather than events themselves significantly reduces errors from overall normalisation of fluxes and cross - sections ; the former can be as large as 30% and the latter around 10% .
smearing in energy @xmath99 and base - length @xmath6 by use of gaussian functions is used to simulate finite detector resolutions .
typical relative widths of 15% are used for both energy and base - length . the allowed parameter space obtained on fitting up / down neutrino and anti - neutrino events ratios in any two of @xmath237
is considered .
since the up / down events ratio has very different dependence on @xmath0 and @xmath2 , in principle , they can be simultaneously determined from such studies .
the proposed detectors such as ical / ino with charge identification capability may be in principle suitable for determination of these parameters . in practice , however , these detectors are not well suited for precision determination of @xmath2 with atmospheric neutrinos .
a precise measurement of this parameter is likely to be made , for instance , by the double - chooz experiment .
we summarise below our observations on the simultaneous determination of @xmath238 and @xmath0 from an analysis of the event rates at these detectors with focus on the determination of @xmath0 .
we distinguish three scenarios .
the best case is when both deviation of @xmath0 from maximality as well as its octant can be established . in some cases ,
a maximal value of @xmath0 is disallowed , so deviations from maximality can be established but the octant discrimination may not be possible . the third is where the _ best - fit _
value lies in one of the octants and the region in parameter space around the mirror to the best - fit value in the other octant is ruled out at the precision under consideration ; however , the maximal value is not ruled out at this precision .
in such a case , the allowed parameter space still lies mostly in one octant and hence we consider that we have octant but not maximality discrimination . *
if @xmath2 is known precisely in the near future , as is likely from double - chooz , then in the energy range @xmath16910 gev , where matter effects are largest , the data from ical / ino will be able to study deviations of @xmath0 from maximality as well as determine the octant of this angle , provided @xmath239 . *
it must be emphasised that the octant discrimination is more easily done than establishing deviation from maximality for larger @xmath2 and the reverse is true for smaller @xmath2 when islands of allowed parameter space begin to appear near the `` wrong octant '' solution .
also , values of @xmath0 in the first octant are more easily distinguished from maximality than those in the second octant .
this result is in contrast to those obtained in ref .
@xcite . * in particular , deviations from maximality and octant discrimination to 99% cl can be obtained if @xmath240 and @xmath241 .
these studies used standard gaussian resolution functions with widths @xmath242 15% to smear the events . similarly , we must have @xmath243 for corresponding results in the second octant .
* in an earlier paper@xcite , it was also shown that the same processes are also eminently suited to determine the ( 23 ) mass ordering .
both these determinations need large exposures of roughly 1000 kton - years as well as , crucially , the charge discrimination . * as discussed in @xcite , the issue of determining the ( 23 ) mass ordering and hence establishing the neutrino mass hierarchy is best done using the difference asymmetry defined in eq .
[ eq : da ] which is the _ difference _ of the up / down events ratios with neutrinos and anti neutrinos .
for the octant discrimination that is the crux of this paper , the relevant observable is in fact the _ sum _ of these two ratios .
* the same experiment can therefore study both these questions , while requiring large exposures , 1000 kton - years , in both cases .
hence , studies of neutrino oscillations with atmospheric neutrinos , while being difficult , are probably the only source of precision measurements , at least until very large megaton detectors and/or neutrino factories become a reality .
even so , it is important to point out that the determination of the hierarchy is likely to be relatively easier than the octant determination .
this is because , as seen from fig .
[ fig : lebin ] , the octant effect rides on the hierarchy issue as a sub - dominant effect . another way to see this is that the octant determination is dominated by the ( anti ) neutrino events ratio in the ( inverted ) normal hierarchy , which are the relevant sectors where matter effects dominate .
since the statistics is smaller by half for anti - neutrinos for the same exposure , the significance of the results deteriorates in the case of inverted hierarchy .
the bulk of the results presented in this paper therefore pertain to the normal mass hierarchy .
results of similar significance can only be established with the inverted mass hierarchy if the exposure is at least twice that for normal hierarchy .
hence , such detectors , with exposures of 1000 kton - year or so , may be able to settle the issue of mass hierarchy but , if the ordering is inverted , the issue of maximality of @xmath0 may remain an open question .
* acknowledgements * : we are grateful to dave casper for making the nuance software freely available , and answering a long list of questions on its use .
the work of n.s . was partially supported by the department of science and technology , india .
one of the earliest analytical methods for the calculation of oscillation probabilities with variable matter density is in ref .
the oscillation probability was derived assuming matter to be made up of a series of slabs through which neutrinos and anti - neutrinos propagate .
each slab has smoothly varying density , but the density itself has discrete jumps at the junction of adjacent slabs . in a recent paper , akhmedov et al .
@xcite , have provided an excellent collection of approximate analytic formulae for the neutrino oscillation probabilities .
exact analytical formulae have also been derived in the case of vacuum @xcite and for matter with constant density slabs @xcite .
though complicated , exact analytic formulae have been derived for non - uniform density with linear @xcite and exponentially @xcite varying density .
approximate analytic formulae with varying assumptions have been derived in a number of papers @xcite . here , we use a standard runge - kutta solver to numerically propagate the neutrinos through earth s matter , using the prem earth density profile for a spherical equivalent earth with radius @xmath244 km .
we outline some details about the numerical calculation .
we also highlight how results from an algorithm using constant density slabs differ from those with the prem profile .
our technique is to numerically evolve the flavour eigenstates using the equation , @xmath245 = \by{1}{2e}\,\left ( u m^2 u^\dagger + { \cal a } \right ) \left [ \nu_\alpha\right]~ , \eqno(6)\ ] ] where @xmath246 is the diagonal matrix , diag@xmath247 and @xmath248 $ ] denotes the vector of eigenstates , @xmath249 , @xmath250 . as usual
, @xmath15 is the mns mixing matrix defined in eq .
[ eq : mns ] , @xmath251 is the mass - squared matrix with the diagonal piece proportional to @xmath252 removed : @xmath253 diag@xmath254 .
note that in a constant density slab approximation , the mass eigenstates are propagated , including the `` sudden '' jumps across the density discontinuities . as mentioned in the text ,
the chief difference between the constant density approximation and the prem profile lies in the resonance effects .
this can be seen from fig .
[ fig : profile ] where the results of using the prem profile in our calculations , and the nuance @xcite result for the same parameters , using constant density slabs , is shown .
here we have set the parameters @xmath38 , @xmath48 ( normal hierarchy ) , @xmath255 and @xmath0 to their best - fit values as given in table [ tab : bestfit ] and set the cp phase to zero .
we use @xmath86 . for a low value of @xmath256 gev , where resonance occurs in the core
, the constant density slab of nuance does not show resonance since the precise density value needed for resonance does not occur .
of course , matter effects are present and are large ; however , this difference causes the two curves for the neutrino survival probability to begin to separate in this region , as can be seen from the top left panel of the figure .
it is seen that the anti - neutrino probabilities , which are relatively insensitive to matter effects in the normal hierarchy , however , match exactly . at larger energies , @xmath257 ,
the `` missed '' resonance occurs in the mantle itself .
hence , deviations between the two curves are seen at almost all zenith angles .
the difference can be up to 1015% .
note also that the prominent effect of core - crossing , which occurs at @xmath231 , can be seen clearly in both curves .
again , the anti - neutrino probabilities match .
we have also checked that , by taking smaller and smaller slabs , and approximating the prem profile as closely as possible , the results from the constant density slab approach those using the exact runge - kutta solver with prem profile . however , the resultant increase in number of slabs is quite large .
ahmed et al .
( sno collaboration ) , phys .
lett . * 92 * , 181301 ( 2004 ) [ arxiv : nucl - ex/0309004 ] ; q.r .
ahmad , et al .
( sno collaboration ) , phys .
lett . * 89 * , 011302 ( 2002 ) ; phys .
lett . * 89 * , 0113010 ( 2002 ) , phys . rev . lett .
* 87 * , 071301 ( 2001 ) ; s. fukuda , et al .
( super - kamiokande collaboration ) , phys .
* b 539 * , 179 ( 2002 ) , phys .
lett . * 86 * , 5656 ( 2001 ) ; phys .
86 * , 5651 ( 2001 ) ; fukuda , y. et al .
( super - kamiokande collaboration ) , phys .
lett . * 82 * , 2430 ( 1999 ) ; phys .
82 * , 1810 ( 1999 ) ; phys .
lett . * 81 * , 1158 ( 1998 ) ; w. hampel , et al .
( gallex collaboration ) , phys .
* b 447 * , 127 ( 1999 ) ; j.n.abdurashitov , et al .
( sage collaboration ) , phys .
lett . * 83 * , 4686 ( 1999 ) ; k.s .
hirata , et al .
( kamiokande - ii collaboration ) , phys . rev . *
d 44 * , 2241 ( 1991 ) ; r. davis , ( homestake ) prog .
. phys . * 32 * , 13 ( 1994 ) .
y. ashie et al .
( super - kamiokande collaboration ) , phys .
rev . lett . * 93 * , 101801 ( 2004 ) ; m. ishitsuka ( super - kamiokande collaboration ) , noon2004 , the 5th workshop on `` neutrino oscillations and their origin '' , 11 - 15 february 2004 , tokyo , japan .
mohan narayan , g. rajasekaran , rahul sinha , mod .
. lett . * a 13 * , 1915 ( 1998 ) ; see also g. rajasekaran , pramana * 55 * , 19 ( 2000 ) ( plenary talk given at workshop in high energy particle physics-6 , chennai , india ) .
a. para , acta phys . polon . * b 31 * , 1313 ( 2000 ) [ arxiv : hep - ph/0005012 ] ; c. andreopoulos , p. stamoulis and g. tzanakos , numi - note - atm@xmath259nu-990 , ua/ phys/ hep/ 11 - 05 - 2003 ( http://www-numi.fnal.gov/ ) . | we study in detail the sensitivity of neutrino oscillation probabilities to the fundamental neutrino parameters and their possible determination through experiments . the first part of the paper is devoted to the broad theme of isolating regions in the neutrino ( and anti - neutrino ) energy and propagation length that are sensitive to the oscillation parameters .
such a study is relevant to neutrinos both from the earth s atmosphere or from a neutrino factory . for completeness
we discuss the sensitivity , however small , to the parameters involved in a three - generation framework , and to the earth matter density profile .
we then study processes relevant to atmospheric neutrinos which are sensitive to and allow precision measurements of the mixing angle @xmath0 and mass - squared difference @xmath1 apart from the mixing angle @xmath2 .
crucial to this analysis is charge identification ; detectors having this capability can isolate these matter effects .
in particular , we address the issue of using matter effects to determine whether the mixing angle @xmath0 is maximal , and , if not , to explore how well its octant can be determined .
when realistic detector resolutions are included , we find that deviations of about 15% ( 20% ) from a maximal value of @xmath3 can be measured at 95% ( 99% ) cl ) provided @xmath2 is non - zero , @xmath4 , and the neutrino mass ordering is normal , with fairly large exposures of 1000 kton - years . |
thermoelectric material is a functional material that converts heat into electricity and vice versa through the seebeck and peltier effects @xcite .
high - efficiency thermoelectric material is of particular importance because of its potential application for energy - conversion technology using waste heat @xcite .
although the thermoelectric conversion efficiency is insufficient at present , various important concepts to improve the efficiency have been recently proposed @xcite .
the fundamental route for efficient thermoelectrics is to increase the dimensionless figure of merit @xmath0 , where @xmath1 , @xmath2 , @xmath3 , and @xmath4 are the seebeck coefficient , electrical conductivity , thermal conductivity , and absolute temperature , respectively . in conventional semiconductors ,
the @xmath5 value ( power factor ) is maximized with optimal carrier concentration of the order of 10@xmath6 @xmath7 owing to a trade - off relation between @xmath1 and @xmath2 in terms of carrier concentration @xcite , and the thermal conductivity is mainly governed by the lattice contribution @xmath8 near room temperature .
thus , optimization of the carrier concentration is a basic step to improve the thermoelectric efficiency , which has been typically achieved by chemical substitutions for parent semiconductors .
it was recently shown that an electric - field application to field effect transistor structure on a thermoelectric material is also an effective tool to control the concentration and improve the thermoelectric property @xcite .
the induced two - dimensional electron gas exhibits unusually large seebeck coefficient , which might be attributed to an enhanced density of states due to the low - dimensional structure @xcite .
the light illumination may be a simple and useful way to optimize the carrier concentration and improve the effciency .
the conduction phenomenon attributed to the excited carriers by the light illumination is known as the photoconduction .
likewise , the seebeck effect contributed from such photo - induced carriers is called `` photo - seebeck effect '' , which was first measured in conventional semiconductor ge by tauc in 1955 @xcite . in 1970s ,
the photo - seebeck effect has been explored in several semiconductors si , gaas , and cds @xcite .
these materials exhibit finite photo - response for the seebeck coefficient as well as the conductivity . the light illumination for ge increases @xmath2 and decreases @xmath9 , naively understood within the photo - doping effect @xcite .
on the other hand , silicon exhibits nontrivial photo - seebeck effect near room temperature : both @xmath2 and @xmath9 are increased by light illumination , which can not be understood within a simple model @xcite .
the elucidation of the mechanism of such unusual photo - transport behaviors possibly provides a new idea for efficient thermoelectrics .
note that the photo - seebeck effect measured with different photon energies might be a powerful tool to investigate detailed electronic band structure as well @xcite . however , there are only a few reports regarding the photo - seebeck effect , and the thermoelectric transport nature of the photo - induced carriers is poorly understood so far . here ,
we report first investigation of the photo - seebeck effect in the wide - gap oxide semiconductor zno ( band gap of @xmath10 ev ) , which is known as a transparent conducting oxide widely used for applications @xcite .
aluminum - doped zno is investigated as a potential thermoelectric oxide as well @xcite .
we construct the photo - seebeck measurement system , and observe the strong variations of the seebeck coefficient and the electrical conductivity under the ultraviolet light illumination , whose photon energy exceeds the band - gap energy of zno . to analyze the observed data , we utilize a simple parallel circuit model and determine the contributions from the photo - induced carriers existing only near the illuminated sample surface .
our results demonstrate that the distinct variations of transport properties originate from the changes of both carrier concentration and mobility with illuminating light , that is discussed within the framework of conventional semiconductor physics . ,
( b ) the thermoelectric voltage @xmath11 , and ( c ) the temperature difference @xmath12 . the voltage and temperature were measured with several heater - power applications .
the illumination was performed with the ultraviolet led ( @xmath13 ev ) .
( d ) the schematic experimental setup for the photo - seebeck measurement . ]
we construct the photo - transport measurement system inside the physical property measurement system ( ppms ) .
the experimental configuration is schematically shown in fig .
1(d ) , which is based on ref . .
we measure the resistivity and the seebeck coefficient under light illumination with green ( @xmath14 ev ) and ultraviolet ( uv ) ( @xmath13 ev ) light - emitting diodes ( leds ) .
the electric contacts are covered with aluminum foils to prevent an extrinsic photo - voltaic effects around such regions .
the electrical resistivity was measured with the two - wire method .
the seebeck coefficient was measured with standard steady - state method .
the temperature difference @xmath12 between hot and cold sides , which is made by applying current @xmath15 to the resistive heater on the hot side , is determined from the thermoelectric voltage of the bridged constantan wire .
all measurements has been performed at @xmath16 k. we used zno single crystals ( typical dimension of 5@[email protected] mm@xmath18 ) purchased from a materials company ( mti corporation ) . led illumination was done to the @xmath19 plane .
we firstly describe our experimental procedure for the photo - seebeck measurements .
we measure the thermoelectric voltage @xmath11 and the temperature difference @xmath12 under illuminations with several different light intensities controlled by the excitation current for led @xmath20 .
figures 1(a - c ) display the measurement time dependence of @xmath20 , @xmath11 , and @xmath12 , respectively . here
we show the results for the uv led illumination with five different light intensities . to eliminate a constant voltage due to a photo - voltaic effect from inhomogeneities inside the crystal , we measure @xmath11 with several different @xmath12 by applying different heater currents @xmath15 . from these data ,
we obtain the seebeck coefficient @xmath21/[\delta t(i_{\rm heater } = 20 \mbox { ma } ) - \delta t(17 \mbox { ma})]$ ] . as shown in fig .
1(b ) , the thermoelectric voltage @xmath11 is significantly varied with led light illumination , change of which from the dark value increases with increasing @xmath20 .
we note that the temperature difference @xmath12 and the averaged temperature between hot and cold sides are slightly changed by illuminations as shown in fig .
1(c ) , however , such temperature change during the measurement is negligibly small , which can not explain the observed large @xmath11 variations under illuminations .
the electrical resistivity is measured using same experimental configuration without thermal gradient . and ( b ) the seebeck coefficient @xmath1 of zno single crystal .
the excitation current for led illumination @xmath20 is also plotted as a function of the time in the right axis of ( a ) . ]
the photo - transport properties of zno obtained from above procedures are plotted in figs .
2(a ) and 2(b ) . figure 2(a ) shows the resistivity variation measured under five different light intensities indicated as @xmath20 in the right axis .
the uv led illumination considerably decreases the resistivity , whereas the green light with lower photon energy negligibly affects the resistivity , indicating that the electron - hole pairs are created by illuminating uv light across the band gap of zno and contribute the conduction .
figure 2(b ) displays the seebeck coefficient measured under illumination .
we observe a significant decrease of @xmath9 with uv illumination , while the green one shows no influence for @xmath1 , consistent with the photoconductivity results shown in fig .
2(a ) . and electrical conductivity @xmath2 .
( b ) the photo - induced seebeck coefficient @xmath22 as a function of the photo - induced electrical conductivity @xmath23 ( the jonker plot for the photo - induced transport properties ) .
the dotted line shows the slope of @xmath24 .
the inset schematically illustrates our parallel - circuit model composed of light - penetrated and non - penetrated regions . ]
figure 3(a ) depicts the relation between the seebeck coefficient @xmath1 and electrical conductivity @xmath2 measured under the uv led illumination . at dark
, this sample exhibits an @xmath25-type conduction ( @xmath1 = -550 @xmath26v / k ) , which might be attributed to impurity hydrogen @xcite . under uv illuminations , similar to earlier studies on ge @xcite ,
the conductivity is increased and the seebeck coefficient is decreased in magnitude as expected in the photo - doping effect . as discussed in ref .
, this @xmath1 variation should be related to the difference of mobilities of valence and conduction bands since the illumination creates electron - hole pairs .
we also note that the absorption coefficient of zno is about @xmath27 @xmath28 at @xmath29 = 3.4 ev @xcite , indicating that the illuminated uv light can penetrate into very thin region ( @xmath30 nm ) from the surface .
we then apply a simple parallel - circuit model composed of thin conducting and thick insulating layers [ the inset of fig .
3(b ) ] in order to analyze the observed photo - transport properties . the measured electrical conductivity @xmath2 and seebeck coefficient @xmath1 are expressed by @xcite @xmath31 @xmath32 where @xmath33 is the sample thickness ( 0.5 mm ) and @xmath34 is the penetration depth ( 50 nm ) . here
@xmath35 ( @xmath36 ) and @xmath37 ( @xmath38 ) are the electrical conductivities ( the seebeck coefficients ) measured at dark and induced by the light illumination , respectively .
now @xmath39 , then the photo - induced transport components @xmath23 and @xmath22 are @xmath40 @xmath41 figure 3(b ) shows the photo - induced seebeck coefficient @xmath22 as a function of the photo - induced electrical conductivity @xmath23 obtained from eqs .
( [ sig0last ] ) and ( [ s0last ] ) ( the jonker plot for the photo - induced transport properties ) .
the dotted line shows the slope of @xmath24 , which is expected in doped semiconductors @xcite .
the obtained @xmath22-@xmath23 behavior roughly coincides with this slope , implying that the thermoelectric transport property of photo - induced carriers can be discussed in terms of conventional semiconductor physics as well . and
( b ) mobility @xmath26 of the photo - induced carriers as a function of the excitation current for led illumination @xmath20 .
arrows indicate the values measured at dark . ]
we next show the photo - induced carrier concentration @xmath25 , which can be obtained from the photo - induced seebeck coefficient @xmath22 . in the degenerate regime ,
the seebeck coefficient is expressed by @xmath42 where @xmath43 is the fermi energy for a simple parabolic band ( @xmath44 is the effective mass , which is assumed to be @xmath45 @xcite ) and @xmath46 is a scattering parameter ( @xmath47 for an electron - phonon scattering and @xmath48 for an impurity scattering ) @xcite . in the non - degenerate regime , it is given by @xmath49 , \label{ndeg}\ ] ] where @xmath50 .
figure 4(a ) displays the excitation current @xmath20 dependence of @xmath25 determined with using eqs .
( [ deg ] ) and ( [ ndeg ] ) . here
, we adopt @xmath47 , which is not crucial to the results . the carrier concentration at dark is determined with the non - degenerate model [ eq .
( [ ndeg ] ) ] , yielding @xmath51 @xmath7 , as shown by the arrow in fig .
4(a ) . under illuminations
, @xmath25 is increased from the dark value in both models , and reaches about @xmath52 @xmath7 at maximum light intensity .
this value corresponds to about 0.1% al - doping effect for zno @xcite and is close to the value of the optimized carrier concentration in conventional thermoelectric materials @xcite .
we have also evaluated the photon number from the output power ( 1.5 mw at @xmath20 = 20 ma ) of the uv led light .
the number of photons is about @xmath53 photons / s .
however , the electron - hole recombination time in zno depends on samples and even a persistent photoconductivity is observed in several samples owing to hole traps in grain boundaries and/or oxygen vacancies @xcite . also , we used a wide - angle led light .
these arbitrary properties make it difficult to quantitatively compare the photo - induced carrier concentration with the number of photons absorbed by the sample .
the mobility @xmath26 of the photo - induced carriers can be evaluated as @xmath54 , shown in fig .
4(b ) as a function of @xmath20 .
in contrast to high - mobility zno samples @xcite , our sample exhibits very low mobility of @xmath55 @xmath56/vs at dark , which is evaluated with using the carrier concentration and the conductivity measured at dark .
the uv illumination considerably increases the mobility roughly by one order of magnitude .
such mobility variations under illumination have been also reported in photo - hall measurements on zno single crystal @xcite , although they measured only transport relaxation process after the illumination . there , the illumination changes both the carrier concentration and the mobility , consistent with our present results . let us discuss more detailed transport mechanism of photo - induced carriers . in zno ,
the conduction phenomena are mainly governed by polaronic electrons in a broad zn 4@xmath57 conduction band , which are supplied from hydrogenic donor . on the other hand ,
the holes in narrow o 2@xmath58 and zn 3@xmath33 valence bands are mostly trapped and negligibly contribute to the transport properties @xcite . under uv illuminations , therefore , only electrons of the photo - induced electron - hole pairs are mobile and contribute to the photoconductivity and photo - seebeck effect .
it is also known that the molecular oxygen in atmosphere significantly affects the transport properties of zno @xcite . at the surface ,
adsorbed oxygen molecules are charged by removing the electrons from the conduction bands , and then tightly bound to the surface .
this oxygen adsorption process thus decreases the conductivity . under illuminations ,
such charged oxygen loses its charge by attracting hole of the photo - induced electron - hole pair .
meanwhile , the electron remains in the conduction bands , leading to an increase of the conductivity .
such effects are usually dominant in thin films and porous materials with high surface - to - volume ratio , but not in single - crystalline bulk materials . in present case
, however , the light - penetrated region is very thin ( @xmath59 nm ) , then such surface effects should be considered . under illuminations ,
the mobility is also increased as well as the carrier concentration , which has been indicated in the photo - hall experiments as well @xcite .
the origin of the enhanced mobility by illumination is an open question . in semiconductors ,
the mobility exhibits a non - monotonic variation as a function of the carrier concentration owing to multiple scattering mechanisms @xcite . in high - concentration region , where the ionized impurity scattering is dominant , the mobility is decreased with increasing carrier concentration @xcite . on the other hand , at low carrier concentrations ,
the mobility is limited by grain - boundary or surface scattering @xcite , even for epitaxial films due to a small amount of mosaic structures @xcite .
this leads to an increase of the mobility with increased carrier concentration , which might be applied in our present results .
the photo - transport measurement using zno samples with different impurity or vacancy levels such as samples annealed in oxygen or hydrogen is a future issue to elucidate the exact mechanism of thermoelectric transport property of photo - induced carriers .
we finally give brief comments on the potential thermoelectrics using light . the light illumination combined with thin film applications may offer a new route toward an efficient thermoelectrics as well as the fundamental understandings of thermoelectric effects induced by photo - excited carriers .
in bulk measurements , we observe the averaged values including large contributions from non - penetrated region . on the other hand ,
if the sample thickness is smaller than the penetration length ( i.e. @xmath60 ) , one can directly observe photo - induced components of the conductivity @xmath23 and the seebeck coefficient @xmath22 .
in present case , the improvement of the mobility with illuminating light directly contributes the @xmath61 enhancement by using thin films .
illumination on semiconductors with large absorption coefficient may produce an unexpectedly large photo - seebeck effect owing to its low dimensionality .
furthermore , light is able to be utilized as a heat source to make a thermal gradient as well , then a light - concentration - type thermoelectric module is possibly an efficient photo - thermoelectrics using thin films .
we have investigated the photo - seebeck effect as well as the photoconductivity in the wide - gap oxide semiconductor zno for the first time .
we observe a significant variation of such transport quantities , which are sensitive to the photon energy of illuminated light .
the ultraviolet light illumination decreases both the absolute value of seebeck coefficient and the resistivity , acting as a photo - doping effect to zno .
a parallel - circuit model is applied to extract the photo - induced contributions from the observed transport data , which is discussed through the jonker analysis .
our results demonstrate that both the carrier concentration and the mobility are increased from dark by illuminations , which can be qualitatively argued within the framework of conventional semiconductor physics .
thin film applications may be efficient for photo - thermoelectrics .
this work was supported by advanced low carbon research and development program ( alca ) from jst and program for leading graduate schools `` integrative graduate education and research program in green natural sciences '' from mext , japan . | we examine how the photo - induced carriers contribute the thermoelectric transport , i.e. the nature of the photo - seebeck effect , in the wide - gap oxide semiconductor zno for the first time .
we measure the electrical conductivity and the seebeck coefficient with illuminating light .
the light illumination considerably changes the seebeck coefficient as well as the conductivity , which is sensitive to the photon energy of the illuminated light . by using a simple parallel - circuit model
, we evaluate the contributions of the photo - induced carriers to the conductivity and the seebeck coefficient , whose relationship shows a remarkable resemblance to that in doped semiconductors .
our results also demonstrate that the light illumination increases both the carrier concentration and the mobility , which can be compared with impurity - doping case for zno .
future prospects for thermoelectrics using light are discussed .
= 10000 |
two distinct mechanisms localize interacting bosons moving in a random environment . either repulsions from local on - site interactions on a @xmath2-dimensional lattice lock the bosons in place or disorder
inhibits tunneling from site to site .
the former insulating state is an incompressible mott insulator ( mi ) , while the second is a compressible bose glass ( bg ) . in this paper , we identify an order parameter and an associated length scale that can describe the vanishing of the compressibility at the mi / bg transition .
the order parameter is the relative variance of the disorder - induced mass distribution .
the key insight leading to this observation is that despite the relevance of the disorder vertex in @xmath3 spatial dimensions , it is possible to identify a mott - insulating fixed point where the mass , corresponding to the mott gap , diverges fast enough such that the relative variance of the mass distribution is renormalized to zero . sufficiently close to the mi , both the mean and the relative variance
diverge , identifying the adjacent phase as an insulating bg , irrespective of the boson filling .
we will show that the length scale underlying this order parameter is the effective distance between rare regions . in their initial treatment of this problem , known as the disordered bose - hubbard ( bh ) model , fisher and colleagues@xcite situated the bg between the mi and superfluid ( sf ) , thereby preventing a direct sf / mi phase transition .
they pointed out however , that it is in principle possible but extremely unlikely that for sufficiently weak disorder the bg phase is completely suppressed at commensurate boson fillings . for more than twenty years , this question remained controversial as simulations and analytical arguments both support@xcite and negate@xcite the possibility of a direct mi / sf transition in the presence of weak disorder . only recently
, strong arguments@xcite based on a mathematical theorem have been presented for the existence of the bg upon the destruction of the mi , thereby precluding a direct transition to the sf .
given this result , it is important to establish precisely how the mi to bg transition obtains . in this paper , we are concerned with the critical theory for the mi / bg transition for strongly interacting bosons on a d - dimensional lattice subject to weak disorder .
of particular interest is the universality of the transition , e.g. the identification of a diverging length scale and the determination of the critical exponent describing the divergence of the inverse compressibility . despite extensive numerical@xcite and analytical@xcite studies of the disordered bh model ,
the nature of the mi / bg transition remains elusive . to be more specific , in the following we consider the disordered bh model in its simplest form , @xmath4 describing bosons with creation and annihilation operators @xmath5 , @xmath6 ( @xmath7 ) which are hopping with amplitude @xmath8 between neighboring sites @xmath9 on a @xmath2-dimensional hyper - cubic lattice and which are subject to an on - site repulsion @xmath10 .
the chemical potential @xmath11 controls the boson filling and disorder enters via random uncorrelated on - site potentials @xmath12 which must be appropriately bounded for the model to exhibit stable mi phases.@xcite since the formation of the mi is a consequence of strong repulsive interactions , the mi / sf transition in the clean system can only be understood within an effective long - wavelength theory which is dual to the original bh model ( [ bhmodel ] ) and derived by a strong coupling expansion around the atomic limit.@xcite the mass in the theory is given by the mott gap which vanishes at the transition and turns negative in the sf . in the latter broken - symmetry state
the finite bosonic order parameter is proportional to the sf density .
although the underlying strong coupling expansion breaks down in the sf where the renormalization group ( rg ) flow is towards weak coupling,@xcite the effective long - wavelength theory can be used to analyze the instability of the mi . since the theory is controlled in the mi state it should serve as a starting point to also analyze the instability towards the formation of the bg in the presence of weak disorder . at the outset
, it would seem that an effective theory in terms of the bosonic order parameter is completely inapplicable to the mi / bg transition since in both insulating phases the mass gap is finite and the sf correlations are short ranged . certainly , the relevant length scale at the mi / bg transition has nothing to do with the sf correlation length but instead should be intrinsically related to the disorder .
such a divergent length scale should exist since the transition is expected to be continuous with a divergent inverse compressibility . in this work , we demonstrate that the order parameter for the mi / bg transition can be extracted from the disorder- averaged replica version of the bosonic order parameter theory .
the relevant quantity is the relative variance of the disorder - induced mass distribution which renormalizes to zero in the mi and diverges in the bg and hence serves as the order parameter for the mi / bg transition .
the corresponding length scale is the distance between rare regions in the bg which causes the breakdown of self - averaging in the system as indicated by a diverging relative variance .
this correlation length is finite in the bg and diverges on approaching the mi . from the analysis of the rg flow of the mean value and relative variance of the mass distribution we obtain the key results that ( i )
the instability of the mi is always towards a bg , in agreement with recent arguments,@xcite and that ( ii ) the mi / bg transitions at incommensurate and commensurate boson fillings are not in the same universality class . in the former case of incommensurate fillings ,
we calculate the critical exponents at 1-loop order : @xmath0 for the correlation length and @xmath1 for the inverse compressibility .
the paper is organized as follows . in sec . [ sec.selfaveraging ] , we argue that the relative variance of the disorder - induced mass distribution serves as the order parameter for the mi / bg transition which vanishes in the mi and acquires a finite value in the bg .
we further identify the corresponding correlation length as the distance between rare regions in the bg which cause a breakdown of self - averaging in the system .
the disorder - averaged replica theory describing the long - wavelength physics at strong coupling is derived in sec .
[ sec.replicatheory ] . in sec .
[ sec.rg ] , the rg equations are reviewed and reformulated in a new set of variables which permits us to extract the scale dependence of the relative variance of the induced random - mass distribution .
the results are presented in sec .
[ sec.results ] . the phase boundary between the mi and the bg
is determined by a numerical integration of the rg equations . for incommensurate boson fillings
, we obtain analytical results for the correlation length and compressibility exponents .
finally , we show that the mi / bg at commensurate boson fillings is in a different universality class . in sec .
[ sec.final ] our main results are discussed .
the mi in the disordered bh model exists in the strong coupling limit . while the disorder must be appropriately bounded for the mi to persist,@xcite the key limit that defines strong coupling
is the ratio @xmath13 , where @xmath10 is the on - site energy and @xmath8 the hopping matrix element . as with all strong coupling problems ,
the natural variables that uncloak the physics are not related straightforwardly to those in the uv - complete hamiltonian .
further , if a critical theory is to correctly describe the destruction of the mi , it should contain the seeds of the bg phase , which fisher and co - workers@xcite argued are due to the physics of rare regions .
consequently , the underlying critical theory might have nothing to do with the bare bosonic propagator and the associated superfluid correlation length but rather a new length scale that is intrinsically related to the disorder .
consider the atomic limit of this problem .
a scaling analysis around this regime@xcite demonstrates that the correlation length , @xmath14 , defined as the length scale beyond which the system encounters rare - region physics is finite in the bg and diverges at the transition to the mi with an exponent of @xmath0 for generic bounded disorder distributions such as the box distribution @xmath15 for @xmath16 and @xmath17 for @xmath18 .
the violation of the lower bound of @xmath19 known as the quantum harris criterium@xcite was attributed to a breakdown of self - averaging@xcite in the bg phase .
that is , the fluctuations are not governed by the central - limit theorem .
several years ago , aharony and harris@xcite showed that a break - down of self - averaging implies that the relative variance of any thermodynamic quantity must be finite . taken in tandem , these results imply that characterizing a transition governed by the physics of rare regions requires two quantities to be finite : 1 ) a finite length scale , @xmath14 , over which a rare region is encountered and 2 ) a finite value of the relative variance of any thermodynamic quantity .
the central idea we advance here is that the nature of the mi / bg transition can be understood using perturbative rg techniques but with a _
new set _ of variables and length scales that contain the information about the rare regions .
the mass @xmath20 in the effective long wavelength theory in the clean system corresponds to the mott gap which renders the correlations of the bosonic order parameter short ranged .
any form of disorder in the microscopic bh model , e.g. the potential disorder @xmath12 following a distribution @xmath21 , will induce an effective disorder distribution @xmath22 of the mass coefficient @xmath20 . since in any insulating phase
the superfluid order parameter must vanish on large scales , in both phases , the mi and the bg , the mean @xmath23 of the distribution must diverge under the rg , @xmath24 .
hence , it can not be used to distinguish between the mi and bg phases . given the general considerations above , the natural quantity to distinguish the bg from the mi is the relative variance @xmath25 of the mass distribution , @xmath26 in the mi phase , the system is self averaging and therefore @xmath27 since relative variances of extensive quantities have to vanish in the thermodynamic limit as a consequence of the central - limit theorem . since the bg is characterized by a breakdown of self averaging , @xmath25 is non - zero in the thermodynamic limit and therefore serves as the order parameter for the mi / bg transition . in this paper ,
break down of self - averaging refers entirely to the fact that @xmath28 .
note , @xmath27 in the mi implies that @xmath14 diverges at the transition ; that is , there is no finite length scale over which rare regions exist .
we point out that a break - down of self averaging , as defined here , has nothing to do with a finite - size effect .
rather , it is a function entirely of whether or not in the _ thermodynamic limit _ , @xmath14 , the distance over which rare regions exist , is finite .
if it is , @xmath25 is necessarily non - zero as we will demonstrate by an explicit calculation . to work out the rg flow of @xmath25
, we will use that the disorder vertex in the disorder - averaged replica theory is proportional to the variance of the mass distribution .
we will proceed by recasting the rg equations in terms of @xmath25 and show that these equations admit a stable mi fixed point at @xmath29 and @xmath30 as well as a critical mi / bg fixed point at @xmath29 and finite @xmath25 .
we point out that such an identification of fixed points is only possible because of the identification of the relevant variables . whereas the standard equations that are used to study this problem only capture a runaway flow away from the mi / sf fixed point of the clean system,@xcite in the new variables we find that the unstable mi / sf fixed point is connected with the new critical mi / bg fixed point by a separatrix .
our analysis will further demonstrate that the mi / bg transition at commensurate fillings belongs to a different universality class .
we analyze the instability of the mi state towards the formation of the bg using the disorder - averaged replica version of the effective long - wavelength theory which is derived by an expansion around the atomic limit of the stable mi state . in this theory , the replica mixing disorder vertex is proportional to the variance of the mass distribution which allows us to extract the rg flow of the relative variance @xmath25 ( [ rel.var ] ) which we have identified as the order parameter for the mi / bg transition . in the absence of disorder , the effective action capturing the long - wavelength physics at strong coupling , @xmath31 , @xmath32 ) and decoupling the off - diagonal hopping terms with a hubbard - stratonovich ( hs ) transformation . in this action , we have rescaled the length and imaginary time @xmath33 to dimensionless units , @xmath34 with @xmath35 the momentum cutoff and @xmath36 .
the precise dependence of the coupling constants on the microscopic parameters is known analytically.@xcite for example , the mass coefficient which corresponds to the mott gap , is given by @xmath37 with @xmath38 , @xmath39 , and @xmath40 the number of bosons per site .
the interaction vertex @xmath41 is proportional to @xmath42 reflecting the underlying strong - coupling expansion around the atomic limit ( @xmath43 ) .
an explicit expression for @xmath41 can be found elsewhere.@xcite at the mean - field level , the mi / sf phase boundary is determined by @xmath44 .
the mi states for different fillings @xmath40 are characterized by a finite mott gap , @xmath45 , and @xmath46 .
the sf obtains for @xmath47 where the sf density is proportional to @xmath48 . as a consequence of the @xmath49 symmetry of the fields,@xcite the temporal gradient terms
are given by @xmath50 and @xmath51 and are therefore related to the slope and the curvature of the mean - field phase boundary .
at the tips of the mi lobes at values @xmath52 of the chemical potential where the boson filling is commensurate with the lattice and the mi states are most stable , @xmath53 , and the critical theory is characterized by a dynamical exponent @xmath54 . at incommensurate fillings , @xmath55 , corresponding to @xmath56 dynamics .
consider now the disordered case .
the site energies @xmath12 will be uncorrelated and chosen from a bounded distribution , @xmath57 $ ] .
the identical hs transformation can be performed@xcite yielding a dual theory with coefficients given by the expressions for the clean system but with the chemical potential @xmath38 shifted by the random potential @xmath58 on each site , e.g. @xmath59 for the mass term .
this expression is well defined on every site if the disorder distribution @xmath60 is bounded , and the chemical potential @xmath61 lies in the intervals @xmath62 $ ] and @xmath63 .
note that this stability condition ensures that the system exhibits stable mi phases in the presence of disorder .
the procedure for constructing the effective action is now standard : restore translational symmetry by performing the disorder average of the free energy using the replica trick and and take the continuum limit . in the resultant expression , @xmath64}-\frac{\bar{g}}{2}\sum_{\alpha\beta}\int_{{{\bm x}},x_0,x_0'}|\phi_\alpha(x_0)|^2|\phi_\beta(x_0')|^2 , \label{srep}\ ] ] the first term corresponds to @xmath65 identical copies of the disorder- averaged action with @xmath66 the replica indices .
note that also for the disorder - averaged coefficients @xmath67 and @xmath68 .
the replica - mixing disorder vertex arises from the quadratic order of the cumulant expansion and is given by the variance of the mass distribution , @xmath69 as is evident , it is non - diagonal in imaginary time as a consequence of the perfect correlation of the disorder along the imaginary time direction .
since the disorder also couples to the temporal gradient terms , the replica theory contains disorder vertices with additional time derivatives.@xcite these terms turn out to be irrelevant and do not change the universality of the transitions between the mi and the bg and hence are not considered any further . as a consequence of the underlying strong - coupling expansion ,
the variance @xmath70 of the mass distribution is not simply given by the variance of the random - site disorder in the bh model.@xcite this is easily seen by expanding @xmath59 for small @xmath58 yielding @xmath71 with @xmath72 the @xmath65-th moment of @xmath73 . hence , the effective disorder @xmath74 is a continuous function of the chemical potential entering via the coefficients @xmath75 and is strongly suppressed at the tips of the mi lobes where @xmath53 .
therefore , the inclusion of higher moments of the distribution @xmath21 is essential not only to account for the boundedness of the distribution but also to capture the coupling to disorder at commensurate fillings . in order to retain the moments to _ infinite _ order , without loss of generality , we use the discrete distribution @xmath76 corresponding to on - site energies which occur with the same probability @xmath77 increased or decreased by @xmath78 .
the resulting mean @xmath79 and variance @xmath80 of the induced mass distribution entering the replica theory ( [ srep ] ) are given by @xmath81,\\ \overline{g } & = & \overline{[r(x-\epsilon_i / u)-\overline{r})]^2}\nonumber\\ & = & ( 1 - 2p)r^2(x)+p[r^2(x-\delta)+r^2(x+\delta)]-\overline{r}^2,\end{aligned}\ ] ] and contain the moments of @xmath21 to infinite order . the mf phase boundary between the @xmath82 mi and the sf obtained by the condition @xmath83 is shown in fig .
whereas at incommensurate fillings the interaction vertex @xmath84 is irrelevant for @xmath85 , the disorder vertex is always relevant in @xmath3 rendering the mf theory meaningless . in order to determine the stability region of the mi and subsequent phase transitions , we employ the rg .
we proceed with the rg analysis of the effective replica field theory ( [ srep ] ) . from the rg flow of the mean @xmath86 and the variance @xmath74 of the disorder - induced mass distribution we can subsequently determine the scale dependence of the relative variance @xmath87 to analyze the instability of the mi towards the formation of a bg . after successively eliminating modes of highest energy corresponding to momenta from the infinitesimal shell @xmath88 and rescaling of momenta ( @xmath89 ) , frequencies ( @xmath90 ) , and fields ( @xmath91 ) , we obtain to one - loop order@xcite @xmath92 [ rgeqs ] where the scaling dimension @xmath93 of the fields has been determined such that the coefficient of the spatial gradient term remains constant .
further , we have absorbed a numerical factor arising from the shell integration by redefining @xmath94 and @xmath95 with @xmath96 the surface of the @xmath2-dimensional unit sphere .
the frequency integrals are defined as @xmath97 , @xmath98^{-1/2}$ ] , @xmath99 , and @xmath100 with @xmath101 and @xmath102 the on - shell propagator . from the rg equations ,
it is evident that for @xmath3 , the variance @xmath103 increases with the scale which suggests that the rg flow is always towards strong disorder .
this interpretation is inconsistent with perturbative arguments which show that the disordered bh model exhibits stable mi phases in the regime of weak disorder,@xcite and does not take into account that in the insulating phases the mean @xmath86 also diverges under the rg . therefore the question of whether disorder is strong or not can only be answered by analyzing the relative variance @xmath104 . in the stable mi phase ,
disorder is irrelevant and we expect that @xmath27 . realizing that @xmath86 diverges for both insulating phases , we implement the approach of aharony and harris@xcite by focusing on the relative variance @xmath25 of the disorder - induced mass distribution . as we argued in sec .
[ sec.selfaveraging ] , @xmath25 serves as the order parameter for the mi / bg transition characterized by a breakdown of self - averaging .
putting this into practice in the rg equations requires a new set of variables . instead of the mass @xmath86
, we use @xmath105 which corresponds to the on - shell propagator in the static limit , @xmath106 , and asymptotically becomes the inverse mass in the insulating phases where @xmath107 .
in addition , we work with @xmath108 ( @xmath109 ) for the temporal gradient terms and @xmath110 , @xmath111 for the interaction and disorder vertices , respectively . in the following ,
we identify the mi phase by the conditions that @xmath107 but @xmath112 . in the bg , @xmath113 . in the large mass regime
, @xmath114 corresponds precisely to the relative variance @xmath25 of the mass distribution .
we start with a numerical integration of the full set of rg equations ( [ rgeqs ] ) in terms of the variables introduced above to see if we can indeed identify the mi / bg transition .
since @xmath115 away from the tips of the mott lobes , we use a dynamical exponent @xmath56 . note that for small @xmath86 close to the mi - sf transition in the clean system , the spatial gradient term @xmath116 serves as a cut - off in @xmath117 and the corresponding frequency integrals .
indeed we find a regime where the system flows towards a stable mi fixed point @xmath118 with @xmath119 and @xmath120 .
therefore , in the mi phase the mean @xmath86 of the mass distribution diverges fast enough that the relative variance @xmath25 of the distribution renormalizes to zero .
as expected , @xmath121 and @xmath122 renormalize to zero in the mi . using the analytic expressions for the bare coefficients as a function of the microscopic parameters , we obtain the mi phases in the presence of disorder .
1(a ) shows the resulting @xmath82 mi lobe in @xmath123 as a function of @xmath38 and @xmath39 for @xmath124 and @xmath125 . sufficiently close to the mi state
, @xmath114 diverges at a certain scale @xmath126 whereas @xmath127 is practically zero .
this clearly identifies the adjacent phase as an insulating bg ! note that the mass distribution spreads faster than it shifts to infinity .
hence it is the tail of the distribution which destabilizes the mi , indicative of rare - region physics . ) and the bg obtained from numerical integration of the rg equations .
the disorder corresponds to site energies increased or decreased by @xmath128 with probability @xmath124 .
the solid line shows the unphysical mean - field phase boundary ( @xmath129 ) between the mi and sf .
( b ) extraction of the correlation length exponent for different values @xmath61 of the chemical potential as indicated in ( a ) . ] in the following , we determine the phase boundary @xmath130 by bisection and follow the divergence of the correlation length @xmath131 as @xmath132 .
note that the correlation length defined here is determined by the scale @xmath126 where @xmath25 diverges and self - averaging breaks down .
the correlation length @xmath14 therefore corresponds to the distance of rare regions in the bg which cause the breakdown of self - averaging ( see sec .
[ sec.selfaveraging ] ) . sufficiently close to the mi , we expect a power - law divergence , @xmath133 , defining the correlation - length exponent @xmath134 . to extract @xmath134 , in fig .
1(b ) we plot @xmath135 as a function of @xmath136 for different values of @xmath61 .
for @xmath132 , the functions indeed converge to a constant value corresponding to @xmath137 .
interestingly , on approaching commensurate boson fillings , the length scale upon which universal behavior is observed increases and possibly even diverges . whereas an increase might be simply a quantitative effect originating from the suppression of the bare disorder vertex by commensuration , a divergence would indicate a different universality class at the tips of the mi lobes . to better understand the nature of the transitions we proceed with an analytical investigation of the _ incommensurate _ case ( @xmath56 ) .
since for @xmath85 the interaction vertex @xmath84 is irrelevant at the mi - sf transition of the clean system , in the vicinity of the fixed point @xmath138 ( @xmath120 , @xmath139 ) we can simplify the rg equations by setting @xmath140 .
the resulting rg equations for @xmath141 and @xmath114 , @xmath142 [ rgeqsic ] are decoupled from @xmath121 .
our analysis of these equations is strictly in the perturbative regime where @xmath143 .
the divergence of @xmath141 which obtains for @xmath144 is spurious and outside the range of validity of the perturbative rg equations .
in addition to @xmath138 and the stable mi fixed point @xmath145 , the rg equations ( [ rgeqsic ] ) exhibit an additional fixed point @xmath146 , which will turn out to be critical fixed point determining the universality of the mi - bg transition at incommensurate fillings .
from linearization around @xmath147 we find @xmath148 indicating that the interaction vertex remains irrelevant . for @xmath120
we obtain a trivial flow equation for @xmath141 capturing the mi - sf of the clean system . for @xmath149 ,
the flow is towards @xmath118 , while for @xmath150 ( @xmath151 ) , @xmath152 signaling the instability toward the formation of a sf . for @xmath3
, @xmath138 is unstable for small disorder , @xmath153 .
however , instead of a runaway flow , we find a separatrix @xmath154 which connects @xmath138 and @xmath147 as shown in the case @xmath123 in fig .
linearizing around the fixed - points , we find that asymptotically @xmath155 in the vicinity of @xmath138 and @xmath156 close to @xmath147 , respectively . for initial values
@xmath157 and @xmath158 , the system flows towards the mi fixed point , whereas for @xmath159 , @xmath160 will eventually diverge . for initial values very close to the mi phase , the trajectory @xmath161 will track the separatrix and consequently , sufficiently close to the transition
, the divergence of @xmath160 is controlled by the fixed point @xmath147 .
linearizing around this point , we find @xmath162 with @xmath163 the infinitesimal distance to the separatrix . from the condition @xmath164 we obtain the correlation length @xmath165 , implying a correlation length exponent of @xmath0 , in perfect agreement with the value extracted from numerical integration of the full set of rg equations for @xmath123 ( see fig .
1(b ) ) and a previous real - space rg and scaling analysis.@xcite and ( b ) @xmath166 for incommensurate boson fillings as a function of @xmath141 and @xmath114 corresponding ( in the limit of large mass @xmath86 ) to the inverse mean and the relative variance of the mass distribution , respectively . ] since @xmath167 at the fixed point and in the bg , the fluctuations are no longer@xcite governed by the central - limit theorem .
this implies that self - averaging breaks down .
consequently , the bound @xmath19 [ ref . ]
no longer applies . that the bg is mediated by rare localized regions
is the efficient cause of this effect . in quantum systems ,
such rare events play a more pronounced@xcite role than in classical systems because of the perfect correlation of the disorder along the imaginary time direction .
it is also the divergence of the relative variance that is responsible for a finite compressibility @xmath168 in the bg . using the asymptotic form of @xmath169 in which only the @xmath23 and @xmath170 terms are retained
, we obtain to leading order @xmath171 which clearly vanishes in the mi .
however , in the bg sufficiently close to the transition , @xmath172 with @xmath1 . for completeness
, we briefly discuss the case @xmath173 in which a small amount of disorder is irrelevant at the mi - sf transition .
the inversion of the rg flow along the @xmath114-direction is a consequence of the presence of an additional unstable fixed point @xmath174 in @xmath175 located on the separatrix ( see fig .
2(b ) ) . therefore
, in @xmath173 the disorder has to exceed a critical value in order to induce a bg phase in contrast to recent claims.@xcite we now focus on _ commensurate _ boson fillings to understand the nature of the transition at the tips of the mi lobes where @xmath176 and @xmath54 .
the smaller dynamical exponent makes the interaction vertex @xmath110 more relevant and in fact marginal at the mi - sf transition of the clean system in @xmath123 .
however , for @xmath177 , @xmath122 remains irrelevant at @xmath138 suggesting that the critical behavior is controlled by the same separatrix as for incommensurate fillings .
interestingly , this is not the case .
a numerical integration of the rg equations shows that in the disordered phase close to the mi , @xmath178 diverges simultaneously with @xmath179 .
this behavior is easily understood upon linearizing around @xmath147 yielding @xmath180 which demonstrates that @xmath122 becomes relevant for @xmath181 .
therefore , for any finite value @xmath182 , the separatrix obtained by projection into the @xmath141-@xmath114 plane does not terminate in @xmath147 but diverges as @xmath183 .
consequently , the mi - bg transition at commensurate fillings is not controlled by @xmath147 but by a different fixed point which is not accessible in the present calculation . however , from the divergence along the seperatrix , @xmath184 we estimate @xmath185 which is significantly larger than the incommensurate value and satisfies the bound @xmath19.@xcite
to summarize , we have analyzed the instabilities of the mi state of the weakly disordered bh model within an rg analysis of the disorder - averaged replica theory valid in the strong - coupling regime .
our analysis shows that the mi always becomes unstable towards a bg regardless of the boson filling , in agreement with recent work@xcite which , based on a mathematical theorem , excludes the possibility of a direct mi / sf transition in the presence of disorder .
while in both insulating phases the mass coefficient diverges on large scales , rendering sf correlations short ranged , the relative variance of the induced random mass distribution renormalizes to zero in the mi and diverges in the bg .
this allowed us to demonstrate that the relative variance serves as the order parameter for the mi / bg transition .
we further extracted the diverging correlation length of the mi / bg transition from the scale where the relative variance diverges under the rg .
since the finite relative variance of an extensive quantity signals a breakdown of the central - limit theorem,@xcite the correlation length corresponds to the separation between rare regions in the bg , causing a breakdown of self - averaging in the system .
we point out that the correlation length defined here has nothing to do with the sf correlation length which remains finite over the mi / bg transition .
the reformulation of the rg equations in terms of new variables , in particular the relative variance , allows us to identify fixed points and to subsequently analyze the universality of the mi / bg transition . that new variables , not apparent in the uv - complete theory , are necessary to access the fixed point is typical of strong - coupling problems . at incommensurate fillings , we find a separatrix which connects the mi / sf fixed point in the clean system with the critical mi / bg fixed point in the weakly disordered system .
the correlation - length and compressibility exponents characterizing the mi / bg transition are determined as @xmath0 and @xmath1 , respectively . in principle , the power - law divergence of the inverse compressibility could be verified either experimentally or numerically provided that the critical region can be accessed sufficiently .
the violation of the strict lower bound @xmath19 known as the quantum harris criterium@xcite has been argued some time ago@xcite to be a key signature for the breakdown of self averaging .
interestingly , our correlation - length exponent @xmath0 is identical to the exponent estimated by a scaling analysis in the atomic limit for a box distribution of site energies.@xcite our analysis shows that the mi / bg transition at commensurate fillings is controlled by a different fixed point which is not accessible within the present rg approach .
the two distinct transitions found here are , in retrospect , not out of the ordinary , since in the clean system , the mi / sf transitions are in the universality class of the @xmath186-dimensional xy model@xcite and of mean - field type@xcite for commensurate and incommensurate fillings , respectively .
this dichotomy persists in the disordered system as well .
what this analysis indicates is that within a replica theory , instabilities driven by local rare regions , indicative of griffiths singularities , are completely accessible if one focuses on the scaling dependence of the relative variance of the mass distribution .
since the relative variance of any extensive quantity acquires a non - zero value @xcite once self - averaging breaks down , the method presented here can in principle be applied to any transition driven by local rare - region physics .
* acknowledgment : * we are grateful for stimulating discussions with e. fradkin , a. j. leggett , and a. g. green and we thank the nsf dmr-0940992 , the samsung scholarship ( s. h. ) , the darpa ole program , and the macarthur professorship endowed by the john d. and catherine t. macarthur foundation at the university of illinois ( f. k. ) for financial support . | we investigate the instabilities of the mott - insulating phase of the weakly disordered bose - hubbard model within a renormalization group analysis of the replica field theory obtained by a strong - coupling expansion around the atomic limit .
we identify a new order parameter and associated correlation length scale that is capable of capturing the transition from a state with zero compressibility , the mott insulator , to one in which the compressibility is finite , the bose glass .
the order parameter is the relative variance of the disorder - induced mass distribution . in the mott insulator ,
the relative variance renormalizes to zero , whereas it diverges in the bose glass .
the divergence of the relative variance signals the breakdown of self - averaging .
the length scale governing the breakdown of self - averaging is the distance between rare regions .
this length scale is finite in the bose glass but diverges at the transition to the mott insulator with an exponent of @xmath0 for incommensurate fillings .
likewise , the compressibility vanishes with an exponent of @xmath1 at the transition . at commensurate fillings , the transition is controlled by a different fixed point at which both the disorder and interaction vertices are relevant . |
in the relativistic ( or dirac- ) brueckner model the dynamical two - nucleon correlations in infinite nuclear matter are studied .
the formalism is based on an effective quantum field theory for mesons and nucleons @xcite and thus ignores the underlying quark nature of the nucleon . in the brueckner model the t - matrix ( or brueckner g - matrix ) serves as an effective in - medium two - body interaction .
it is determined by a self - consistent summation of the ladder diagrams in a quasipotential approximation ( thompson equation ) to the bethe - salpeter equation , see , e.g. , ref .
the bare nucleon - nucleon ( ladder ) interaction is described by one - boson exchange potentials . in the present work
we apply the bonn potentials a , b , c of refs .
@xcite and for comparison also a simple @xmath0-potential which is well known from the walecka model @xcite .
the nucleon inside the nuclear medium is treated as a dressed dirac particle where the influence of the coupling to the surrounding nucleons via the t - matrix serving as an effective interaction is expressed by the nucleon self - energy .
the self - energy has a lorentz structure with large scalar and four - vector components and modifies the single - particle spectrum and the spinor wave functions .
the relativistic brueckner model has been developed in the last decades by various authors . at the beginnig of the 80 s
the relativistic problem was considered by the brooklyn group @xcite in first order perturbation theory . in the following the covariant formalism to a self - consistent treatment of the thompson equation
was developed by horowitz and serot @xcite and is in detail outlined in ref .
. however , in these works the nucleon - nucleon interactions was described within the framework of the @xmath0-model .
calculations with realistic nucleon - nucleon interactions have been performed by brockmann and machleidt @xcite and later on by ter haar and malfliet @xcite .
a more rigorous derivation of the brueckner approach in the framework of relativistic green s function can be found in ref .
@xcite . in the present work we refer to the formalism described in refs .
@xcite . the main difference between the relativistic brueckner approach of refs .
@xcite and previous works is the use of medium dependent spinors and the requirement of self - consistency for both the spinor wave functions and the single - particle spectrum .
the dirac spinors of the nucleon in the medium differ considerably from the free spinors due to the large scalar and vector self - energies .
these introduce an additional density dependence into the spinor matrix elements of the one - boson exchange potentials and thus into the t - matrix .
this dynamical effect is absent in non - relativistics .
another new aspect of the modern brueckner approach is that the inertial frames where the t - matrix is determined and where it is used have to be distinguished carefully .
due to the neglection of the retardation in the meson propagators the one - boson exchange potentials and the correspondingly t - matrix serve as quasi - potentials in the two - particle center - of - mass ( c.m . )
therefore the relativistic thompson equation is solved in this system which can be done by numerical standard techniques , as , e.g. , in the non - relativistic case of ref .
@xcite . on the other hand single - particle quantities like the self - energy of the nucleon
are calculated in the nuclear matter rest frame , where the single - particle distribution , the fermi sea , is naturally defined .
thus a covariant representation of the t - matrix is required to connect both frames .
even though , e.g. , the single particle potential can be calculated with some additional approximations avoiding this representation , as done in ref .
@xcite , the covariant technique is unavoidable in order to calculate the self - energy components themselves . in the relativistic brueckner approach
the self - consistent spinors are restricted to positive energy states .
a different approach in the full dirac space which is also based on one - boson exchange potentials is used in refs .
but in a full treatment one should further include renormalisation effects like the vacuum fluctuation contributions to the self - energy , as discussed in the context of the relativistic hartree approximation in refs .
@xcite , and solve the full bethe - salpeter equation self - consistently .
however , then the one - boson exchange potentials would be inappropriate .
but these are still outstanding problems . in the standard brueckner model which we discuss in the present work one gets around the problem of renormalization .
the masses and coupling constants in the model are physical renormalized quantities and as a consequence one neglects the effective negative - energy states .
however , the effective positive - energy spinors still contain admixtures of free negative - energy states because they are a superposition of free positive- and negative - energy spinors . to lorentz transform the t - matrix as we have mentioned above the positive energy on - shell t - matrix
is projected onto five invariant amplitudes . by the restriction to positive energy states
an ambiguity arises @xcite which is discussed in terms of the pseudo - scalar and the pseudo - vector representation of the t - matrix in the dirac space . from analogy to the one - pion exchange coupling in the nucleon - nucleon potential
the pseudo - vector choice is more natural and further it is known from meson theory analysis @xcite that this choice suppresses the coupling to the negative - energy states .
however , the difference between both choices also states a measure for the uncertainty concerning the determination of the self - energy components due to the restriction on positive energy spinors .
one issue of the present work is to estimate this uncertainty .
thus , the aim of this work is to determine the self - energy components , to estimate their uncertainty due to different covariant representations of the t - operator and different nucleon - nucleon potentials and to examine the sensitivity of some commonly used observables , i.e. , the equation of state and the optical potential on the self - energies .
we are investigating in detail the density and the momentum dependence of the scalar and the vector components and compare our results to those of other groups @xcite . thereby the momentum dependence can just be determined in a first approximation because it is neglected in the self - consistency treatment of the brueckner model .
commonly this approximation is justified by the assumption of a weak momentum dependence , at least inside the fermi sea , but there have been also indications that this assumption depends on the chosen one - boson exchange potential @xcite .
we show here for the first time the momentum dependence over its full range , i.e. , below and above the fermi surface , and we find it in the case of the bonn a , b , c potentials of the order of several 100 mev , i.e. , to be strikingly strong . in difference to the most publications in the field we focus our discussion to the self - energy components themselves .
it is meanwhile well known that the nuclear matter saturation properties are much better reproduced in the relativistic than in the non - relativistic brueckner approach , i.e. , the predictions for the saturation point of a variety of nucleon - nucleon potentials are located along the _ coester band _ which meets in the relativistic case the empirical area @xcite .
the same also holds , e.g. , for the nucleon optical potential .
however , as we also find in the present analysis these commonly studied observables are not really sensitive to the self - energy components since they mainly depend on a cancellation of the components .
the situation changes in the application of the brueckner theory beyond infinite nuclear matter . when applying relativistic brueckner results to the description of finite nuclei , e.g. , in the framework of an effective ( density dependent ) relativistic mean field theory @xcite the self - energies start to play a decisive role .
this fact is , however , most pronounced in the investigation of heavy ion collisions .
different from resting nuclear matter here the scalar and vector components are out of balance which is due to their different lorentz transformation character and diverse observables are in particular sensitive to the magnitude of the fields @xcite
. furthermore the momentum space configuration can be highly anisotropic and rather corresponds to two counterstreaming currents of nuclear matter which can be described by two lorentz - elongated fermi ellipsoids .
approximate determinations of relativistic brueckner mean fields for such configurations have been performed in @xcite and encouraging applications to heavy ion collisions in @xcite .
thus the density and momentum dependence of the nuclear self - energy components as the quantity which most prominant characterizes the medium effects is a question of particular interest .
this paper is organized in the following way : in the next section we review the relativistic brueckner approach .
there are still some technical differences in the _ standard _ brueckner model between ref .
@xcite and refs .
here we follow the approach of horowitz and serot @xcite which allows the use of standard techniques for the solution of the thompson equation known from earlier works @xcite . however , we use the modern one - boson exchange potentials , treat the isospin channels and also the real and imaginary part of the t - matrix separately and project on both , the pseudo - scalar and the pseudo - vector representation . in the following section we discuss our numerical results .
the calculations with the @xmath0-potential are compared with results from ref .
we further discuss the density and momentum dependence of the scalar and vector self - energy , the influence of pseudo - scalar and pseudo - vector choice is analyzed and we compare to the results of brockmann and machleidt @xcite and the groningen group @xcite .
further the equation of state and the optical potential are shown and their sensitivity to the self - energy components is analyzed . in the final section
a summary is given .
in the relativistic brueckner approach the nucleon inside the nuclear medium may be viewed as a dressed particle in conseqeunce of its two - body interaction with the surrounding nucleons .
this problem is stated as a coupled set of three non - linear integral equations @xmath1 - g t ) .
\label{eq3}\end{aligned}\ ] ] the t - matrix is calculated in the ladder approximation of the bethe - salpeter equation ( [ eq1 ] ) and the bare nucleon - nucleon interaction is described by one - boson exchange potentials @xmath2 .
the intermediate nucleons are given by the two - body propagator @xmath3 which is , however , usually replaced by an effective propagator @xcite , i.e. , the blankenbecler - sugar propagator or the thompson propagator . in the present work the thompson choice is applied .
the pauli operator @xmath4 accounts for the influence of the medium by the pauli principle and projects the intermediate scattering states outside the fermi sea .
the dressed one - body green propagator @xmath5 is calculated via the dyson equation ( [ eq2 ] ) from the free propagator @xmath6 .
the influence of the coupling to the surrounding nucleons is expressed by the self - energy @xmath7 of the nucleon . in the brueckner formalism
this self - energy is determined by summing up the direct and exchange interactions with all the nucleons inside the fermi sea @xmath8 , see eq .
( [ eq3 ] ) . here
the t - matrix plays the role of an effective medium dependent two - body interaction . in the relativistic brueckner approach
one introduces additional approximations for solving the equations ( [ eq1]-[eq3 ] ) which are sketched in the following , for further reading see , e.g. , refs .
@xcite . the most general form of the lorentz - structure of the self - energy compatible with translational invariance , hermiticity , parity conservation , time - reversal invariance and rotational invariance ( in the nuclear matter rest frame ) is @xmath9 with a scalar part @xmath10 and a vector part @xmath11 . due to the dyson equation ( [ eq2 ] )
the full green function has the formal solution @xmath12\ ] ] where we have introduced the effective mass and the kinetic momentum @xmath13 re and i m denote real and imaginary part since ( above the fermi surface ) the self - energy in general will be complex . here
we adopt the _ quasiparticle approximation _
@xcite , i.e. , the i m @xmath14 $ ] will be neglected in eq .
( [ eq5 ] ) .
this means that the decay width of the dressed nucleon is set equal to zero , resulting in an infinite lifetime of this quasiparticle. in the relativistic brueckner approach the full green function @xmath5 is replaced by its positive energy part . thus the self - energy in eq .
( [ eq3 ] ) is determined by the part proportional to the fermi sea @xmath8 , the so - called dirac green function @xmath15 @xcite @xmath162\pi
i \delta(k^{\ast 2 } - m^{\ast 2}(k))\theta(k^{\ast 0 } ) \theta(k_f-|{\bf k}|)\ ] ] which eliminates the divergent contributions from the negative energy sea .
however , in the form of eq .
( [ eq7 ] ) @xmath15 is restricted to the description of nuclear matter at rest . in order to achieve general convenience
the fermi sphere @xmath17 has to be replaced by an fermi ellipsoid @xmath18 with the fermi energy @xmath19 and the boost 4-velocity @xmath20 @xcite . in eq .
( [ eq7 ] ) the momenta of the quasiparticles are put on mass - shell , which is expressed by the @xmath21-distribution and thus @xmath22 .
hence the self - energy @xmath23 depends only on the three - momentum @xmath24 and , of course , on the respective density or @xmath25 .
furthermore , in the relativistic brueckner model one applies a mean field approximation to @xmath7 , eq .
( [ eq4 ] ) , which allows a simple formulation of the self - consistency problem .
the explicit momentum dependence of the self - energy which enters via the term @xmath26 can be dealt with by introducing the reduced kinetic momentum @xmath27 and the reduced effective mass @xmath28 .
further one neglects the momentum dependence of the effective mass ( @xmath29 or @xmath30 ) .
thus , the nucleons fulfill a quasi - free dirac equation @xmath31 u_{\lambda}(k ) = 0\ ] ] and using the normalization of ref .
@xcite the self - consistent positive - energy spinors of helicity @xmath32 are defined as @xmath33 with @xmath34 being a pauli spinor .
the dirac spinors depend on the effective mass and thus on the nuclear density .
the matrix - elements of the t - matrix and the one - boson exchange potentials @xmath2 , eq . ( [ eq1 ] ) , are calculated with these spinors .
in contrast to non - relativistic brueckner theory this fact introduces an additional density dependence into the interaction which is one major reason for the great success of the relativistic treatment .
this density dependence is mediated by the additional parameter @xmath35 which is fixed at a reference point , usually at @xmath36 .
@xmath35 is obtained as the solution of the non - linear equation ( _ reference spectrum approximation _ ) @xmath37 which follows from the definition given above .
self - consistency is achieved by determining for a start value of @xmath30 the medium - dependent spinors of eq .
( [ eq9 ] ) , next the t - matrix from eq .
( [ eq1 ] ) and finally the self - energy and the new @xmath30 . ] .
this procedure is repeated until convergence is reached . the t - matrix , in eq . ( [ eq1 ] ) , is most easily determined in the two - particle center - of - mass ( c.m . ) system while the self - energy , eq . ( [ eq3 ] ) ,
is calculated in the nuclear matter rest frame .
therefore we will determine in a second step covariant amplitudes of the t - matrix .
the thompson propagator ( and similar the blankenbecler sugar propagator ) projects the intermediate nucleons onto positive energy states and restricts the exchanged energy transfer by @xmath38 to zero .
thus equation ( [ eq1 ] ) is reduced to a three dimensional integral equation of the lippmann - schwinger type , the so called thompson equation @xcite @xmath39 here the thompson equation is given in the two - nucleon c.m .- frame where @xmath40 is the relative momentum of the final states and similar @xmath41 are the relative momenta of the initial and intermediate states , respectively .
the thompson propagator and similar the blankenbecler - sugar propagator imply that the timelike component of the momentum transfer in @xmath2 and @xmath42 is set equal to zero which is a natural constraint in the c.m .- frame , however , not a covariant one .
hence we solve the thompson equation in the c.m .- frame in contrast to , e.g. , the approach of ref .
@xcite . the starting energy in eq . ( [ eq11 ] )
is already fixed : @xmath43 .
the pauli operator @xmath4 explicitely depends on the chosen frame , i.e. , on the boost 3-velocity @xmath44 into the c.m .- frame .
with @xmath45 we denote the set of parameters @xmath46 on which the t - matrix actually depends .
we solve the thompson equation ( [ eq11 ] ) for the on - shell t - matrix @xmath47 in the c.m .- system and thereby apply standard techniques which are described in detail by erkelenz @xcite .
we construct the positive - energy helicity t - matrix elements from the @xmath48-scheme . in the on - shell case
only five of the sixteen helicity matrix elements are independent which follows from general symmetries , see ref . @xcite . after a partial wave projection onto the @xmath48-states
the integral reduces to a one - dimensional integral over the relative momentum @xmath24 and eq .
( [ eq11 ] ) decouples into three subsystems of integral equations for the uncoupled spin singlet , the uncoupled spin triplet and the coupled triplet states .
the pauli operator @xmath4 is replaced by an angle averaged pauli operator @xmath49 @xcite . since the fermi sphere is deformed to a fermi ellipsoid in the two - nucleon c.m .-
frame @xmath49 is evaluated for such a fermi - ellipsoid .
@xmath50 with @xmath51 and @xmath52 .
we are solving the integral equations by the matrix inversion techniques of haftel and tabakin @xcite .
real and imaginary parts of the t - matrix are calculated separately by the principal - value treatment given in ref .
@xcite . due to the antisymmetry of the two - fermion states we can restore the total isospin i of the two - nucleon system ( i=0,1 ) with the help of the selection rule : @xmath53 . from the five independent on - shell amplitudes in the @xmath48-representation
the five independent partial wave amplitudes in the helicity representation ( for i=0,1 and real and imaginary part separately ) are obtained by inversion of eq . ( 3.32 ) and
then of eq .
( 3.28 ) of ref .
the summation over the total angular momentum @xmath54 ( we have taken @xmath55 and for comparison @xmath56 ) yields the full helicity matrix element @xmath57 d^j_{\lambda \lambda^{\prime } } ( \theta ) < |{\bf p}|\lambda^{\prime}_1 \lambda^{\prime}_2 |t^{j,{\rm i}}(x)| \ , |{\bf
q}| \lambda_1 \lambda_2 > = < { \bf p } \lambda^{\prime}_1 \lambda^{\prime}_2 { \rm i } { \rm i}_3 | t(x ) @xmath58 is the scattering angle between @xmath59 and @xmath60 and @xmath61 .
the reduced rotation matrices @xmath62 are those defined by rose @xcite .
the matrix element ( [ eq12 ] ) is actually independent of the third component of the isospin @xmath63 .
the next step is the projection of the five independent helicity amplitudes onto five covariant amplitudes .
therefore the t - operator is expanded into the basis matrices in the dirac spinor space @xmath64 the subscripts @xmath65 denote the particle on which the matrix acts .
this expansion , however , is not unique since we have specified the dirac matrix structure of the t operator by its action only on positive energy states .
the expansion of eq .
( [ eq13 ] ) is given with @xmath66 a pseudo - vector interaction @xmath67 .
it is defined as @xmath68 with the transferred momentum at the vertex @xmath69 where @xmath70 indicates final and initial momentum .
we will , however , also adopt a pseudo - scalar interaction where the fourth term in eq .
( [ eq13 ] ) is replaced by @xmath71 as it was also done by horowitz and serot @xcite .
the pseudo - vector choice is , e.g. , made by the groningen group .
this choice is supported by the argument that it agrees with the one - pion exchange coupling in the nucleon - nucleon potential and furthermore suppresses the coupling to the negative energy states which are neglected in the brueckner approach @xcite . in the present work
we apply both variants since we want to examine the influence of the different choices on the self - energy .
next we take the helicity matrix elements of eq .
( [ eq13 ] ) . on the r.h.s .
the helicity states are acting directly on the matrix operators which are now abbreviated by @xmath72 for i=1,2 and @xmath73 @xmath74 the helicity matrix elements on the r.h.s .
are given explicitely , e.g. , in eqs .
( 2.11 - 2.17 ) of ref .
for on - shell scattering between positive energy states the matrix elements of the pseudo - scalar ( ps ) and the pseudo - vector ( pv ) matrix operators are identical @xcite . this vertex equivalance follows immediately from the dirac eq .
( [ eq8 ] ) : @xmath75 .
thus the covariant amplitudes @xmath76 are identical as well .
the difference between both representations of the t - matrix in dirac space enters only at the one - body level when calculating the self - energy @xcite . for fixed parameters @xmath77 equation ( [ eq14 ] )
is a matrix relation between the five independent helicity amplitudes @xmath78 ( five of the sixteen amplitudes numbered with @xmath79 ) on the l.h.s . and the five unknown covariant amplitudes @xmath80 @xmath81 this corresponds to eq .
( 3.23 ) of ref .
@xcite with respect to our different normalization of the spinor basis , eq .
( [ eq9 ] ) .
the covariant amplitudes are determined by a matrix inversion of equation ( [ eq15 ] ) .
this , however , only has to be done for two angles since for on - shell scattering in the two - particle c.m .- frame only two scattering angles appear , i.e. , @xmath82 for the direct and @xmath83 for the exchange interaction .
however , now the matrix becomes singular because two of the five helicity amplitudes vanish .
therefore it is useful to extract the leading angular dependence of the matrix @xmath84 as well as of the helicity amplitudes @xmath78 from the rotation matrices @xmath85 in eq .
( [ eq12 ] ) .
thus , the limit @xmath86 can be performed analytically .
this method has been developed in appendix c of ref .
@xcite and formula ( [ eq15 ] ) can be found there for the limiting angles @xmath87 ( eqs .
( c.10,11 ) ) . proceeding this way we determine the real and the imaginary part of the five covariant amplitudes for the direct and exchange contribution in both , the isospin singlet and triplet channels of the t - matrix @xmath88 .
now we are able to calculate the self - energy . inserting the green function , eq .
( [ eq7 ] ) , and the dirac representation of the t - matrix , eq .
( [ eq13 ] ) , into the definition of the self - energy , eq .
( [ eq3 ] ) , yields @xmath89 \right . \\
\label{eq16a } \left .
+ { \not q}^{\ast}_{\alpha\beta } \left [ 4 t^{\rm v}_d - t^{\rm s}_x + 2 t^{\rm v}_x + 2 t^{\rm a}_x - \frac{(k^{\ast\mu } - q^{\ast\mu})^2}{4m^{\ast 2 } } t^{\rm p}_x\right ] - ( { \not q}^{\ast } - { \not k}^{\ast})_{\alpha\beta } \frac{2 q^{\ast}_{\mu } ( k^{\ast\mu } - q^{\ast\mu})}{4m^{\ast 2 } } t^{\rm p}_x \right\ } \end{aligned}\ ] ] where the @xmath90 denote both , dirac ( spin ) and isospin indices .
the direct and exchange parts of the t - matrix denoted by @xmath91 and @xmath92 are calculated from isospin sums of the isospin projection operators which yields @xmath93 and @xmath94 . the self - energy as a one - body operator in spinor space posseses components proportional to the unity matrix 1 and to @xmath95 as postulated in equation ( [ eq4 ] ) . from this relation
the self - energy components @xmath96 follow by taking traces @xmath97 \qquad = \int \frac{d^3{\bf q}}{(2\pi)^3 } \theta(k_f-|{\bf q}| ) \frac{m^{\ast}}{e^{\ast}({\bf q } ) } t_s({\bf k , q};x ) , \\
\label{eq17 } \sigma_{0}(|{\bf k}|;m^{\ast},k_f ) & = & - \frac{1}{8 } tr[\gamma^{0 } \sigma ] \quad = \int \frac{d^3{\bf q}}{(2\pi)^3 } \theta(k_f-|{\bf q}| ) t_{(0)}({\bf k , q } ; x ) , \\
\label{eq18 } \sigma_{v}(|{\bf k}|;m^{\ast},k_f ) & = & - \frac{1}{8 } tr[\mbox{\boldmath$\gamma$\unboldmath } \cdot{\bf k } \sigma ] = \int \frac{d^3{\bf q}}{(2\pi)^3 } \theta(k_f-|{\bf q}| ) \frac{{\bf
q}^{\ast}\cdot{\bf k}}{|{\bf k}|^2 e^{\ast}({\bf q } ) } t_{v}({\bf k , q};x ) \end{aligned}\ ] ] where we have used the abbreviations @xmath98 , \\
\label{eq20 } t_{(0)}({\bf k , q } ; x ) & = & \frac{1}{4}\left[- 4 t^{\rm v}_d + t^{\rm s}_x -2 t^{\rm v}_x - 2 t^{\rm a}_x + \frac{e^{\ast}({\bf k})}{e^{\ast}({\bf q } ) } \frac{m^{\ast 2 } - k^{\ast}_{\mu } q^{\ast\mu}}{2m^{\ast 2 } } t^{\rm p}_x\right ] , \\
\label{eq21 } t_{v}({\bf k , q } ; x ) & = & \frac{1}{4}\left[- 4 t^{\rm v}_d + t^{\rm s}_x -2 t^{\rm v}_x - 2 t^{\rm a}_x + \frac{|{\bf k}|}{q_z } \frac{m^{\ast 2 } - k^{\ast}_{\mu } q^{\ast\mu}}{2m^{\ast 2 } } t^{\rm p}_x\right ] .\end{aligned}\ ] ] in the pseudo - scalar case @xcite all the cofactors of @xmath99 are replaced just by -1 which yields @xmath100 .
the self - energy of a nucleon is calculated in the nuclear matter rest frame by integrating the effective two - body interaction over the fermi sea .
the 3-momenta @xmath101 of a nucleon pair are given in this frame and we define @xmath102 . on the other hand
the covariant amplitudes are determined in the two - nucleon c.m .
system and thus depend on the respective quantities in that frame @xmath103 from the invariant mass @xmath104 , the relative momentum in the c.m .- frame follows by @xmath105 . as already mentioned above , @xmath45 abbreviates the set of parameters @xmath106 , where the boost 3-velocity ( from the nuclear matter into the c.m .- frame ) is now given by @xmath107 .
in the two - particle c.m .- frame there exist only two possible scattering angles , i.e. , @xmath82 for the direct and @xmath83 for the exchange amplitude .
numerically we solve the thompson equation ( [ eq11 ] ) and determine the covariant amplitudes @xmath80 for a suitable two - dimensional range of the quantities @xmath108 . using the azimuthal symmetry
we calculate the self - energy , eqs .
( [ eq16]-[eq18 ] ) , by a two - dimensional integral over the fermi sphere .
therefore for each nucleon pair we have to interpolate the covariant amplitudes at the respective values of @xmath109 and @xmath110 .
the resulting self - energy , taken at the fermi surface @xmath36 , yields the new value of the effective mass @xmath29 , eq .
( [ eq10 ] ) .
this value serves as the input for the next iteration .
this procedure is repeated until a convergence of the effective mass is achieved .
for the solution of the thompson equation ( [ eq11 ] ) we apply the bonn a , b , c potentials of refs .
@xcite as the nucleon - nucleon interaction @xmath2 .
the potentials are calculated with the obnns code of r. machleidt @xcite .
the bonn potentials are based on the exchange of the six nonstrange bosons with masses below 1 gev @xmath111 . for the pion
a derivative pseudovector coupling is applied .
the three parametrizations a , b and c differ essential in the @xmath112nn form - factor and as a consequence in the strength of the nuclear tensor force .
actually we find only minor differences of a few mev in the resulting self - energies , see below , and thus we present mainly the bonn c results .
for comparison with the results of horowitz and serot we also employ the simple @xmath0-meson exchange of ref .
@xcite with @xmath113 and use the form - factor given therein .
first we want to demonstrate the reliability of our approach by comparison with the results of ref . @xcite . in fig.1
we show the self - consistent effective mass @xmath29 , eq .
( [ eq10 ] ) , as a function of the fermi momentum @xmath25 which is a measure for the density @xmath114 .
our results with the @xmath0-meson exchange potential and the pseudo - scalar decomposition of the t - matrix reproduce quite well the values taken from ref .
@xcite . the slight deviation for densities above saturation density
( here @xmath115 is partly due to the different effective two - nucleon propagators .
as mentioned above , we are using the thompson propagator , eq .
( [ eq11 ] ) , while in ref .
@xcite the blankenbecler - sugar propagator is applied .
furthermore , in ref .
@xcite an additional approximation has been made . as discussed above
, we are solving the thompson equation for a two - dimensional grid of the relative momentum @xmath116 and the c.m .-
momentum @xmath117 respectively the boost - velocity @xmath118 and use this t - matrix for the determination of @xmath7 . in ref .
@xcite the @xmath117-dependence is replaced by an `` average @xmath117 approximation '' for each @xmath116 which causes deviations mainly at high density systems . for the bonn
c potential the effective mass is significantly different from that of the @xmath0-model .
here we applied the pseudo - vector choice . in fig.2
we show the density dependence of the self - energy , i.e. , the scalar part @xmath10 as well as the time - like @xmath119 and the space - like component @xmath120 of the vector self - energy .
the self - energies are determined for the three bonn potentials a , b , c and again we the pseudo - vector choice is applied . for the sake of a better comparison
the sign of the vector self - energies is inversed and the dimensionless space - like component @xmath121 is multiplied by @xmath25 .
the scalar and also the vector self - energy component @xmath122 differ only slightly for the three bonn potentials .
the difference amounts about 5 mev at saturation density ( here @xmath123 ) and is still less than 10 mev at @xmath124 . in
absolute magnitude bonn c yields the weakest and bonn a the strongest self - energies .
the space - like vector self - energy @xmath120 is equal within line - width for the three bonn potentials and is small in comparison to @xmath125 , e.g. , about 7% of @xmath126 at saturation density . for further analysis
we therefore concentrate on the large components @xmath125 calculated with the bonn c potential .
the two different representations of the t - matrix , i.e. , the pseudo - scalar and the pseudo - vector case , represent an inherent ambiguity .
this is , however , a common feature of all relativistic brueckner calculations which specify the dirac matrix structure of the t - matrix by using only positive energy t - matrix elements . in order to clarify the influence of this ambiguity in figs.3 .
we show the self - energy components @xmath127 calculated exemplaryly with the bonn c potential for both choices . in the pseudo - scalar case ,
see fig.3 , the self - energies are larger by about 40 - 80 mev ( decreasing with increasing density ) than in the pseudo - vector case ( fig.4 ) .
this difference may be interpreted as a measure of the general uncertainty of the self - energy components arising from any particular projection of the t - operator on positive energy matrix elements only .
this uncertainty is larger at low densities . to get more insight into the origin of this difference we decompose the self - energy in the two contributions stemming from the direct and the exchange t - amplitude respectively .
e.g. , in the pseudo - scalar case the direct part of the self - energy @xmath10 is determined by the integral over @xmath128 while the exchange part is given by the integral over the sum of all exchange - amplitudes @xmath129 , see eq .
( [ eq19 ] ) and the subsequent comment .
we find that in the pseudo - scalar case direct and exchange parts of @xmath127 are identical within less than 0.5 mev , see fig.3 .
however , in the pseudo - vector case the factor accompanying the @xmath99 amplitude in eqs .
( [ eq19],[eq20 ] ) reduces the exchange part by roughly 50% . for comparison in the @xmath0-model , which lacks of pion exchange ,
the @xmath99 amplitude is generally smaller and the effect , i.e. , the reduction of the exchange part , amounts to roughly 25% .
of course also the direct part is different from that of the pseudo - scalar case since we consider the self - consistently determined fields .
we conclude that the pseudo - vector choice in general reduces the total self - energy which is due to a considerable reduction of the exchange contributions while the direct contributions are even enhanced . in fig.5
we compare our results for @xmath122 obtained with the pseudo - vector choice and the bonn c potential with the results of two other groups , i.e. , the groningen group @xcite and the approach of brockmann and machleidt @xcite . in ref .
@xcite the relativistic thompson equation is solved by standard methods @xcite directly in the nuclear matter rest frame instead of the two - particle c.m .- frame
. therefore the thompson eq .
( [ eq11 ] ) is galilei transformed to the rest frame with the c.m .-
momentum @xmath130 .
the relativistic single particle potential @xmath131 in the nuclear matter rest frame can be evaluated from this t - matrix by @xmath132 with the definition of @xmath133 as in eq .
( [ eq3 ] ) .
the advantage of this method is that one is not forced to lorentz transform the matrix elements between the nuclear matter rest frame and the two - particle c.m .- frame and thus a projection onto covariant amplitudes is obsolete which simplifies the task considerably .
however , now the self - energy components @xmath122 can not be calculated directly as in the projection method of eqs .
( [ eq16]-[eq18 ] ) .
instead , one has to extract them from the single particle potential @xmath131 . with the form of @xmath7 as in eq .
( [ eq4 ] ) , however , thereby neglecting @xmath121 one finds @xmath134 i.e. , the single particle potential results from a cancellation of the huge scalar and vector components . using a mean field ansatz for @xmath127 ,
i.e. , neglecting their momentum dependence , one is able to determine them by a fit procedure to @xmath135 . in ref .
@xcite the self - energy components are constructed by this potential - fit method . in fig.5
we compare these results with our calculations both for the bonn c potential . in the density range considered the fitted self - energies are generally smaller in absolute values by 50 up to more than 150 mev , e.g. , at saturation density they are reduced by about 75 mev . in spite of this discrepancy both calculations have several features in common .
e.g. , the single particle potential , defined in eq .
( [ eq24 ] ) , is almost identical and as we will see later on the same holds for the equation of state . concerning these two quantities
the main difference between both treatments lies in the reference frame where the t - matrix is actually calculated while the projection method on covariant amplitudes is not essential . at moderate densities
the t - matrix depends only weakly on the c.m .-
momentum @xmath136 and respectively on the chosen frame .
thus the numerical results show only minor differences .
another feature in common is that also the potential - fit method yields self - energies for the three types of bonn potentials which are equal within a few mev @xcite . in fig.5
we further compare to the results of the groningen group @xcite .
they also perform a full relativistic brueckner calculation solving the thompson equation in the two - particle c.m .- system and using a projection of the t - matrix onto a pseudo - vector representation .
the difference to our method is on the one hand the application of a different one - boson exchange potential including different form - factors , the so called groningen potential . on the other hand as - a more technical point - in ref .
@xcite the thompson equation is solved in full momentum - spin space by the use of pad approximants , for further details see , e.g. , ref .
@xcite . for densities above saturation density
the self - energies calculated with the groningen potential are roughly the same as the ones obtained by us , however , in the low density range they are much smaller in magnitude . up to now
we have discussed the self - energy of a nucleon in nuclear matter obtained at the fermi surface ( @xmath36 ) , i.e. , the density dependence of the self - energy . in the relativistic brueckner theory
the self - energy at the fermi surface and the t - matrix are determined self - consistently , as described above .
the dressed nucleon propagators respectively the self - consistent spinor basis , eq . ( [ eq9 ] ) , entering into the calculation of the t - matrix are constructed with the effective mass , eq .
( [ eq10 ] ) .
@xmath29 itself depends on the density dependent self - energies @xmath137 .
thus only the effective mass acts as a self - consistency or iteration parameter . in a next step
the self - energies above and below the fermi surface can be evaluated from eqs .
( [ eq16]-[eq18 ] ) .
this has to be considered just as the first iteration step in the determination of self - consistent momentum dependent functions @xmath138 and stands in contrast to the non - relativistic brueckner treatment where self - consistency is required for the full positive single particle energy spectrum . however , these approximations are not based on the same footing . in the relativistic treatment
we have an intrinsic momentum dependence due to the dirac structure of the nucleon which introduces a momentum dependence in the single particle potential @xmath131 even for constant values of @xmath96 , see eq .
( [ eq25 ] ) . a further argument often given in connection with the reference spectrum approximation
@xcite is that the momentum dependence of the self - energies in the relativistic brueckner approach is found to be rather soft .
however , in the present work we demonstrate that this is in general not the case but strongly depends on the special form of the one - boson exchange potential . in fig.6
we show the momentum dependence of the nucleon self - energies @xmath127 at saturation density calculated with the @xmath0-exchange potential .
our results nicely agree with the results of ref .
@xcite within 10 mev which might be due to different technical details already mentioned in the discussion of fig.1 .
we find that @xmath10 decreases in the range from @xmath139 to @xmath25 by 14 mev and @xmath140 by 24 mev , i.e. , below the fermi surface the self - energies are constant within a few percent .
in fig.7 we show the momentum dependence of the self - energies for calculations with the bonn c potential ( pseudo - vector choice ) and the @xmath0-potential ( pseudo - scalar choice ) , both at @xmath141 gev .
more strictly this is the real part of the self - energy since for momenta greater than the fermi momentum the self - energy is complex .
the results obtained with the bonn c potential show a strong momentum dependence : over the entire momentum range up to 0.8 gev corresponding to a single particle energy of around 300 mev the self - energies decrease by nearly 60% with respect to the central value .
the most pronounced change occurs in the region around the fermi surface while deep inside the fermi sea and far above the fermi momentum the momentum dependence nearly saturates . for the @xmath0-potential the self - energies are rather weakly momentum dependent : over the full interval they decrease by about 20% .
for comparison we also show two results obtained by the groningen group . in ref .
@xcite the self - energies above the fermi surface are calculated in a similar way as in the present work , see discussion of fig.5 .
the results obtained by the groningen group @xcite are in general softer k - dependent than those of the present work obtained with the bonn potentials .
the results of ref .
@xcite show a momentum dependence as strongly as ours .
however , in this case a comparison is not fully appropriate since these calculations include an additional self - consistent treatment of the pion polarisation . in fig.8
we show the equation of state , i.e. , the energy per particle e / a as a function of the nuclear matter density @xmath142 , calculated with bonn a , b , c . in the relativistic brueckner theory
the energy per particle is defined in analogy to the non - relativistic hartree - fock method as the kinetic plus half of the potential energy @xmath143 - m \end{aligned}\ ] ] with the self - consistent spinors @xmath144 given in eq .
( [ eq9 ] ) . by this way we have determined e /
a by integrating the self - energy components of eqs .
( [ eq16]-[eq18 ] ) over the fermi sphere , see eq .
( [ eq27 ] ) . in fig.8
we compare to the results of ref .
@xcite where e / a is calculated directly from the single particle potential , eq .
( [ eq24 ] ) , using the expression ( [ eq26 ] ) . for moderate densities
the results agree with an accuracy better 0.5 mev and even for @xmath145 the deviation is less than 3 mev .
firstly , this result is an additional test for the accuracy of our numerical treatment , i.e. , the projection of the t - matrix onto covariant amplitudes which are subsequently used to determine the self - energy components and the equation of state .
secondly , these results again demonstrate that the equation of state is quite insensitive to the self - energy components themselves . actually , as can be seen clearly from the definition eq .
( [ eq26 ] ) , the equation of state depends directly on the single - particle potential , i.e. , the difference of @xmath10 and @xmath126 , eq .
( [ eq25 ] ) , and relatively weak on the effective mass via the kinetic energy part .
nevertheless , the relativistic effects are responsible in contrast to non - relativistic calculations that the minima of the equations of state for the three different bonn potentials reveal a coester band which meets the empirically found saturation area of nuclear matter . in order to demonstrate the influence of the two different choices ,
i.e. , pseudo - vector and pseudo - scalar on the equation of state in fig.9 we show the respective equation of state , again for bonn c. actually the expectation value @xmath146 of the full self - energy operator , eq .
( [ eq16a ] ) , is identical for both choices @xcite .
however , as can be seen from fig.9 the equation of state turns out to be rather sensitive on the different choices .
this significant dependence is only due to the self - consistently iterated effective mass which differs in both approaches .
this effect deepens the equation of state in the pseudo - scalar choice by around 1 to 2 mev and shifts the minimum towards higher densities .
thus the saturation point is shifted away from the empirical point towards the results of non - relativistic brueckner calculations as , e.g. , published in ref .
this observation additionally supports the pseudo - vector choice . in view of the strong momentum dependence of the self - energy components for the bonn potentials one might expect that the results depend sensitive on the actual momentum reference value where the selfconsistent effective mass is determined . as a consequence
a selfconsistent treatment of the full momentum dependence might appear to be indispensable .
therefore we have also examined the influence of a different choice of the selfconsistent effective mass . in the reference spectrum approximation ,
( [ eq10 ] ) , @xmath29 is determined at the fermi surface which is the conventional treatment in the relativistic brueckner approach . however , this choice is by no means unique .
thus we also determine @xmath29 by its mean value in analogy to eq .
( [ eq10 ] ) @xmath147 the self - energies @xmath148 are not simply taken at @xmath36 but replaced by their values averaged over the fermi sea @xmath149 these averages are lorentz scalars .
the extension of the reference spectrum approximation , eq .
( [ eq28 ] ) , is especially meaningful with respect to the treatment of anisotropic nuclear matter , as done in ref .
@xcite . here
it serves as a test of the influence of the momentum dependence . due to
the strong momentum dependence observed , e.g. , in fig.7 the mean value @xmath150 differs considerably from @xmath151 , i.e. , by about 57 mev at saturation density ( with bonn c ) .
respectively for the different effective masses we obtain a mean value @xmath152 mev which is smaller than the selfconsistent mass @xmath153 mev .
but if we use the alternatively defined effective mass , eq .
( [ eq28 ] ) , as iteration parameter we find for the final iterated mean value @xmath154 mev . thus the mean value is increased , i.e. , it is shifted towards the previous self - consistent result or , in other words , the selfconsistent value is rather stable .
the comparison of the self - energy components @xmath155 in both iterations , see fig.2 , shows just slightly deviations , e.g. , about 10 mev at saturation density . and
even more , the equation of state , shown in fig.9 , remains nearly unchanged .
these results demonstrate that the strong momentum dependence has only minor influence on the density dependence of the self - energy components and that the equation of state remains unaffected .
another quantity which is more sensitive on the momentum dependence of @xmath96 is the schroedinger equivalent optical potential which a nucleon feels inside the nuclear medium .
the optical potential is given by @xmath156 it contains counteracting linear and quadratic terms in the self - energy components . for nucleons with @xmath157
the momentum space is not completely blocked for scattering processes and thus above the fermi surface the self - energies and the optical potential become complex . the single - particle energy @xmath158 is a monotonic function of @xmath24 which allows by inversion to determine @xmath159 as function of the single - particle energy @xmath160 .
we want to stress that neglecting the momentum dependence of the fields @xmath161 and neglecting completely @xmath121 the optical potential @xmath162 would simply be a linear function of @xmath160 . however , this is not the case here .
real and imaginary part of the optical potential calculated for bonn c ( pseudo - vector choice ) are shown in fig.10 .
the real part agrees well with the groningen results of ref .
while the imaginary part of the present calculations is about twice as large . further we compare to the empirical optical potential hama i , table 2 of ref . @xcite .
for the full range of single - particle energies up to 700 mev , shown in fig.10 , our results deviate from the empirical re @xmath162 by less than 10 mev . up to the pion production threshold at around 300 mev
we observe a remarkable agreement of the imaginary part of @xmath162 with the empirical data . for single - particle energies above this region
the results appear to be no more fully reliable since then meson - nucleon resonances , e.g. the @xmath163(1232 ) resonance , should be taken into account .
in this work in particular the self - energy of a nucleon inside the nuclear medium has been studied in the framework of the relativistic brueckner approach .
therefore the dependence of the lorentz scalar and vector self - energy components on the nuclear density and the nucleon momentum have been examined . as bare nucleon - nucleon interaction various one - boson exchange potentials
have been employed , i.e. , the bonn a , b , c potentials and for comparison the @xmath0-exchange potential . for the latter we are able to nicely reproduce the density dependence of the self - energies of ref .
the self - energy components themselves deviate for the three different bonn potentials over a wide density range only by a few mev and thus we restricted our discussion mainly to the bonn c potential . from the comparison with our results we find that the potential - fit method of ref .
@xcite underpredicts the magnitude of the self - energies by about 50 up to more than 150 mev .
further we analyzed the influence of different possible covariant representations of the t - matrix .
the positive energy on - shell t - matrix is projected on five invariant amplitudes and by the restriction to positive energy states an ambiguity arises which is discussed in terms of the pseudo - scalar and the pseudo - vector representation . although the t - amplitudes of both choices are equal on the one - body level , i.e. , concerning the respective self - energy components a significant influence of these representations is observed .
this is due to the fact that in the pseudo - scalar case direct and exchange contributions to the self - energies are equal while in the pseudo - vector choice the exchange part is suppressed . for the bonn potentials this suppression lies between 40 - 80 mev and
is most prominant at low densities .
this difference between both choices has to be interpreted as an inherent measure for an uncertainty of the approach due to the restriction on positive energy spinors .
commonly the momentum dependence of the self - energy components is supposed to be weak .
this assumption is also reflected in the self - consistency treatment in the relativistic brueckner model which is managed by just one density dependent parameter , i.e. , the self - consistent effective mass .
therefore we have carefully examined this momentum dependence which , however , can be done within the bruckner scheme just in a first approximation . as in ref .
@xcite we find that taking the @xmath0-exchange potential the self - energies are found to be constant below the fermi surface within a few percent .
however , this result does not hold for the more realistic bonn potentials . here
we observe a strong momentum dependence which may be due to the strong pion exchange .
the self - energies decrease from their central value , i.e. , at @xmath164 to the fermi surface by about one fourth and further to @xmath165 gev by more than one half .
this striking momentum dependence makes the treatment of the self - consistency in the model somewhat questionable . but a test with a different defined self - consistent effective mass which is determined by a mean value over the fermi sea and not taken at @xmath25 as usually done shows that the self - energy components are rather stable at not too high densities .
thus the self - consistency treatment in the model still appears to be justified up to two or three times saturation density . in this context
the question raises which observables are sensitive to the self - energy components and their momentum dependence .
it is remarkable that the full self - energy and , respectively , the single particle potential show ( absolutely ) a much weaker momentum dependence .
as it is well known from mean field theory the large scalar and vector self - energy components contribute to these quantities with opposite signs and thus cancel each other to hight extent .
the resulting observables , e.g. , the single particle potential are smaller by nearly one order of magnitude . here
we found that this holds not only at @xmath25 but also over the full momentum range considered .
thus the momentum dependence cancels to a large extent .
also the equation of state depends mainly on the difference of the scalar and the vector self - energy and thus the absolute values of the single component are of minor importance .
this can be seen most clearly by the use of just density dependent self - energy components determined with the potential - fit method of brockmann and machleidt @xcite which deviate from our results ( at @xmath25 ) by about 75 mev .
the resulting binding energy , however , only deviates by about 1 mev .
similarly the self - energy components obtained in the pseudo - scalar and the pseudo - vector choice differ considerable but the equations of state differ by less than one to two mev . on the other hand , small variations of the self - energy due to
the different three bonn potentials produce a significant change in the equations of state .
summarizing , the equation of state is not sensitive to the self - energy components themselves but to their difference .
a bit more sensitive is another quantity , i.e. , the optical potential .
it depends not only on the cancellation of the scalar and vector self - energy but also contains quadratic terms .
we find a good agreement of our results with the experimental data of ref .
@xcite for , both , the real and the imaginary part up to a single particle energy of several hundred mev . as a resum ,
the self - energy components in the relativistic brueckner model are still affected with some uncertainty . for realistic one - boson exchange potentials
we find them to be strongly momentum dependent .
however , for this attribute we have to be aware of the limits of the model .
one major success of the relativistic approach was that the equation of state comes close to the empirical saturation point .
the equation of state is rather insensitive to the self - energy components themselves .
similarly , the nucleon optical potential is only little more sensitive .
however , a precise knowledge of the size of each self - energy component is necessary , e.g. , in the much more complicate situation of heavy ion collisions .
the authors acknowledge stimulating discussions with f. de jong and h. mther .
99 j. d. walecka , ann .
( n. y. ) * 83 * , 497 ( 1974 ) ; + b. d. serot , j. d. walecka , advances in nuclear physics , * 16 * , 1 , eds .
j. w. negele , e. vogt , ( plenum , n.y . , 1986 ) g. e. brown , a. d. jackson , _ the nucleon nucleon interaction _ ( north holland , amsterdam , 1976 ) . r. machleidt , advances in nuclear physics , * 19 * , 189 , eds
. j. w. negele , e. vogt , ( plenum , n.y . , 1989 ) .
m. r. anastasio , l. s. celenza , c. m. shakin , phys .
rev . * c23 * , 2258 ( 1981 ) ; phys
. rev . * c23 * , 2273 ( 1981 ) .
m. r. anastasio , l. s. celenza , w. s. pong , c. m. shakin , phys .
rep . * 100 * , 327 ( 1983 ) . c. j. horowitz , b. d. serot , phys .
* b137 * , 287 ( 1984 ) . c. j. horowitz , b. d. serot , nucl
a464 * , 613 ( 1987 ) .
r. brockmann , r. machleidt , phys .
* b149 * , 283 ( 1984 ) .
r. brockmann , r. machleidt , phys . rev . *
c42 * , 1965 ( 1990 ) .
b. t. haar , r. malfliet , phys . rep . * 149 * , 207 ( 1987 ) .
b. t. haar , r. malfliet , phys . rev . *
c36 * , 1611 ( 1987 ) .
r. malfliet , progress in particle and nuclear physics , * 21 * , 207 , ed .
a.faessler , ( oxford , 1988 ) ; + w. botermans , r. malfliet , phys . rep . * 198 * , 115 ( 1990 ) ; + f. d. jong , r. malfliet , phys . rev . * c44 * , 998 ( 1991 ) .
m. i. haftel , f. tabakin , nucl . phys . *
a158 * , 1 ( 1970 ) . p. poschenrieder , m. k. weigel , phys . rev . *
c38 * , 471 ( 1988 ) . c. nuppenau , y. j. lee , a. d. mackellar , nucl
a504 * , 839 ( 1989 ) .
j. a. tjon , s. j. wallace , phys . rev . *
c32 * , 267 ( 1985 ) ; phys . rev . *
c32 * , 1667 ( 1985 ) .
f. d. jong , b. t. haar , r. malfliet , phys .
b220 * , 485 ( 1989 ) . c. fuchs , h. lenske , h. h. wolter , phys
. rev . * c52 * , 3043 ( 1995 ) . c. fuchs , e. lehmann , l. sehn , f. scholz , t. kubo , j. zipprich , a. faessler , nucl .
* a603 * , 471 ( 1996 ) .
l. sehn , h. h. wolter , nucl . phys . *
a601 * , 473 ( 1996 ) ; + c. fuchs , l. sehn , h. h. wolter , nucl . phys . *
a601 * , 505 ( 1996 ) ; + h. elsenhans , l. sehn , a. faessler , h. mther , n. ohtzuka , h. h. wolter , nucl .
* a536 * , 750 ( 1992 ) . c. fuchs , t. gaitanos , h. h. wolter , phys
b381 * , 23 ( 1996 ) .
k. erkelenz , phys . rep . *
c13 * , 191 ( 1974 ) .
j. d. bjorken , s. d. drell , relativistic quantum mechanics ( mcgraw - hill , n.y . , 1964 ) ; + j. d. bjorken , s. d. drell , relativistic quantum fields ( mcgraw - hill , n.y . , 1965 ) .
m. trefz , a. faessler , w. h. dickhoff , nucl .
a443 * , 499 ( 1985 ) .
m. rose , elementary theory of angular momentum ( wiley , n.y . , 1957 ) .
f. d. jong , ph.d .
thesis , groningen , the netherlands ( 1992 ) ( unpublished ) .
r. machleidt , in : computational nuclear physics 2 nuclear reactions , k. langanke , j. a. maruhn , and s. e. koonin , eds .
( springer , new york , 1993 ) , chapter 1 , p. 1 .
s. hama , b. c. clark , e. d. cooper , h. s. sherif , r. l. mercer , phys .
rev . * c41 * , 2737 ( 1990 ) .
* self - consistent effective mass @xmath29 in units of the bare mass ( @xmath166=939 mev ) as a function of the fermi momentum @xmath25 .
the relativistic brueckner results obtained with the @xmath0-exchange potential ( full line ) are compared to those of ref .
@xcite ( diamonds ) .
the effective mass for the bonn c potential ( dashed line ) differs significantly from that of the @xmath0-model . * self - energy components for the pseudo - vector choice , calculated with various bonn potentials .
the dotted , dashed and full lines represent the bonn a , b and c potential . the scalar self - energy @xmath10 ( lower half - plane ) and the time- and space - like vector self - energies @xmath167 ( upper half - plane ) are shown as functions of @xmath25 .
for the three bonn potentials the small space - like vector self - energies @xmath168 are equal within line width .
the remaining curves ( dashed dotted ) involve a different definition of the self - consistent effective mass which is given in the text later on . * self - energy components @xmath122 calculated with the pseudo - scalar choice and the bonn c potential .
the curves represent the total self - energy components ( full lines ) , their direct ( dir . ) and their exchange ( exc . )
parts which are practically equal ( dashed dotted lines ) .
* self - energy components @xmath122 calculated with the pseudo - vector choice and the bonn c potential .
the curves represent the total self - energy components ( full lines ) , their direct ( dir . )
parts ( dashed lines ) and their exchange ( exc . ) parts ( dotted lines ) .
* self - energy components @xmath122 calculated with the pseudo - vector choice and the bonn c potential ( full lines ) .
they are compared with the results of ref .
@xcite ( dashed lines ) and of ref .
@xcite ( dotted lines ) . * self - energy components @xmath122 at saturation density @xmath169 gev ) as function of the nucleon momentum @xmath170 .
the self - energies are calculated with the @xmath0-exchange potential ( pseudo - scalar choice ) and compared with values ( symbols ) taken from ref .
@xcite . * self - energy components @xmath122 as function of the momentum @xmath24 of the nucleon for @xmath141 gev .
the self - energies are calculated with the bonn c ( pseudo - vector choice , solid line ) and the @xmath0-exchange potential ( pseudo - scalar choice , dotted line ) and compared with results of ref .
@xcite ( dashed line ) and of ref .
@xcite ( @xmath171 gev , dashed dotted line ) .
* equation of state , i.e. , energy per particle @xmath172 as a function of the nuclear matter density @xmath142 , for the three bonn potentials a , b and c ( from below to top : dashed line , dotted line and solid line ) .
the symbols represent the respective results of ref .
@xcite . * equation of state for the bonn c potential calculated in the pseudo - vector ( solid line ) and the pseudo - scalar ( dashed line ) choice .
the remaining curve ( dotted ) involves a different definition of the self - consistent effective mass given in the text . * real and imaginary part of the schroedinger equivalent optical potential @xmath162 as function of the single - paricle energy @xmath173 .
the results for the bonn c potential ( real part : solid line , imaginary part : dotted line ) are compared to the results of ref .
@xcite ( real part : dashed line , imaginary part : dashed dotted line ) and to the empirical optical potential taken from ref .
@xcite ( real part : diamonds , imaginary part : circles ) . | the formalism of the relativistic ( or dirac- ) brueckner approach in infinite nuclear matter is described .
as nucleon - nucleon interaction the one - boson exchange potentials bonn a , b , c and for comparison the walecka model are used .
the t - matrix is determined from the thompson equation and is projected onto five covariant amplitudes . by the restriction to positive energy states
an ambiguity arises in the relativistic brueckner approach which is discussed here in terms of the pseudo - scalar and the pseudo - vector projection .
the influence of the coupling of the nucleon via the t - matrix as an effective two - nucleon interaction to the nuclear medium is expressed by the self - energy . in particular
we investigate the scalar and vector components of the self - energy for the different one - boson exchange potentials and discuss their density and momentum dependence .
we estimate the uncertainty of the self - energy due to the pseudo - scalar and the pseudo - vector choice .
usually the momentum dependence of the self - energy is thought to be weak , however , we find that this depends on the one - boson exchange potentials . for the bonn potentials , in contrast to the @xmath0-potential ,
the momentum dependence is strikingly strong above as well as below the fermi surface .
we compare with the results of other groups and study the effects on the equation of state and the nucleon optical potential . |
the description of strong interactions in the nonperturbative regime is still one of the most important open problems in particle physics .
it is believed that the strong interaction lagrangian supports an su(2 ) chiral symmetry which is dynamically broken at low energies , and pions play the rle of the corresponding goldstone bosons .
a simple scheme including these properties is the well known nambu@xmath0jona - lasinio ( njl ) model @xcite , proposed more than four decades ago .
the njl model has been widely used as an schematic effective theory for qcd @xcite , allowing e.g. the description of light mesons as fermion - antifermion composite states . in the njl model quarks
interact through a local , chiral invariant four fermion coupling . due to the local nature of this interaction ,
the corresponding schwinger - dyson and bethe - salpeter equations become relatively simplified . however , the main drawbacks of the model are direct consequences of this locality : loop integrals are divergent ( and therefore have to be regulated somehow ) , and the model is nonconfining . the absence of confinement is basically related to the fact that the dynamically generated constituent quark masses are momentum independent .
this imposes a severe restriction on the range of applicability of the model , since a @xmath1 continuum opens up at energies of twice the constituent mass .
since the njl model is nonrenormalizable , in practice it is necessary to apply some form of ultraviolet regularization , with the introduction of a finite cutoff parameter .
the adopted regularization scheme has to be regarded as a part of the model itself .
a variety of schemes have been used in the literature , such as sharp three- or four - momentum cutoffs , and proper time or pauli - villars regulators .
although the model does contain regularization - independent information @xcite , and results obtained within various regularization schemes are found to be qualitatively similar @xcite , the choice of any particular scheme lacks a strong physical motivation .
moreover , it can be seen that one needs further regularization prescriptions in order to carry out calculations beyond the leading order in the @xmath2 expansion ( a new cutoff is required for meson loops @xcite ) . on the other hand ,
the introduction of a finite cutoff is somewhat problematic in the anomalous sector : if low energy theorems for anomalous processes such as @xmath3 are assumed to be valid , then the corresponding quark loop integrals should include a complete set of quark states . in the njl framework , this can be achieved either by leaving the anomalous diagrams ad - hoc unregulated @xcite or by including additional terms in the lagrangian in order to recover the anomalous ward identities @xcite . as a way to improve upon the njl model , extensions which include nonlocal interactions
have been proposed ( see ref .
@xcite and references therein ) .
in fact , nonlocality arises naturally in the context of several well established approaches to low energy quark dynamics , as e.g. the instanton liquid model @xcite and the schwinger - dyson resummation techniques @xcite .
lattice qcd calculations @xcite also indicate that quark interactions should act over a certain range in the momentum space .
moreover , it has been argued that nonlocal extensions of the njl model do not show some of the above mentioned inconveniences of the local theory .
indeed , nonlocal interactions regularize the model in such a way that anomalies are preserved @xcite and charges are properly quantized , the effective interaction is finite to all orders in the loop expansion and therefore there is not need to introduce extra cutoffs @xcite , soft regulators such as gaussian functions lead to small next - to - leading order corrections @xcite , etc .
in addition , it has been shown @xcite that a proper choice of the nonlocal form factor and the model parameters can lead to some form of quark confinement , in the sense that the effective quark propagator has no poles at real energies @xcite . in order to bring the problem into a tractable form , most of the calculations reported in the literature deal with nonlocal interactions which are separable in momentum space .
in fact , basically two alternative schemes to introduce nonlocality in a separable way have been considered .
one of them @xcite is inspired on the instanton liquid picture of qcd , which indeed gives rise to an effective separable nonlocal four quark vertex .
the other one @xcite might be understood as a separable approximation to an effective one - gluon exchange interaction .
considerable work has been done using one scheme or the other .
this includes studies of the mesonic @xcite and the baryonic @xcite sectors , as well as the behavior of quark matter at finite temperature and densities @xcite .
the aim of the present work is to present a detailed comparison of both approaches and their relation with the njl model , analyzing the respective input parameter ranges for different form factors and discussing the validity of low energy theorems .
the article is organized as follows . in sec .
ii we introduce a common framework to deal with the above mentioned approaches to nonlocal separable interactions .
analytical expressions for mean field quantities , meson masses and decay constants are given for both schemes . in sec .
iii , the validity of various chiral relations is explicitly shown . in sec .
iv we quote some numerical results for definite nonlocal form factors .
we consider here the widely used gaussian form factor , as well as a family of lorentzian functions which allow us to interpolate between soft form factors and sharp njl - like cutoffs .
finally , our main results and conclusions are outlined in sec .
let us begin by stating the euclidean action for the nonlocal chiral quark model in the case of two light flavors , @xmath4 here @xmath5 is the current quark mass , which is assumed to be equal for @xmath6 and @xmath7 quarks .
as mentioned in the introduction , separable nonlocal interactions might be introduced in two alternative ways . in what follows , we will call them `` scheme i '' and `` scheme ii '' . these schemes
are characterized by the form of the nonlocal currents @xmath8 in eq .
( [ action ] ) . in the case of scheme
i @xcite , the effective interactions are based in an instanton liquid picture of qcd .
the currents can be written as @xmath9 where we have defined @xmath10 , @xmath11 being the pauli matrices acting on flavor space .
on the other hand , scheme ii @xcite arises from a separable form of the effective one - gluon exchange picture . in this case
the nonlocal currents @xmath8 are given by @xmath12 the functions @xmath13 and @xmath14 in eqs .
( [ uno ] ) and ( [ cuoge ] ) , respectively , are nonlocal form factors characterizing the corresponding interactions .
it is convenient to translate them into momentum space .
since lorentz invariance implies that they can only be functions of @xmath15 , we will use for the fourier transforms of these regulators the forms @xmath16 and @xmath17 from now on . a schematic representation of the four quark interactions in both schemes is given in fig . 1 .
in order to deal with meson degrees of freedom , one can perform a standard bosonization of the theory .
this is done by considering the corresponding partition function @xmath18 $ ] , and introducing auxiliary fields @xmath19 , where @xmath20 and @xmath21 are scalar and pseudoscalar mesons , respectively . integrating out the quark fields we get @xmath22 \ , \end{aligned}\ ] ] where @xmath23 the operator @xmath24 reads , in momentum space , @xmath25 where @xmath26 & & { \rm ( scheme\ ii ) } \ \ .
\\ \end{array } \right.\end{aligned}\ ] ] at this stage we assume that the @xmath27 field has a nontrivial translational invariant mean field value @xmath28 , while the mean field values of the pseudoscalar fields @xmath29 are zero .
thus we write @xmath30 replacing in the bosonized effective action , and expanding in powers of the meson fluctuations , we get @xmath31 here the mean field action per unit volume reads @xmath32 + \frac{\bar \sigma^2}{2 g } \ , \end{aligned}\ ] ] where @xmath33 .
the quadratic terms are given by @xmath34 \
, \ ] ] where @xmath35}{\left [ ( q^+)^2 + \sigma^2(q^+ ) \right ] \left [ ( q^-)^2 + \sigma^2(q^-)\right ] } \label{vamopincha}\end{aligned}\ ] ] with @xmath36 .
in order to find the mean field values @xmath28 , one has to minimize the action @xmath37 .
a straightforward exercise leads to the gap equation @xmath38 now the chiral condensates are given by the vacuum expectation values @xmath39 .
they can be easily obtained by performing the variation of @xmath40 $ ] with respect to the corresponding current quark masses .
we obtain @xmath41 in what follows we consider in particular the case of pseudoscalar mesons .
the expressions for scalar mesons are completely equivalent , just replacing the upper index `` @xmath0 '' by `` @xmath42 '' where it corresponds . the pion mass can be obtained by solving the equation g^-(-m_^2 ) = 0 .
[ pieq ] now , one has to perform a wave function renormalization of the pseudoscalar fields .
the renormalized pion field @xmath43 is defined by fixing the residue of @xmath44 at the pion pole , i.e. by demanding that in the vicinity of the pole the corresponding contribution to the quadratic lagrangian is given by @xmath45 in this way , one obtains @xmath46 finally the @xmath47 coupling constant @xmath48 is given by the residue of the pion propagator at the pole , leading to @xmath49 by definition the pion weak decay constant @xmath50 is given by the matrix element of the axial current @xmath51 between the vacuum and the renormalized one - pion state at the pion pole : @xmath52 in order to obtain an explicitly expression for the axial current , we have to `` gauge '' the effective action @xmath53 by introducing a set of axial gauge fields @xmath54 . for a local theory this `` gauging '' procedure
is usually done by performing the replacement @xmath55 in the present case owing to the nonlocality of the involved fields one has to perform an additional replacement , namely @xmath56 here @xmath57 and @xmath58 are the integration variables in the definitions of the nonlocal currents ( see eqs .
( [ uno ] ) and ( [ cuoge ] ) ) , and the function @xmath59 is defined by @xmath60 \ , \ ] ] where @xmath61 runs over an arbitrary path connecting @xmath62 with @xmath57 .
this corresponds to the introduction of a parallel transport @xcite .
once the gauged effective action is built , it is easy to get the axial current as the derivative of this action with respect to @xmath63 , evaluated at @xmath64 . performing the derivative of the resulting expressions with respect to the renormalized meson fields
, we can finally identify the corresponding meson weak decay constants .
after a rather lengthy calculation we obtain @xmath65 where @xmath66 is given by @xmath67 with @xmath68}{\left [ ( q^+)^2 + \sigma^2(q^+ ) \right ] \left [ ( q^-)^2 + \sigma^2(q^-)\right ] } \nonumber \\
k(p^2 ) & = & 8 \,n_c \int \frac{d^4 q}{(2 \pi)^4}\ h^2(q^+,q^- ) \ ; \frac { \left [ ( q^+ \cdot q^- ) + \sigma(q^+ ) \sigma(q^-)\right]}{\left [ ( q^+)^2 + \sigma^2(q^+ ) \right ] \left [ ( q^-)^2 + \sigma^2(q^-)\right ] } \label{integrals}\end{aligned}\ ] ] ( here , as before , @xmath69 ) . notice that in terms of the functions @xmath70 and @xmath71 eqs .
( [ gapeq ] ) and ( [ pieq ] ) can be written as @xmath72 replacing these expressions in eq .
( [ fpieq ] ) , we get the relation @xmath73 let us analyze in our context the anomalous decay of the @xmath75 meson into two photons .
in general , the amplitude for this process can be written as @xmath76 where @xmath77 is the fine structure constant and @xmath78 , @xmath79 stand for the momenta and polarizations of the outgoing photons respectively .
the partial decay width is then given by @xmath80 the coefficients @xmath81 are given by quark loop integrals .
besides the usual `` triangle '' diagram , given by a closed quark loop with one pion and two photon vertices , in the present nonlocal schemes one has a second contribution @xcite in which one of the quark - photon vertices arises from the gauge fields entering the nonlocal currents [ see eq .
( [ gauge ] ) ] .
the sum of both contributions leads to @xmath82 where @xmath83 \ ; [ ( q+k_1)^2+\sigma^2(q+k_1 ) ] \ ; [ ( q - k_2)^2+\sigma^2(q - k_2)]}\nonumber \\ & & \times \left\ { \sigma(q)\ + \ \frac{q^2}{2}\ ; \left[\frac{[\sigma(q - k_2 ) - \sigma(q)]}{(k_2\cdot q ) } \ - \ \frac{[\sigma(q+k_1 ) - \sigma(q)]}{(k_1\cdot q ) } \right ] \right\}\ .
\label{iloop}\end{aligned}\ ] ] notice that for on - shell photons the above integral is a function of the scalar product @xmath84 only , and in euclidean space one has @xmath85 .
we start by discussing the goldberger - treiman ( gt ) relation , which states that in the chiral limit ( @xmath86 ) one has f_,0 g_q|q,0 = |_0 [ gt ] ( here , and in the following , subindices @xmath87 mean that quantities are evaluated in the chiral limit ) . in our case , from eqs .
( [ fpieq ] ) and ( [ jk ] ) one has @xmath88_{p^2=0 } \
= \ - \,\bar\sigma_0\ ; z_{\pi,0}^{1/2 } \;k'_0(0 ) \ , \label{gt1}\ ] ] where prime denotes derivation with respect to @xmath15 .
now , an explicit calculation shows that @xmath89 can be written in terms of the momentum - dependent constituent masses @xmath90 as @xmath91 ^ 2 } \ \ . \label{fpich}\ ] ] then , taking into account eqs .
( [ zpi ] ) and ( [ gpiqq ] ) , and noting that @xmath92 , one gets @xmath93 the gt relation ( [ gt ] ) follows immediately from ( [ gt1 ] ) and ( [ gt2 ] ) .
next , let us consider the gell - mann - oakes - renner ( gor ) relation , which can be obtained using eq .
( [ relation ] ) . by performing a chiral expansion , and keeping only the lowest nonzero order at both sides of this equation
, one has @xmath94 in addition , from eq .
( [ cond ] ) it is easy to see that @xmath95 thus using the gt relation one arrives to @xmath96 which is the form taken by the gor relation in the isospin limit . finally , we study the anomalous effective coupling @xmath81 in the chiral limit . considering eq .
( [ iloop ] ) , one can perform an expansion in powers of the momenta @xmath97 and @xmath98 and take the limit @xmath99 .
this leads to @xmath100 now , for @xmath101 , using the gt relation one gets @xmath102 which is the expected result according to low energy theorems and chiral perturbation theory .
we stress that the derivation of the above chiral relations holds for both schemes i and ii .
in what follows we consider two types of form factors which have been frequently used in the literature .
one is the gaussian form factor , @xmath103 ^ 2 = \exp ( - p^2/\lambda^2 ) \ .\ ] ] the other is the lorenztian form factor @xmath104 ^ 2 = \frac{1}{1 + \left(p^2/\lambda^2\right)^n } \ , \ ] ] where @xmath105 is an integer that allows us to interpolate between soft form factors ( such as the gaussian one ) and sharp form factors as e.g. a step function @xmath106 . with the choice @xmath107 ^ 2 $ ] , both schemes i and ii lead to the same expressions for the mean field quantities .
moreover , in the chiral limit , the expressions for the pion decay constant converge in both cases to the result in eq .
( [ fpich ] ) .
this simplifies the comparison of the present type of nonlocal models with other frameworks such as instantaneous nonlocal schemes and the usual njl model . in the following subsection we will present a comparative analysis , considering in particular the chiral limit .
results corresponding to the case of finite quark masses will be discussed in subsection b. in this limit the pion is massless .
moreover , a simple dimensional analysis shows that , for the kind of models under study , the chiral condensate can be expressed as @xmath108^{1/3 } \ = \
f_{\pi,0 } \ { \cal f}(g \,\lambda^2 ) \ .\ ] ]
thus , for a given value of @xmath89 the chiral condensate only depends on a single dimensionless parameter , @xmath109 .
this allows to choose a typical value for the pion decay constant in the chiral limit , say @xmath110 mev , and study the behavior of the chiral condensate as a function of @xmath109 for various models .
we have performed this analysis , considering the range of values for which the different approaches are compatible with the empirical bounds @xmath111 to @xmath112 mev @xcite .
our results are displayed in fig .
2 . in the upper panel
we show the curves corresponding to covariant models , with different form factors . for comparison , in the lower panel
we show the results obtained for the case of instantaneous nonlocal models @xcite , and for the usual njl model . for different models in the chiral limit .
upper and lower panels correspond to covariant and instantaneous interactions , respectively . ]
let us start by analyzing the results in the upper panel of fig . 2 .
we observe that in all cases it is possible to choose a value of @xmath109 so that the corresponding condensate falls within the empirical range .
for the case of smooth form factors ( g and l2 ) we note , however , that condensates in the lower half of the empirical range require quite large values of @xmath109 .
an important observation is that when the form factors become sharper ( ln with increasing n ) the corresponding curves approach the values obtained for the step form factor .
this means that although eq .
( [ fpich ] ) includes a term with a derivative of the form factor , which naively should be divergent for @xmath113 , this limit is well defined for @xmath89 . on the other hand , as it can be seen in the upper panel of fig . 2 , in these models
the step limit does not match the curve obtained for the 4dnjl model , i.e. the njl model regularized by a four - dimensional covariant cutoff .
this fact can be understood as follows . from eqs .
( [ cond ] ) and ( [ fpich ] ) , the explicit form of the function @xmath114 for a covariant model is given by @xmath115 where @xmath116 , and @xmath117 ^ 2 } \ \ .
\label{deriv}\end{aligned}\ ] ] here we have defined @xmath118 ^ 2 $ ] , whereas @xmath119 .
the value of @xmath62 depends implicitly on the dimensionless parameter @xmath109 through the gap equation @xmath120 where @xmath121 in general , the integrals @xmath122 can not be performed analytically .
however , they can be evaluated for the case of a step form factor by considering the function @xmath123 and taking the limit @xmath124 . by doing this
, we find that in the step limit the integrals @xmath125 and @xmath126 are equal to each other , and they are also coincident with their 4dnjl model counterparts ( the latter can be found e.g. in ref .
the actual results are @xmath127 \
.\end{aligned}\ ] ] on the other hand , for @xmath128 we get @xmath129 where @xmath130 \ , \qquad \qquad \delta
i_2(x ) = \frac{n_c}{4 \pi^2 } \
\frac{1}{\,4\,(1+x^2 ) } \ \ .\end{aligned}\ ] ] due to the presence of the @xmath131 contribution , the step limit of the covariant nonlocal models does not match the 4dnjl result ( as shown in fig . 2 ) .
it is not hard to check that @xmath131 comes from the integral of the term with the derivative in eq .
( [ deriv ] ) , which is nonzero only in the `` surface '' region @xmath132 .
the contribution of this term vanishes in the 4dnjl according to the usual regularization procedure @xcite .
it is interesting to compare the results obtained within covariant models with those provided by instantaneous nonlocal models @xcite .
the latter are shown in the lower panel of fig . 2 .
we observe that , in contrast with covariant models , instantaneous nonlocal theories with soft form factors such as g and l2 are not able to yield quark condensates that fall within the empirical range .
this can be achieved only for rather sharp form factors , like e.g. l5 .
another important difference is that for instantaneous nonlocal models the curves obtained for ln do converge to the corresponding njl results ( i.e.the results obtained within the njl model with three - dimensional covariant cutoff , 3dnjl ) .
it is worth to mention , however , that for this type of models @xmath50 is frame dependent , owing to the noncovariance . here , to obtain the results displayed in the lower panel of fig . 2 we have used the pion rest frame , which is the only frame where low energy theorems , such as the gt relation , are fulfilled .
in the case of finite current quark masses , the model parameters @xmath5 , @xmath133 and @xmath134 are usually determined by fitting the pion mass and decay constant to their empirical values @xmath135 mev and @xmath136 mev , and fixing the chiral condensate at some phenomenologically acceptable value .
since this last quantity is not well determined , we consider here four representative values within the empirical range , namely @xmath137 , 220 , 240 and 260 mev .
it should be noticed that , in the case of finite @xmath5 , for schemes i and ii one gets different expressions for the integrals in eqs .
( [ vamopincha ] ) and ( [ integrals ] ) , which determine the values of @xmath138 and @xmath50 .
therefore , the fitted model parameters are also expected to be different in both schemes .
our results are shown in table i , where we quote the parameter values for schemes i and ii , considering different form factors and quark condensates .
interestingly , the current quark mass @xmath5 is basically equal for both schemes , while the remaining two parameters ( @xmath109 and @xmath134 ) might be rather different , especially in the case of soft form factors and low absolute values of the quark condensate .
perhaps the most striking case corresponds to the l2 form factor : for scheme ii it is not even possible to find a parameter set leading to a condensate of either @xmath139 mev or @xmath140 mev .
for the gaussian form factor ( i.e. the other `` soft '' function considered here ) , although we do find compatible parameter sets , the obtained values of @xmath134 in scheme ii for the case of low condensates are too low to be reliable for phenomenological applications .
in table ii we quote the values obtained for @xmath141 and @xmath142 , together with the position of the first pole of the quark propagator , for both covariant schemes i and ii , and different form factors and quark condensates . as discussed in the literature @xcite , the absence of purely imaginary poles in the quark propagator can be understood as a realization of quark confinement ( we are working here in euclidean space , thus a purely imaginary pole corresponds to a real pole in minkowski space ) .
we remark that in the nonlocal covariant models studied here , instead of one single purely imaginary pole as in the case of the usual njl model one has in general an arbitrary number of purely imaginary and/or complex poles @xcite ( we denote by `` complex pole '' a pole which has _ both _ non - vanishing real and imaginary components ) .
in fact , previous analysis show that for most applications only the first pole , i.e. the closest one to the imaginary axis , is phenomenologically relevant @xcite . in the case of scheme
i , as one can see from table ii , among the cases considered the first pole is complex only for a condensate of 200 mev and form factors g and l10 . moreover ,
the case of l10 is not phenomenologically viable in view of the low value of @xmath141 .
thus , we conclude that if one wants to avoid ( following confinement arguments ) the presence of low lying purely imaginary poles , the only acceptable situation for scheme i is that which corresponds to a soft ( gaussian ) form factor and a low value of @xmath143 .
if a low lying purely imaginary pole is admitted , then scheme i with a gaussian form factor and somewhat larger absolute values of the condensate can be accepted .
sharper form factors would be excluded , leading in all cases to too small values of @xmath141 . in the case of scheme
ii , one observes that sharp form factors also tend to yield small values of @xmath141 , the sole exceptions being the cases l5 and l10 with @xmath144 mev . in these cases , however , the situation is also problematic , since the first poles , although complex , have a very small imaginary part .
this implies the existence of a very low @xmath1 pseudo - threshold , which would require additional prescriptions to preserve meson stability already at energies of a few hundreds of mev @xcite .
given the above discussed limitations of scheme ii for soft form factors and low absolute values of the condensate , we conclude that the sets corresponding to g and l2 form factors and values of @xmath143 between 240 and 260 mev would provide the more acceptable results .
notice that for these limiting values of the condensate the lowest pole turns out to be complex ( `` confinement '' situation ) and purely imaginary , respectively .
concerning the values obtained for @xmath142 , it is seen that in all cases they are compatible with the experimental range , @xmath145 mev@xmath146 . now in table iii we quote the parameter values and the results obtained for @xmath141 and @xmath142 in the 4dnjl model .
we recall that in this case the propagator has a single pole , located at the constituent mass @xmath147 on the imaginary axis .
notice that , as commented in the introduction , the model does not provide a proper description of the @xmath148 decay width .
phenomenologically acceptable values of @xmath142 can be obtained instead if the cutoff is ( ad - hoc ) relaxed to infinity in the triangle loop integrals .
ccccccccc @xmath149 & form factor & & & + & ' '' '' & @xmath5 & @xmath150 & @xmath134 & & @xmath5 & @xmath150 & @xmath134 + 200 & g & 9.7 & 18.82 & 651.9 & & 9.8 & 71.11 & 459.7 + & l2 & 9.7 & 12.45 & 539.9 & & & & + & l5 & 9.6 & 8.71 & 799.0 & & 9.6 & 13.38 & 660.2 + & l10 & 9.8 & 7.57 & 885.0 & & 9.7 & 10.68 & 716.5 + 220 & g & 7.4 & 16.98 & 772.0 & & 7.4 & 29.06 & 604.0 + & l2 & 7.4 & 10.98 & 642.2 & & & & + & l5 & 7.4 & 8.40 & 924.4 & & 7.4 & 10.34 & 790.3 + & l10 & 7.5 & 7.47 & 1019.3 & & 7.4 & 9.06 & 843.2 + 240 & g & 5.8 & 15.82 & 902.4 & & 5.8 & 20.65 & 752.2 + & l2 & 5.8 & 10.14 & 751.8 & & 5.8 & 16.06 & 586.8 + & l5 & 5.8 & 8.20 & 1059.2 & & 5.8 & 9.27 & 925.7 + & l10 & 5.8 & 7.39 & 1163.0 & & 5.8 & 8.36 & 978.6 + 260 & g & 4.6 & 15.08 & 1042.2 & & 4.6 & 17.53 & 903.4 + & l2 & 4.6 & 9.61 & 868.0 & & 4.6 & 11.77 & 736.1 + & l5 & 4.6 & 8.07 & 1202.8 & & 4.6 & 8.73 & 1067.7 + & l10 & 4.6 & 7.34 & 1315.6 & & 4.6 & 7.98 & 1122.5 + ccccccccc @xmath149 & form factor & & & + & ' '' '' & @xmath151 & first pole & @xmath152 & & @xmath151 & first pole & @xmath152 + 200 & g & 318 & 180 + 454i & 7.43 & & 1356 & 499 + 241i & 7.16 + & l2 & 296 & 284i & 7.37 & & & & + & l5 & 181 & 191i & 7.57 & & 366 & 640 + 169i & 7.37 + & l10 & 149 & 875 + 131i & 7.39 & & 302 & 711 + 98i & 7.24 + 220 & g & 282 & 356i & 7.50 & & 620 & 421 + 372i & 7.45 + & l2 & 259 & 259i & 7.49 & & & & + & l5 & 173 & 180i & 7.74 & & 276 & 757 + 217i & 7.45 + & l10 & 145 & 152i & 7.64 & & 248 & 835 + 120i & 7.46 + 240 & g & 255 & 288i & 7.57 & & 424 & 288 + 510i & 7.49 + & l2 & 233 & 237i & 7.55 & & 475 & 485 + 304i & 7.48 + & l5 & 166 & 153i & 7.87 & & 238 & 243i & 7.52 + & l10 & 142 & 147i & 7.96 & & 221 & 968 + 142i & 7.57 + 260 & g & 235 & 255i & 7.61 & & 339 & 430i & 7.52 + & l2 & 216 & 219i & 7.65 & & 330 & 323i & 7.50 + & l5 & 160 & 172i & 7.94 & & 215 & 220i & 7.61 + & l10 & 139 & 143i & 8.11 & & 204 & 1110 + 165i & 7.66 + .model parameters and values of @xmath141 and @xmath152 for representative values of the chiral condensate in the 4dnjl model ( @xmath152 given in mev@xmath153 , @xmath5 , @xmath134 and @xmath141 given in mev ) [ cols="^,^,^,^,^,^,^",options="header " , ]
we have presented a comparative description of chiral quark models which include nonlocal covariant four fermion interactions .
these approaches inspired in features of the underlying qcd theory represent natural extensions of the usual njl model and allow to overcome some of its difficulties , in particular those related with the regularization of ultraviolet divergences .
we have concentrated here in two alternative ways of introducing the nonlocality , which we have called scheme i and scheme ii . using a common notation , we have derived the main expressions to obtain the values of the pion mass , the pion decay constant and some important mean field quantities in terms of the model parameters @xmath5 , @xmath133 and @xmath134 .
in addition , for both schemes we have shown the validity of various well - known relations in the chiral limit .
we have performed numerical calculations for gaussian and lorentzian form factors , studying the behavior of model parameters , constituent masses and propagator poles for given values of the chiral @xmath154 condensate within a phenomelogically acceptable range .
the parameters have been fitted so as to reproduce the empirical values of the pion mass and decay constant . in the chiral limit ,
the results for covariant nonlocal models have been compared to those obtained within instantaneous nonlocal models .
we have found that , in contrast to covariant models , the latter can be hardly accommodated to yield phenomenologically acceptable values of the @xmath154 condensate . in the case of finite current quark masses , it has been found that sharp form factors such as l5 and l10 are also problematic , leading to either too low values of constituent quark masses or to the appearance of a low @xmath154 pseudothreshold , which would need additional prescriptions to preserve meson stability . on the other hand ,
the results seem to favor quark interactions driven by smooth form factors ( in our analysis , g and l2 ) . in the case of scheme
ii , large values of @xmath155 are found to be preferred , since lower condensates lead to too low values of the `` cutoff '' @xmath134 .
for scheme i , instead , lower absolute values of the condensate seem to be favored in order to avoid relatively small constituent quark masses . as it is shown in table
ii , in some of these cases one finds that the first pole of the quark propagator has both a real and an imaginary part , which can be understood as a manifestation of confinement .
this pole structure is a characteristic of the form factor under consideration , and can be extended to more involved approaches such as e.g. the instanton liquid model .
finally , we have considered the case of the njl model regularized by a four dimensional cutoff ( 4dnjl ) , analyzing its deviation from a nonlocal model with a step - like form factor .
the authors are glad to thank d. aguilera , d. blaschke and h. grigorian for useful comments and discussions .
this work has been supported in part by conicet and anpcyt ( argentina ) , under grants pip 02368 , pip 6084 , pip 6009 , pict00 - 03 - 08580 and pict02 - 03 - 10718 .
s. p. klevansky , rev .
phys . * 64 * , 649 ( 1992 ) . t. hatsuda and t. kunihiro , phys .
* 247 * , 221 ( 1994 ) .
j. bijnens , e. de rafael and h. q. zheng , z. phys .
c * 62 * , 437 ( 1994 ) ; j. bijnens and j. prades , phys .
b * 320 * , 130 ( 1994 ) ; j. bijnens and j. prades , z. phys .
c * 64 * , 475 ( 1994 ) .
t. meissner , e. ruiz arriola and k. goeke , z. phys .
a * 336 * , 91 ( 1990 ) .
v. dmitrasinovic , h. j. schulze , r. tegen and r. h. lemmer , annals phys .
* 238 * , 332 ( 1995 ) ; e. n. nikolov , w. broniowski , c. v. christov , g. ripka and k. goeke , nucl . phys . a * 608 * , 411 ( 1996 ) ; c. schuren , f. doring , e. ruiz arriola and k. goeke , nucl .
a * 565 * , 687 ( 1993 ) .
g. ripka , _ quarks bound by chiral fields _ ( oxford university press , oxford , 1997 ) .
t. schafer and e. v. shuryak , rev .
phys . * 70 * , 323 ( 1998 ) . c. d. roberts and a. g. williams , prog .
phys . * 33 * , 477 ( 1994 ) ; c. d. roberts and s. m. schmidt , prog .
* 45 * , s1 ( 2000 ) .
m. b. parappilly , p. o. bowman , u. m. heller , d. b. leinweber , a. g. williams and j. b. zhang , phys .
d * 73 * , 054504 ( 2006 ) .
e. ruiz arriola and l. l. salcedo , phys .
b * 450 * , 225 ( 1999 ) .
d. blaschke , y. l. kalinovsky , g. roepke , s. m. schmidt and m. k. volkov , phys .
c * 53 * , 2394 ( 1996 ) .
g. ripka , nucl .
a * 683 * , 463 ( 2001 ) ; r. s. plant and m. c. birse , nucl .
a * 703 * , 717 ( 2002 ) .
r. d. bowler and m. c. birse , nucl .
phys . a * 582 * , 655 ( 1995 ) .
r. s. plant and m. c. birse , nucl .
a * 628 * , 607 ( 1998 ) .
m. stingl , phys .
* d 34 * , 3863 ( 1986 ) [ erratum - ibid .
* d 36 * , 651 ( 1987 ) ] ; h. j. munczek , phys .
* b 175 * , 215 ( 1986 ) ; c. j. burden , c. d. roberts and a. g. williams , phys . lett . *
b 285 * , 347 ( 1992 ) ; g. krein , c. d. roberts and a. g. williams , int .
j. mod .
phys . * a 7 * , 5607 ( 1992 ) ; d. blaschke , g. burau , y. l. kalinovsky , p.
maris and p. c. tandy , int.j .
a * 16 * , 2267 ( 2001 ) .
m. buballa and s. krewald , phys .
b * 294 * 19 ( 1992 ) .
h. ito , w. buck and f. gross , phys .
c * 43 * ( 1991 ) 2483 ; h. ito , w. w. buck and f. gross , phys .
c * 45 * 1918 ( 1992 ) .
s. m. schmidt , d. blaschke and y. l. kalinovsky , phys .
c * 50 * 435 ( 1994 ) .
a. scarpettini , d. gomez dumm and n. n. scoccola , phys . rev .
d * 69 * , 114018 ( 2004 ) .
s. noguera , arxiv : hep - ph/0502171 ; s. noguera and v. vento , arxiv : hep - ph/0505102 .
b. golli , w. broniowski and g. ripka , phys .
b * 437 * , 24 ( 1998 ) ; w. broniowski , b. golli and g. ripka , nucl .
phys . a * 703 * , 667 ( 2002 ) .
a. h. rezaeian , n. r. walet and m. c. birse , phys .
c * 70 * , 065203 ( 2004 ) ; a. h. rezaeian and h. j. pirner , nucl . phys .
a * 769 * , 35 ( 2006 ) .
i. general , d. gomez dumm and n. n. scoccola , phys .
b * 506 * , 267 ( 2001 ) ; d. gomez dumm and n. n. scoccola , phys .
d * 65 * , 074021 ( 2002 ) ; phys .
c * 72 * , 014909 ( 2005 ) ; r. s. duhau , a. g. grunfeld and n. n. scoccola , phys .
d * 70 * , 074026 ( 2004 ) ; d. gomez dumm , d. b. blaschke , a. g. grunfeld and n. n. scoccola , phys .
d * 73 * , 114019 ( 2006 ) .
j. w. bos , j. h. koch and h. w. l. naus , phys .
c * 44 * , 485 ( 1991 ) . | we present a comparative analysis of chiral quark models which include nonlocal covariant four - fermion couplings .
we consider two alternative ways of introducing the nonlocality , as well as various shapes for the momentum - dependent form factors governing the effective interactions . in all cases
we study the behavior of model parameters and analyze numerical results for constituent quark masses and quark propagator poles .
advantages of these covariant nonlocal schemes over instantaneous nonlocal schemes and the standard njl model are pointed out . |
in recent years there has been a growing attention to braneworld scenarios , in which our observable universe is realized as a brane embedded in a higher dimensional bulk spacetime @xcite . among various models a simple setup called the randall - sundrum type ii model @xcite is of particular interest because , despite gravity propagating into an infinite extra dimension , four - dimensional general relativity can be recovered on the brane due to the warped geometry of anti - de sitter spacetime @xcite .
cosmological consequences of the randall - sundrum type braneworld have been widely discussed @xcite , especially regarding cosmological perturbations @xcite .
the generation and evolution of perturbations are among the most important issues in cosmology because of their direct link to cosmological observations such as the stochastic gravitational wave background and the temperature anisotropy of the cosmic microwave background ( cmb ) , by which we can probe the early universe .
while the cosmological perturbation theory in the conventional four - dimensional universe is rather established @xcite , calculating cosmological perturbations in the braneworld still remains to be a difficult problem .
although some attempts have been made concerning scalar perturbations @xcite , further progress is awaited to give a clear prediction about the cmb anisotropy in the randall - sundrum braneworld .
almost the same is true for gravitational wave ( tensor ) perturbations @xcite .
however , since their generation and evolution depend basically only on the background geometry , they are slightly easier to handle . during inflation super - horizon
gravitational wave perturbations are generated from vacuum fluctuations of gravitons .
the pure de sitter braneworld is the special case that allows definite analytical computation of quantum fluctuations .
thanks to the symmetry of the de sitter group , the perturbation equation becomes separable and can be solved exactly @xcite .
the time variation of the hubble parameter generally causes mixing of a massless zero mode and massive kaluza - klein modes , and this effect was investigated based on the `` junction '' models , which assume an instantaneous transition from a de sitter to a minkowski brane @xcite or to another de sitter brane @xcite .
these works discussed the quantum - mechanical generation of gravitational waves within such limited and simplified models .
as for the classical evolution of gravitational waves during the radiation dominated epoch , several studies have been done @xcite , assuming that a given single initial mode for each comoving wave number dominates .
the focus of these works is mainly on the evolution of modes which re - enter the horizon in the high - energy regime . while the late time evolution is worked out analytically in refs .
@xcite by resorting to low - energy approximation methods . in this paper
we consider the generation of primordial gravitational waves during inflation in more general models of the randall - sundrum type .
for definiteness , we adopt a simple setup in which both initial and final phases are described by de sitter braneworlds .
two de sitter phases with different values of the hubble expansion rate are smoothly interpolated .
this work is an extension of the work on the `` junction '' models by the present authors in collaboration with h. kudoh @xcite . in the previous analysis
the transition of the hubble rate was abrupt and the gap was assumed to be infinitesimal .
here we extend the previous results to more general models with smooth transition by using numerical calculations with a refined formulation .
this paper is organized as follows . in the next section
we briefly summarize past studies @xcite on the generation of gravitational waves via quantum fluctuations during inflation in the randall - sundrum braneworld , emphasizing the mapping formula introduced in ref . @xcite . in sec .
[ formulation ] we describe our numerical scheme to investigate the generation of gravitational waves , and then in sec .
[ results ] we present results of our calculations .
section [ discussion ] is devoted to discussion .
the background spacetime that we consider is composed of a five - dimensional ads bulk , whose metric is given in the poincar coordinates by @xmath0 and a friedmann brane at @xmath1 . here
@xmath2 is the bulk curvature length , which is constrained by table - top experiments as @xmath3 mm @xcite .
first let us consider the generation of gravitational waves from pure de sitter inflation on the brane @xcite .
the coordinate system appropriate for the present situation is @xmath4,\end{aligned}\ ] ] which is obtained from eq .
( [ poincare ] ) by a coordinate transformation @xmath5 where @xmath6 is an arbitrary constant and @xmath7 is the conformal time , which is negative .
the de sitter brane is located at @xmath8 and the hubble parameter on the brane is given by @xmath9 the gravitational wave perturbations are described by the metric @xmath10 + d\xi^2 \right\},\end{aligned}\ ] ] and we decompose the perturbations into the spatial fourier modes as @xmath11 where @xmath12 is the transverse - traceless polarization tensor and @xmath13 is the fundamental mass scale which is related to the four - dimensional planck mass @xmath14 by @xmath15 .
the prefactor @xmath16 is chosen so that the kinetic term for @xmath17 in the effective action is canonically normalized .
hereafter we will suppress the subscript @xmath18 .
the linearized einstein equations give the following klein - gordon - type equation for @xmath19 : @xmath20\phi=0 , \label{k - g_for_ds}\end{aligned}\ ] ] which is obviously separable .
assuming the @xmath21-symmetry across the brane , the boundary condition on the brane is given by @xmath22 where @xmath23 is an unit normal to the brane .
equation ( [ k - g_for_ds ] ) admits one discrete zero mode , @xmath24 where @xmath25 is a normalization constant , which will be fixed later , as well as massive kaluza - klein ( kk ) modes , @xmath26 where @xmath27 and @xmath28 obey @xmath29\psi_{\nu}(\eta)=0,\label{kk - eta } \\ & & \left [ \sinh^3\!\xi\frac{\partial}{\partial\xi } { 1\over \sinh^3\!\xi}\frac{\partial}{\partial\xi } + \nu^2+\frac{9}{4 } \right]\chi_{\nu}(\xi)=0,\label{kk - xi}\end{aligned}\ ] ] and the corresponding kaluza - klein mass is given by @xmath30 due to the mass gap @xmath31 @xcite between the zero mode and the lowest kk mode , all the massive kk modes decay on the brane at super - horizon scales during inflation .
the explicit forms of the kk mode functions are shown in appendix [ kkmodefn ] .
in inflationary cosmology , fluctuations in the graviton field ( and other fields such as the inflaton ) are considered to be generated quantum - mechanically . following the standard canonical quantization scheme we quantize the graviton field . treating @xmath19 as an operator
, it may be expanded as @xmath32 the annihilation and creation operators @xmath33 and @xmath34 with @xmath35 or @xmath36 satisfy the usual commutation relations and the modes @xmath37 and @xmath38 are normalized so that they satisfy @xmath39 where @xmath40 is the wronskian defined by @xcite @xmath41 from the conditions ( [ wronskian_condition ] ) we can determine the normalization of the zero mode as @xmath42 with @xmath43^{-1/2 } , \end{aligned}\ ] ] which is the same as that introduced in refs .
@xcite , and behaves like @xmath44 for @xmath45 and @xmath46 for @xmath47 .
this function plays an important roll throughout the present paper .
the expectation value of the squared amplitude of the vacuum fluctuation in the zero mode is given by @xmath48 where the expression in the second line is obtained by evaluating the perturbation in the super - horizon regime / at a late time , and hence this is the amplitude of the _ growing mode_. in terms of the power spectrum defined by @xmath49 we have @xmath50 where we have used @xmath51 .
this is the standard flat spectrum up to the overall factor @xmath52 . in conventional four - dimensional cosmology the power spectrum of the primordial gravitational waves from de sitter inflation
is given by @xcite @xmath53 thus just by rescaling the amplitude in four - dimensional cosmology as @xmath54 the braneworld result ( [ desitter_5d ] ) is exactly obtained .
pure de sitter inflation on the brane described in the previous subsection is a special case where the amplitude of the growing zero mode can be obtained completely analytically . as a next step to understand gravitational waves from more general inflation with @xmath55 const , a discontinuous change in the hubble parameter was considered in ref .
in such a `` junction '' model the hubble parameter is given by @xmath56 and @xmath57 is assumed .
the evolution of gravitational wave perturbations can be analyzed by solving the equation of motion backward using the advanced green s function @xcite .
a set of mode functions can be constructed for the initial de sitter stage , and another for the final de sitter stage , either of which forms a complete orthonormal basis for the graviton wave function . the final ( growing ) zero mode may be written as a linear combination of the initial zero and kk modes .
the bogoliubov coefficients give the creation rate of final zero mode gravitons from the initial vacuum fluctuations in the kaluza - klein modes as well as in the zero mode .
thus the power spectrum may be written as a sum of two separate contributions , @xmath58 where @xmath59 and @xmath60 are the parts coming from the initial zero and kaluza - klein modes , respectively .
an important result brought by analyzing this junction model is that @xmath61 is well approximated by the rescaled spectrum @xmath62 obtained by using a simple map speculated by the exact result of pure de sitter inflation [ eq .
( [ mapds ] ) ] .
more precisely , let @xmath63 be the gravitational wave spectrum evaluated in the standard four - dimensional inflationary universe with the same time dependence of the hubble parameter ( [ dis_hub ] ) , and then the rescaled spectrum is defined by @xmath64 comparing the rescaled spectrum with @xmath61 obtained from a five - dimensional calculation , one finds that the difference between these two is suppressed to be second order like @xcite @xmath65 it might be worth noting here that the kk contribution @xmath60 is necessary for realizing this interesting agreement between the braneworld result and the ( basically ) four - dimensional result especially at high energies @xmath66 . the above agreement ( [ agreement ] ) raised a speculation that the mapping formula @xmath67 may work with good accuracy in more general inflation models with a smoothly changing expansion rate .
the central purpose of the present paper is to check whether this speculation is correct or not .
now let us explain the formulation that we use to study the generation of gravitational waves without assuming a pure de sitter brane .
our formulation is based on double null coordinates , which are presumably the most convenient for numerical calculations .
the metric ( [ poincare ] ) can be rewritten by using double null coordinates @xmath68 in the form of @xmath69 the trajectory of the brane can be specified arbitrarily by @xmath70 by a further coordinate transformation @xmath71 we obtain @xmath72 ^ 2}\left [ -q'(v)dudv+\delta_{ij}dx^idx^j \right],\end{aligned}\ ] ] where a prime denotes differentiation with respect to the argument .
now in the new coordinates the position of the brane is simply given by @xmath73 we will use this coordinate system for actual numerical calculations .
the induced metric on the brane is @xmath74 ^ 2}\left [ -q'(v)dv^2+\delta_{ij}dx^idx^j \right],\end{aligned}\ ] ] from which we can read off the conformal time @xmath7 and the scale factor @xmath75 , respectively , as @xmath76 and hence the hubble parameter on the brane is written as @xmath77,\end{aligned}\ ] ] or equivalently @xmath78 given the hubble parameter as a function of @xmath7 , one can integrate eqs .
( [ eta - v ] ) and ( [ q - h ] ) to obtain @xmath79 as a function of @xmath80 .
when the brane undergoes pure de sitter inflation ( and thus @xmath81 constant ) , the following relation between @xmath82 and @xmath83 will be useful : @xmath84^{1/2},\end{aligned}\ ] ] and @xmath85 the klein - gordon - type equation for a gravitational wave perturbation @xmath19 in the @xmath82 coordinates reduces to @xmath86\phi=0,\!\!\ !
\cr \label{kg_in_uv}\end{aligned}\ ] ] supplemented by the boundary condition @xmath87\phi\bigr\vert_{u = v}=0 . \label{bc}\end{aligned}\ ] ] the expression for the wronskian evaluated on a constant @xmath80 hypersurface is given by @xmath88^{3}\ !
( x\partial_uy^*\!-y^*\partial_u x),\cr\end{aligned}\ ] ] which is independent of the choice of the hypersurface .
when inflation on the brane deviates from pure de sitter one , the decomposition into the zero mode and kk modes becomes rather ambiguous .
for this reason we require the initial and final phases of inflation to be pure de sitter , though arbitrary cosmic expansion is allowed in the intermediate stage . in both de sitter phases
, @xmath89 can be fit by eq .
( [ qvequal ] ) , and @xmath90 and @xmath6 are determined , respectively .
hence we have two sets of de sitter coordinates @xmath91 and @xmath92 .
we distinguish the coordinates in the final phase by associating them with tilde . in the final de sitter
phase the mode will be well outside the horizon , and hence we expand the graviton field in terms of the growing and decaying zero mode solutions @xmath93 and @xmath94 as @xmath95 where @xmath96,\\ & & \tilde\phi_d : = { \rm re}\left[\tilde \phi_0\right],\end{aligned}\ ] ] and the mode functions with tilde are defined in the same way as @xmath37 and @xmath97 with the substitution of @xmath98 for @xmath99 .
it can be easily seen that the growing and the decaying modes are normalized as @xmath100 where subscript @xmath101 means that the expression is to be evaluated in the final de sitter phase .
notice that de sitter mode functions thus defined as functions of @xmath82 through @xmath99 ( or @xmath102 ) do not satisfy the equation of motion outside the initial ( or final ) de sitter phase .
back in the initial de sitter phase the graviton field can be expanded as @xmath103 we assume that initially the gravitons are in the de sitter invariant vacuum state annihilated by @xmath104 and @xmath105 , @xmath106 we would like to evaluate the expectation value of the squared amplitude of the vacuum fluctuation in the growing mode at a late time , @xmath107 or equivalently , the power spectrum , @xmath108 where @xmath109 is the hubble parameter in the final de sitter phase . in order to obtain the final amplitude , it is not necessary to solve the evolution of all the ( infinite number of ) degrees of freedom with their initial conditions set by eq .
( [ initial_mode_decompose ] ) .
in fact , we have only to solve the backward evolution of the final decaying mode , as explained below .
suppose that a solution @xmath110 is chosen so as to satisfy @xmath111 in the final de sitter phase . from the wronskian condition ( [ g - d - wronski - condi ] ) we see that @xmath112 where subscript @xmath113 means that the expression is to be evaluated in the initial de sitter phase .
in the above we used the fact that the wronskian is constant in time .
thus we obtain @xmath114,\end{aligned}\ ] ] from which we can calculate the power spectrum of the primordial gravitational waves .
it is obvious that the spectrum is written in the form of eq .
( [ sumspectrum ] ) with @xmath115 here we describe the algorithm that we employ to solve the backward evolution of the gravitational perturbations numerically .
we set the boundary conditions for the mode @xmath116 so as to be identical to a decaying zero mode @xmath94 at a time @xmath117 in the final de sitter phase .
we decompose @xmath116 as @xmath118 and solve the equation for @xmath119 instead of @xmath120 .
the equation of motion for @xmath121 is obtained as @xmath122\delta\phi \nonumber\\ & & \!\!\!\!=-\left[4\partial_u\partial_v + \frac{6}{q(v)-u}(\partial_v - q'(v)\,\partial_u ) + q'(v)\,k^2\right]\tilde\phi_d . \nonumber\\\label{evolution_eq}\end{aligned}\
] ] since both @xmath116 and @xmath94 satisfy the boundary condition of the form of eq .
( [ bc ] ) , the boundary condition for @xmath123 is also written as @xmath87\delta\phi\bigr\vert_{u = v}=0 . \label{bc2}\end{aligned}\ ] ] we immediately find that we do not have to solve the backward evolution in the final de sitter phase , since @xmath121 identically vanishes there . in some cases @xmath116
is not disturbed so much from its final configuration @xmath124 , especially when we discuss a small deviation from the pure de sitter case .
it is advantageous then to use a small quantity @xmath123 as a variable in numerical calculations .
of course , there is no problem even when @xmath123 does not stay small .
our scheme to obtain the values of @xmath121 at the initial surface @xmath125 is as follows .
we give the boundary conditions at @xmath117 as @xmath126 , where @xmath127 a sketch of the numerical grids is shown in fig .
[ fig : scheme.eps ] . here
@xmath128 is taken to be sufficiently large , and we use the same grid spacing @xmath129 both in the @xmath130 and @xmath80 directions . at a virtual site @xmath131 we set @xmath132 so that the boundary condition ( [ bc2 ] ) is satisfied at @xmath133 . from @xmath134 , @xmath135 , and @xmath136
we can determine the value of @xmath137 by using the equation of motion ( [ evolution_eq ] ) at @xmath133 , and then from @xmath135 , @xmath138 , and @xmath137 we can determine @xmath139 , and so on .
we repeat the same procedure in the subsequent time steps until we obtain @xmath140 at @xmath141 . in our model .
this is a plot for the model of fig .
[ fig : high1.eps].,height=287 ] as stated above , to make the problem well posed , we consider models which have initial and final de sitter phases . to perform numerical calculations , as a concrete example
, we adopt a simple toy model in which the hubble parameter is given by an analytic form @xmath142 where @xmath143 with @xmath144 and @xmath109 the initial and final values of the hubble parameter , @xmath145 is a parameter that controls the smoothness of the transition , and @xmath146 indicates a transition time . taking the initial time @xmath147 and the final time @xmath148 so that @xmath149 and @xmath150 , we have @xmath151 for @xmath152 and @xmath153 for @xmath154 .
the scale factor for this inflation model is given by @xmath155 -\bar h
\eta + \delta \eta_0 \right\}^{-1 } , \nonumber\\\end{aligned}\ ] ] where the integration constant is determined by imposing that @xmath156 at @xmath157 .
the slow - roll parameter , @xmath158 , is obtained as @xmath159-\bar h\eta+\delta \eta_0 \ } } { s [ \bar h \cosh((\eta-\eta_0)/s)-\delta \sinh((\eta-\eta_0)/s)]^2},\end{aligned}\ ] ] which takes maximum at @xmath160 .
the maximum value is @xmath161 the behavior of the hubble parameter and the slow - roll parameter is shown in fig .
[ fig : model.eps ] .
here we should mention the limitation of this simple toy model . for fixed values of @xmath144 , @xmath109 , and
@xmath162 one can prolong the period of the transition @xmath163 measured in conformal time by taking large @xmath145 .
however , using the definition of @xmath164 ( [ ep_max ] ) , the transition time scale in _ proper _ time is found to be given by @xmath165 .
therefore we can not change the transition time scale independently of the other parameters , @xmath144 , @xmath109 , and @xmath162 .
red diamonds ( upper ones ) indicate the result including the contributions from the initial kaluza - klein modes , while green diamonds ( lower ones ) represent the contribution from the initial zero mode . although the rescaled four - dimensional spectrum is shown by blue diamonds , they are almost hidden by the red ones since the result of the five - dimensional calculation is well approximated by the rescaled four - dimensional spectrum .
bottom panel : difference between the five - dimensional and rescaled four - dimensional spectra.,height=321 ] , but the parameters are different.,height=321 ] , but the parameters are different.,height=321 ] , but the parameters are different.,height=321 ] , but the parameters are different.,height=321 ] , but the parameters are different.,height=321 ] , but the parameters are different.,height=321 ] , but the parameters are different.,height=321 ] as a function of @xmath36 .
the parameters are given by those of fig .
[ fig : high1.eps ] and @xmath166.,height=132 ] .[tab : table1 ] parameters used for the numerical calculations presented in the figures . [ cols="<,>,>,>,>,>",options="header " , ] using various parameters shown in table .
[ tab : table1 ] we performed numerical calculations , the results of which are presented in figs .
[ fig : low1.eps][fig : high2.eps ] . numbers of grids are @xmath167 , and the grid separation @xmath129 is chosen to be @xmath168 depending on the energy scale of inflation .
note here that the step width in conformal time @xmath7 is given by @xmath169 [ eq .
( [ eta - v ] ) ] , and @xmath170 becomes smaller for a larger value of @xmath171 [ eq . ( [ q - h ] ) ] .
our choice of @xmath129 makes @xmath172 about the same size , @xmath173 , in all the calculations .
integration over @xmath36 is performed up to @xmath174
@xmath175 with an equal grid spacing of @xmath176
. the typical behavior of the integrand @xmath177 is shown in fig .
[ fig : kk.eps ] . in order to compare the five - dimensional power spectrum with a four - dimensional counterpart
, we solve the conventional evolution equation for gravitational waves in four - dimensions , @xmath178 where @xmath179 is given by the same function as used for the corresponding five - dimensional computation , and then we calculate the _ rescaled _ power spectrum obtained from the four - dimensional ` bare ' spectrum by using the mapping formula ( [ mapping ] ) .
we introduce a parameter that represents a difference between the five - dimensional power spectrum and the rescaled four - dimensional spectrum : @xmath180 in all the cases of our calculations we clearly see that @xmath181 and thus we conclude that _ the primordial spectrum of the gravitational waves in the braneworld is quite well approximated by the rescaled four - dimensional spectrum_. we also find that the difference becomes larger for larger values of the slow - roll parameter @xmath182 , but @xmath183 is no greater than @xmath184 even for models with @xmath185 .
notice that the universe is not inflating any more if @xmath182 is greater than unity . at high energies @xmath186 it can be seen that dependence of @xmath187 on @xmath171 is weak , while at low energies like @xmath188 , @xmath187 is further suppressed compared to the high energy cases .
contributions from the initial kk modes can be significant for small wavelength modes .
for example , for the modes with @xmath189 we see that @xmath190 at low energies , but the kk contribution can become as large as @xmath191 in our most ` violent ' model with @xmath192 and @xmath185 .
in this paper we have investigated the generation of gravitational waves during inflation on a brane and computed the primordial spectrum . extending the previous work @xcite ,
we have considered a model composed of the initial and final de sitter stages , and the transition region connecting them smoothly .
we have numerically solved the backward evolution of the final _ decaying _ mode to obtain the amplitude of the _ growing _ zero mode at a late time by making use of the wronskian method , we found that the power spectrum @xmath61 is well approximated by the rescaled spectrum @xmath62 basically calculated in standard four - dimensional inflationary cosmology . here the rescaling formula for the amplitude is given by a simple map @xmath193 , where @xmath194 is the normalization factor of the zero mode .
although the difference of the two spectra , @xmath195 , depends on the energy scale of inflation and the slow - roll parameter , our numerical analysis clearly shows that in any case the mapping formula works with quite good accuracy , yielding @xmath196 .
this implies that the mapping relation between the two spectra holds quite generally in the randall - sundrum braneworld , which we believe is a useful formula .
we have taken into account the vacuum fluctuations in initial kaluza - klein modes as well as the zero mode , and found that both of them contribute to the final amplitude of the zero mode .
the amount of the initial kk contribution can be large on small scales , and it increases as the energy scale of inflation @xmath171 becomes higher .
this gives rise to a quite interesting picture .
when the expansion rate changes during inflation , zero mode gravitons escape into the bulk as kk gravitons , but at the same time bulk gravitons come onto the brane to compensate for the loss , and these two effects almost cancel each other .
this seems to happen , irrespective of the energy scale , in a wide class of the inflation models even with a not - so - small slow - roll parameter .
it is suggested that this is not the case in the radiation dominated decelerating universe @xcite , where the decay into kk gravitons reduces the amplitude of the gravitational waves on the brane . the junction model of ref .
@xcite joining de sitter and minkowski branes also show the suppression of the gravitational wave amplitude at high energies . what is the reason for the remarkable agreement of the braneworld spectrum and the rescaled four - dimensional spectrum ?
extremely long wavelength modes , which leave the horizon during the initial de sitter stage much before the hubble parameter changes , have a squared amplitude of @xmath197 , and the amplitude of the perturbations stays constant during the subsequent stages limit , @xmath19=const is a growing solution of eq .
( [ kg_in_uv ] ) . ] . the same is true for four - dimensional inflationary cosmology , and so the mapping formula is applicable to these long wavelength modes . on the other hand , a mode
whose wavelength is much shorter than the hubble horizon scale at the transition time will feel the transition as adiabatic and hence the particle production is exponentially suppressed , @xmath198 leading to @xmath199 .
the same argument can be applied to the conventional cosmology .
thus it is not surprising that the mapping formula works for such short wavelength modes .
however , at present there seems no simple reason for the amplitude of the modes with @xmath200 to coincide with the rescaled one .
we thank shinji mukohyama for valuable discussions .
this work is supported in part by jsps research fellowships for young scientists no .
1642 , by grant - in - aid for scientific research , nos .
14047212 and 16740141 , and by that for the 21st century coe `` center for diversity and universality in physics '' at kyoto university , both from the ministry of education , culture , sports , science and technology of japan .
in the de sitter braneworld the kaluza - klein mode functions in the time direction satisfy eq .
( [ kk - eta ] ) and are given by @xmath201 where @xmath202 is the hunkel function .
they are normalized so that @xmath203 the mode functions in the extra direction satisfies eq .
( [ kk - xi ] ) and are given by @xmath204,\end{aligned}\ ] ] where @xmath205 and @xmath206 are the associated legendre functions .
the constants @xmath207 and @xmath208 are determined by the normalization condition @xmath209 and the boundary condition ( [ bc ] ) , respectively , as @xmath210^{-1/2 } , \\
c_2&= & \frac{p^{-1}_{-1/2+i\nu}(\cosh\xi_b)}{q^{-1}_{-1/2+i\nu}(\cosh\xi_b)}.\end{aligned}\ ] ] for a nice review , see , e.g. , r. maartens , living rev .
* 7 * , 7 ( 2004 ) [ arxiv : gr - qc/0312059 ] .
l. randall and r. sundrum , phys .
lett . * 83 * , 4690 ( 1999 ) [ arxiv : hep - th/9906064 ] .
j. garriga and t. tanaka , phys .
lett . * 84 * , 2778 ( 2000 ) [ arxiv : hep - th/9911055 ] , s. b. giddings , e. katz and l. randall , jhep * 0003 * , 023 ( 2000 ) [ arxiv : hep - th/0002091 ]
. t. shiromizu , k. i. maeda and m. sasaki , phys .
d * 62 * , 024012 ( 2000 ) [ arxiv : gr - qc/9910076 ] .
p. binetruy , c. deffayet and d. langlois , nucl .
b * 565 * , 269 ( 2000 ) [ arxiv : hep - th/9905012 ] , p. binetruy , c. deffayet , u. ellwanger and d. langlois , phys .
b * 477 * , 285 ( 2000 ) [ arxiv : hep - th/9910219 ] , s. mukohyama , phys .
b * 473 * , 241 ( 2000 ) [ arxiv : hep - th/9911165 ] , d. ida , jhep * 0009 * , 014 ( 2000 ) [ arxiv : gr - qc/9912002 ] , p. kraus , jhep * 9912 * , 011 ( 1999 ) [ arxiv : hep - th/9910149 ] .
a. hebecker and j. march - russell , nucl .
b * 608 * , 375 ( 2001 ) [ arxiv : hep - ph/0103214 ] , d. langlois , l. sorbo and m. rodriguez - martinez , phys .
lett . * 89 * , 171301 ( 2002 ) [ arxiv : hep - th/0206146 ] , d. langlois and l. sorbo , phys .
d * 68 * , 084006 ( 2003 ) [ arxiv : hep - th/0306281 ] , e. leeper , r. maartens and c. f. sopuerta , class .
* 21 * , 1125 ( 2004 ) [ arxiv : gr - qc/0309080 ] , k. ichiki , m. yahiro , t. kajino , m. orito and g. j. mathews , phys .
d * 66 * , 043521 ( 2002 ) [ arxiv : astro - ph/0203272 ] .
j. garriga and m. sasaki , phys .
043523 ( 2000 ) [ arxiv : hep - th/9912118 ] . s. w. hawking , t. hertog and h. s. reall , phys .
043501 ( 2000 ) [ arxiv : hep - th/0003052 ] . s. kobayashi , k. koyama and j. soda , phys .
b * 501 * , 157 ( 2001 ) [ arxiv : hep - th/0009160 ] , y. himemoto and m. sasaki , phys .
d * 63 * , 044015 ( 2001 ) [ arxiv : gr - qc/0010035 ] , y. himemoto , t. tanaka and m. sasaki , phys .
d * 65 * , 104020 ( 2002 ) [ arxiv : gr - qc/0112027 ] , t. tanaka and y. himemoto , phys .
d * 67 * , 104007 ( 2003 ) [ arxiv : gr - qc/0301010 ] .
k. koyama and k. takahashi , phys .
d * 67 * , 103503 ( 2003 ) [ arxiv : hep - th/0301165 ] , k. koyama and k. takahashi , phys .
d * 68 * , 103512 ( 2003 ) [ arxiv : hep - th/0307073 ] , t. kobayashi and t. tanaka , phys .
d * 69 * , 064037 ( 2004 ) [ arxiv : hep - th/0311197 ] .
r. maartens , d. wands , b. a. bassett and i. heard , phys .
d * 62 * , 041301 ( 2000 ) [ arxiv : hep - ph/9912464 ] , s. mukohyama , phys .
d * 62 * , 084015 ( 2000 ) [ arxiv : hep - th/0004067 ] , h. kodama , a. ishibashi and o. seto , phys .
d * 62 * , 064022 ( 2000 ) [ arxiv : hep - th/0004160 ] , d. langlois , phys .
d * 62 * , 126012 ( 2000 ) [ arxiv : hep - th/0005025 ] , d. langlois , phys .
lett . * 86 * , 2212 ( 2001 ) [ arxiv : hep - th/0010063 ] , h. a. bridgman , k. a. malik and d. wands , phys . rev .
d * 63 * , 084012 ( 2001 ) [ arxiv : hep - th/0010133 ] , d. langlois , r. maartens , m. sasaki and d. wands , phys .
d * 63 * , 084009 ( 2001 ) [ arxiv : hep - th/0012044 ] , h. a. bridgman , k. a. malik and d. wands , phys .
d * 65 * , 043502 ( 2002 ) [ arxiv : astro - ph/0107245 ] , k. koyama and j. soda , phys .
d * 62 * , 123502 ( 2000 ) [ arxiv : hep - th/0005239 ] , k. koyama and j. soda , phys .
d * 65 * , 023514 ( 2002 ) [ arxiv : hep - th/0108003 ] , b. gumjudpai , r. maartens and c. gordon , class .
quant . grav .
* 20 * , 3295 ( 2003 ) [ arxiv : gr - qc/0304067 ] .
h. kodama and m. sasaki , prog .
* 78 * , 1 ( 1984 ) .
v. f. mukhanov , h. a. feldman and r. h. brandenberger , phys .
* 215 * , 203 ( 1992 ) .
k. koyama , phys .
d * 66 * , 084003 ( 2002 ) [ arxiv : gr - qc/0204047 ] .
k. koyama , phys .
lett . * 91 * , 221301 ( 2003 ) [ arxiv : astro - ph/0303108 ] . c. s. rhodes , c. van de bruck , p. brax and a. c. davis , phys .
d * 68 * , 083511 ( 2003 ) [ arxiv : astro - ph/0306343 ] .
k. koyama , jcap * 0409 * , 010 ( 2004 ) [ arxiv : astro - ph/0407263 ] .
k. koyama , d. langlois , r. maartens and d. wands , jcap * 0411 * , 002 ( 2004 ) [ arxiv : hep - th/0408222 ] .
h. yoshiguchi and k. koyama , phys .
d * 71 * , 043519 ( 2005 ) [ arxiv : hep - th/0411056 ] .
d. langlois , r. maartens and d. wands , phys .
b * 489 * , 259 ( 2000 ) [ arxiv : hep - th/0006007 ] .
d. s. gorbunov , v. a. rubakov and s. m. sibiryakov , jhep * 0110 * , 015 ( 2001 ) [ arxiv : hep - th/0108017 ] .
t. kobayashi , h. kudoh and t. tanaka , phys .
d * 68 * , 044025 ( 2003 ) [ arxiv : gr - qc/0305006 ] . t. hiramatsu , k. koyama and a. taruya , phys .
b * 578 * , 269 ( 2004 ) [ arxiv : hep - th/0308072 ] .
t. hiramatsu , k. koyama and a. taruya , phys .
b * 609 * , 133 ( 2005 ) [ arxiv : hep - th/0410247 ] .
k. ichiki and k. nakamura , phys .
d * 70 * , 064017 ( 2004 ) [ arxiv : hep - th/0310282 ] .
k. ichiki and k. nakamura , arxiv : astro - ph/0406606 .
r. easther , d. langlois , r. maartens and d. wands , jcap * 0310 * , 014 ( 2003 ) [ arxiv : hep - th/0308078 ] .
r. a. battye , c. van de bruck and a. mennim , phys .
d * 69 * , 064040 ( 2004 ) [ arxiv : hep - th/0308134 ] .
r. a. battye and a. mennim , phys .
d * 70 * , 124008 ( 2004 ) [ arxiv : hep - th/0408101 ]
. t. tanaka , arxiv : gr - qc/0402068 .
t. kobayashi and t. tanaka , jcap * 0410 * , 015 ( 2004 ) [ arxiv : gr - qc/0408021 ] .
j. c. long , h. w. chan , a. b. churnside , e. a. gulbis , m. c. m. varney and j. c. price , arxiv : hep - ph/0210004 .
a. r. liddle and d. h. lyth , `` cosmological inflation and large - scale structure , '' cambridge university press , 2000 . | we study the quantum - mechanical generation of gravitational waves during inflation on a brane embedded in a five - dimensional anti - de sitter bulk . to make the problem well - posed , we consider the setup in which both initial and final phases are given by a de sitter brane with different values of the hubble expansion rate . assuming that the quantum state is in a de sitter invariant vacuum in the initial de sitter phase , we numerically evaluate the amplitude of quantum fluctuations of the growing solution of the zero mode in the final de sitter phase .
we find that the vacuum fluctuations of the initial kaluza - klein gravitons as well as of the zero mode gravitons contribute to the final amplitude of the zero mode on small scales , and the power spectrum is quite well approximated by what we call the rescaled spectrum , which is obtained by rescaling the standard four - dimensional calculation following a simple mapping rule .
our results confirm the speculation raised in ref .
@xcite before . |
photometric redshifts are of key importance to current and future galaxy surveys . in a typical application
, galaxies might be binned according to photometric redshift and then some analysis conducted on these binned source sets . avoiding systematic error in this case requires knowledge of the true redshift distribution of each photometric redshift bin . ideally , photometric redshift methods themselves produce reliable confidence intervals for this purpose , but external verification is important for controlling and quantifying systematic errors . for deep surveys , direct verification with a spectroscopic sample representative of the photometric sample is very difficult , and this has led to the development of other methods such as cross - correlating each photometric redshift bin with other photometric redshift bins ( schneider _ et al . _
2006 , erben _ et al . _
2009 ) or with spectroscopic redshift bins ( newman 2008 , matthews & newman 2010 ) . here
we present a new and independent method for reconstructing source redshift distributions , using the shear from weak gravitational lensing .
the shear induced by a given lens grows with source redshift in a well - understoood way which depends on cosmographic parameters .
other authors ( jain & taylor 2003 , bernstein & jain 2004 ) have suggested using this dependence to constrain the cosmographic parameters from the observed shear - vs - redshift relation around identified lenses .
( note that this is distinct from cosmic shear , for which the redshift dependence also involves the growth of structure . ) in this paper we reverse the question and ask how well the redshift distribution of a set of galaxies can be constrained by shear measurements around lenses at a range of redshifts , if the background cosmology is already well determined ( see appendix [ sec - cosmo ] for a demonstration that cosmological uncertainties are negligible here ) .
this may prove useful for applications which take the background cosmology as given , and it also provides an internal consistency check for surveys .
although the shear around any given lens is rather weak and might be expected to provide very little constraint , a very large survey such as lsst ( ivezi _ et al .
_ 2008 ) will contain billions of sources lensed by tens of thousands of galaxy clusters . assuming the photometry is uniform over the sky , the entire dataset can be used to constrain the source redshift distribution . in this paper
we use fisher matrix formalism to estimate an upper bound on the effectiveness of this method , and we discuss challenges to implementing the method on real data .
because this method requires a very large survey , we refer to lsst throughout .
we refer the reader to the lsst science book ( lsst science collaborations _
2009 ) for more details on lsst survey parameters .
the tangential shear @xmath5 at a projected distance @xmath6 around an axisymmetric lens is : @xmath7\ ] ] where @xmath8 , @xmath9 , and @xmath10 are respectively the angular diameter distances from lens to source , observer to lens , and observer to source , @xmath11 is the projected surface mass density at @xmath6 , and @xmath12 is the mean projected surface mass density within @xmath6 ( miralda - escud 1991 ) .
figure [ fig - dratio ] ( top panel ) shows the behavior of the distance factor @xmath13 as a function of source redshift , for various fixed lens redshifts .
this depends on the assumed cosmological model ; we assume a wmap7 @xmath14cdm cosmology of @xmath15 km s@xmath2 mpc @xmath2 , @xmath16 , and @xmath17 throughout .
it is clear that very low - redshift ( @xmath18 ) source galaxies will have no measurable shear around any of the lenses in the figure , moderate - redshift ( @xmath19 ) sources will have measurable shear only around the lower - redshift lenses , and high - redshift ( @xmath20 ) sources will have measurable shear around all lenses .
this is the conceptual basis for constraining the redshift distribution of a source set : one should be able to solve for the source redshift distribution which , when passed through the appropriate response curves , best matches the observed shears .
the method requires lenses at a range of redshifts and , to obtain good constraints , some prior knowledge of the lens masses . to simplify the exposition we treat the lens propertise as known in this section , and explore requirements on prior knowledge of lens properties in the next section . as a function of source redshift , for various fixed lens redshifts .
the lens redshift for each curve can be read off at the point where the curve departs from zero . _
bottom : _ as for the left panel , but plotting @xmath21 .
[ fig - dratio],title="fig : " ] + as a function of source redshift , for various fixed lens redshifts .
the lens redshift for each curve can be read off at the point where the curve departs from zero . _
bottom : _ as for the left panel , but plotting @xmath21 .
[ fig - dratio],title="fig : " ] + individual galaxies as well as clusters of galaxies can serve as lenses .
typically , the shear one arcminute off axis might be of order 0.5 for clusters and 0.01 for galaxies .
galaxies have near - isothermal mass profiles ( gavazzi _ et al .
_ 2007 ) , in which @xmath11 and @xmath12 scale as @xmath22 ; specifically , @xmath23 where @xmath24 is the velocity dispersion .
because this scales as @xmath22 , it scales as @xmath25 in terms of projected _ angular _ distance @xmath26 , and @xmath27 can be written @xmath28 isothermal lenses thus have the convenient property that @xmath27 can be separated into a part which depends _ only _ on the cosmology and a part which depends _ only _ on the mass or velocity dispersion of the lens .
mass profiles which are not scale - free can not be written this way because the conversion from physical to angular scales invokes the cosmological model through @xmath9 .
this convenience is not required for our argument , but will substantially simplify the exposition here .
although cluster mass profiles depart from isothermality more than galaxy profiles do , for this section we will assume isothermality to simplify our argument .
we can now write a very simple model for the shear of the @xmath29th source galaxy due to the @xmath30th lens .
we henceforth refer to @xmath24 as @xmath31 to eliminate confusion with the symbol @xmath32 which we will use later to indicate uncertainties on measurements .
we also define @xmath33 ( taking @xmath34 to be zero where @xmath35 ) ; the @xmath34 curves for various lenses are shown in the bottom panel of figure [ fig - dratio ] .
we can then write @xmath36 where @xmath37 is the angular separation between lens @xmath30 and source @xmath29 , @xmath38 is the distance ratio between the two , and @xmath39 is the velocity dispersion of lens @xmath30 .
the redshift distribution of the source set can be parameterized in many ways . here
we adopt the simplest approach , uniform bins in redshift . with @xmath40 bins
, we parametrize the distribution with @xmath41 , where @xmath42 is the fraction of the galaxies in the set which are actually in redshift bin @xmath43
. the highest - redshift bin should extend to arbitrarily high redshift so that none of the measured shear can come from an unmodeled part of the distribution .
becomes fairly flat at high redshift , a very wide high - redshift bin is no more problematic than a modest - width low - redshift bin in terms of variation of @xmath34 within each bin . ] consequently , we do not require that @xmath44 ; any difference between unity and @xmath45 can be interpreted as due to sources at such low redshift that they are never lensed .
( a derived parameter @xmath46 can be defined which encodes the fraction of sources at ultralow redshifts , but is not a parameter we solve for directly . ) for sources from an unknown redshift distribution , the expected value of @xmath34 is simply @xmath47 , where @xmath48 is now the mean distance ratio between lens @xmath30 and source redshift bin @xmath43 . thus the expected shear for a lens - source combination is @xmath49 in this section , we will assume that @xmath39 is perfectly known ; we will consider uncertainty in lens masses in [ sec - mass ] .
the observable is the total shear where @xmath50 is the convergence , but the argument still holds .
the effect of @xmath50 is to change the profile somewhat , which is not important for the toy model here . ] on each source galaxy due to all lenses which , in the weak lensing limit , is @xmath51 .
however , we must add shears componentwise because in the multiple - deflector context there is no longer a single `` tangential '' component .
thus shear is usually written as a complex quantity : @xmath52 where @xmath53 is the position angle of source @xmath29 with respect to lens @xmath30 , and the @xmath29 in the exponent represents @xmath54 .
the total expected shear on the source is then @xmath55 we could form a fisher matrix with each two - component @xmath56 contributing two observables ; the derivatives of these observables with respect to the model parameters @xmath42 are very straightforward .
however , this is computationally inefficient because most sources will be far from any modeled lens ( assuming we use massive clusters rather than individual galaxies as the lenses ) and thus will contribute negligible information to the fisher matrix ; and of the remaining sources , most are near only one lens so that their contribution can be captured by a single ( tangential ) component .
thus , at the expense of very little information , we can efficiently group sources according to their relevant lenses . to do this , we group the observables in equation [ eqn - shearij2 ] on one side and the model parameters on the other : @xmath57 the @xmath58 factor no longer appears because with only one lens involved the tangential component is unambiguous .
we can collapse this down to a single observable for each lens by taking a mean of sources relevant to that lens : @xmath59 where @xmath60 denotes averaging over all sources projected close enough to lens @xmath30 to add non - negligible information to the fisher matrix .
accordingly , @xmath61 to construct the fisher matrix elements , we also need the variance of @xmath62 . because there is essentially no uncertainty on @xmath37 ,
the variance of @xmath63 is @xmath64 where @xmath65 is the uncertainty on the measured shear of galaxy @xmath29 .
furthermore , @xmath65 is uncorrelated with position or lens , so we can approximate it as a constant @xmath66 and pull it out of any sums . weighting sources by inverse variance when computing @xmath62
, we then find that @xmath67 this sum primarily involves properties of the survey rather than the lens .
we can rewrite it as the integral of the areal density of the source set , @xmath68 , over the `` footprint '' of the lens , that is , the area over which the lens contributes non - negligible fisher information : @xmath69 .
( note that @xmath68 here is in units of galaxies per steradian , because @xmath26 is in units of radians .
we will convert to the more typically quoted units of arcmin@xmath70 at the end of the calculation . )
we shall see that the value of this integral is only weakly related to the properties of the lens .
integrating from some @xmath71 to some @xmath72 yields @xmath73 because the dependence is only logarithmic , we can make a rough estimate of @xmath74 for a typical lens and apply it to all lenses without much loss of precision .
sources less than @xmath75@xmath76 from the lens center are typically not used in weak lensing analyses because of the risk of contamination by cluster members and/or errors on their shear measurements due to background gradients from cluster members , so we adopt this value of @xmath71 . at large enough @xmath26 , shear from unrelated structures will dominate the shear from the cluster .
although information about the cluster is still there , as a practical matter the effort to recover it may not be justified given the logarithmic dependence on @xmath72 .
the shear around a massive cluster can clearly be observed out to at least @xmath77@xmath76 from the center ( e.g , kling _
et al . _
2005 ) , so @xmath78@xmath76 should be a conservative value .
thus equation [ eqn - ln ] is approximately equal to @xmath79 , and only weakly dependent on the adopted values of @xmath71 and @xmath72 .
we expect the dependence on assumed shear profile to be weak as well , given that many ground - based weak - lensing cluster studies have difficulty discriminating between different forms of shear profiles ( wittman 2002 ) . inserting this result into equation [ eqn - sigmagamma ] , we find that @xmath80 assuming a gaussian likelihood model , we now have an @xmath40 by @xmath40 fisher matrix : @xmath81 where we have adopted @xmath82 , which represents the irreducible `` shape noise '' due to the random intrinsic orientations of sources .
smaller , fainter , sources will have larger shear uncertainties due to measurement uncertainties : @xmath83 .
we account for this by defining @xmath68 as the _ effective _ source density , that is , the equivalent number of perfectly measured sources provided by the source set : @xmath84 where the sum is taken over a unit area of sky .
this is the basis on which surveys or source sets should be quoted ; see the explanation of equation 14.7 in the lsst science book ( lsst science collaborations _
_ 2009 ) for a more detailed justification . as a rough estimate of the potential of this method
, we compute a fisher matrix forecast for an lsst - depth ( @xmath85 arcmin@xmath70 ) survey covering nineteen lenses with @xmath1 km s@xmath2 at redshifts @xmath86 , which roughly corresponds to using the most massive lens in the sky at each redshift .
we model six redshift bins : five covering @xmath87 with equal spacing , plus a sixth bin containing all galaxies at @xmath20 .
the forecast uncertainties on @xmath42 are then @xmath88 .
the same fisher information can be obtained using a larger number of less - massive lenses , but going beyond a small number of well - studied lenses introduces a new problem : lens parameters such as @xmath24 will not be known precisely for all these lenses , and we must examine the effect of marginalizing over uncertainties in the lens properties .
the two lens parameters in this model are redshift and mass ( or equivalently , velocity dispersion ) .
spectroscopic redshifts are available in the literature for thousands of clusters , particularly for the most massive ones which contribute the most fisher information . but with large imaging surveys now finding more clusters than can be followed up with spectroscopic redshift confirmation , photometric redshifts for the lenses must be considered .
photometric redshifts are much more precise for clusters than for individual galaxies , even beyond the statistical effect of averaging many member galaxies .
because cluster members are preferentially early - type galaxies with well - established spectral energy distributions , their redshifts can be established with very high confidence .
furthermore , the accuracy of cluster photometric redshifts can be easily verified by spectroscopy of a representative subsample of clusters , whereas representativeness is very difficult to achieve in spectroscopic subsamples of individual galaxies . the lsst science book ( lsst science collaborations _
et al _ 2009 ) forecasts photometric redshift uncertainties for modest ( 10@xmath89 ) clusters as a function of redshift ; their figure 13.12 shows that the uncertainties are 0.01 or less for all @xmath90 after a single visit ( and for all @xmath91 after ten years of the survey ) .
this fixes the @xmath34 factors to within 0.01 , which is far more precise than the masses will be known _ a priori_. we can therefore safely ignore uncertainties in lens redshifts . in the remainder of this section , we focus on marginalizing over lens mass uncertainties .
we now regard the @xmath92 as nuisance parameters .
we choose @xmath92 rather than @xmath39 simply for convenience : @xmath92 is conceptually convenient because it is linearly proportional to the shear , and this proportionality makes it fall out of the relevant fisher matrix elements . from equation [ eqn - gammadef ]
we have @xmath93 we now have a fisher matrix which is @xmath94 elements square .
the lower - right @xmath95 block is diagonal , because @xmath96 where @xmath97 .
these diagonal elements look like the square of the first case in equation [ eqn - betasum ] .
the upper - left @xmath98 block contains only elements of the form in equation [ eqn - betas ] , and the other blocks ( the @xmath99 block in the upper right and its transpose in the lower left ) contain mixed terms .
note , however , that many of the elements in these latter blocks will be zero , corresponding to lens / bin combinations for which the bin redshift is lower than the lens redshift , and thus for which @xmath100 . in the upper left block
, no elements will vanish unless _ all _ lenses are at a higher redshift than the lowest - redshift bin .
note that the @xmath42 now appear in the fisher matrix , so we must adopt fiducial values .
fisher matrix estimates should not be trusted far from the fiducial values ; see albrecht _
( 2006 ) for a clear exposition of fisher matrix limitations .
we therefore adopt a uniform fiducial distribution for now , and examine the dependence on the fiducial distribution at the end of this section .
next , we need a prior on the lens masses . with no prior , the lowest - redshift lens would have no constraining power at all , as any shear ( or lack thereof ) in the lowest - redshift bin could be explained as a result of arbitrarily large ( or small ) lens mass . successively higher - redshift lenses would have some constraining power , but not much .
we therefore explore the importance of this prior by running fisher matrix forecasts with the lens set of the previous section , but with priors of 1% , 2% , 5% , 10% , and 20% on the lens masses assuming the fiducial @xmath101 km s@xmath2 .
we implement this by adding to the fisher matrix the inverse variance representing a gaussian prior before inverting the fisher matrix to obtain the covariance matrix . to implement the 1% prior , for example , we add @xmath102 to @xmath103 , which are the nineteen elements representing the lens @xmath92 .
figure [ fig - masspriors ] shows the results .
for the low - redshift bins , obtaining constraints close to those of the previous section requires a 1 - 2% prior , but for higher - redshift bins the required priors are more modest , 10 - 20% .
priors tighter than 1% ( not shown ) do not yield any visible improvement in this figure . with just one massive lens at each redshift .
the redshift - varying prior allows for weaker lens - mass priors at higher lens redshifts where priors are both less necessary and more difficult to establish observationally ; this particular form is @xmath104 which brings the prior to 20% at @xmath105 .
more lenses of the same mass will tighten the constraints as @xmath106 .
[ fig - masspriors ] ] the most efficient observational approach may thus be to obtain tight priors on low - redshift lenses using velocity dispersions , x - ray analyses , external lensing analyses , etc , while obtaining only very rough priors on the higher - redshift lenses .
we model this scenario by testing a mass prior which is 5% at @xmath107 but loosens as @xmath108 ( i.e. , a 20% prior at @xmath109 ) ; this is is shown as the dashed curve in figure [ fig - masspriors ] . this yields source redshift constraints as good as those obtained with a 5 - 10% prior , but without much difficult high - redshift work .
indeed , optical richness alone can provide a 20 - 30% prior on the mass ( rykoff _ et al . _
2011 ) , thus obviating the need for followup at the high - redshift , assuming there is no redshift above which the optical richness estimator breaks down . to a reasonable approximation
, each additional lens tightens the constraints mostly on the 1 - 2 source redshift bins nearest the lens .
therefore , we can tweak the lens sample to obtain roughly equal constraints on all source redshift bins .
for example , we see from figure [ fig - masspriors ] that the maximum uncertainty for the efficient redshift - varying mass prior is for source redshifts 0.7 - 0.9 , at a level of roughly twice the minimum uncertainty . tripling the number of lenses at redshifts 0.5 - 0.95 , for a total of 39 massive lenses , yields roughly equal constraints on most source redshift bins , averaging 0.11 ( middle curve in figure [ fig - sourcedep ] ) . reaching yet lower levels of uncertainty
requires more lenses overall .
for example , reaching 0.01 uncertainty across the board will require @xmath110 times more lenses than the previous scenario , or about 5000 total .
however , when so many lenses are considered , we can not use only very massive lenses .
therefore , the required number of lenses will scale much more steeply than the inverse square of the desired constraint . the fisher information corresponding to 5000 lenses with @xmath111 km s@xmath2 could be accumulated with tens of thousands of less - massive lenses , which is well within expectations for current and future large imaging surveys .
computationally , using so many lenses should be feasible because most of the tens of thousands of nuisance parameters can be marginalized over separately , reflecting the nonoverlapping nature of most lenses
. however , the 1% goal will be difficult to achieve at low redshift due to the difficulty of obtaining the 5 - 8% lens - mass priors necessary to do so well there .
these factors provide insight into why we do not propose using individual galaxies as lenses . for every @xmath112 @xmath113 cluster ,
a survey would need one million @xmath114 @xmath113 galaxies to provide the same fisher information , all other factors being equal .
although surveys such as lsst will have billions of galaxies , all other factors are not equal .
the photometric redshift uncertainty on each individual galaxy is much larger than for clusters , and individual - galaxy lens photometric redshifts bring in exactly the problems this method is designed to avoid .
these problems could be avoided by using only early - type galaxies with well - behaved photometric redshifts and mass priors from the fundamental plane , but then far too few lenses would be available .
furthermore , galaxy - scale lenses do not dominate their local shear in a way which allows parallelized modeling of designated patches of sky , and thus would introduce additional computational challenges .
a foreseeable application for this method is to input source sets corresponding to each of several photometric redshift bins , to determine the outlier fraction and distribution in each .
therefore , we examine the dependence on the redshift distribution of the source set , which we have assumed to be flat until now . figure [ fig - sourcedep ] shows the results using the redshift - varying mass prior and the 39 massive lenses with three different source sets : flat ( as described above , middle curve ) , all in the @xmath18 bin ( lower curve ) , and all in the @xmath115 bin ( upper curve ) .
characterization of outliers will be more precise for the low-@xmath116 source set than for the high-@xmath116 source set , although this is partly due to the choice of lens - mass prior which is tighter at lower redshift . ]
to this point we have assumed that prior knowledge of a cluster s velocity dispersion translates perfectly into prior knowledge of its integrated shear statistic , given a source redshift distribution .
however , the relationship between these two observables is sure to be more complicated .
clusters with the same velocity dispersion may have different profiles , resulting in different integrated shear estimates . if this is a purely stochastic effect , then the scatter in the observable - observable relation weakens the prior .
but in principle , the loss of information due to this scatter can be compensated by analyzing more clusters .
conceptually more worrisome would be a redshift - dependent trend or an overall calibration error in this relationship , which would cause systematic errors .
we therefore insert nuisance parameters describing the uncertainties in calibration and redshift evolution , and explore how the results degrade as priors on these parameters are relaxed .
we allow equation [ eqn - gammadef ] to be multiplied by a factor @xmath117 , where @xmath118 is a nuisance parameter for the calibration , @xmath119 is a nuisance parameter for the redshift evolution , and @xmath120 is the redshift of the @xmath30th lens , and repeat the analysis of the previous section with the flat source distribution and the redshift - dependent lens mass prior .
figure [ fig - evol ] shows the reconstruction precision for a range of calibration and evolution priors , with a fiducial model of @xmath121 and @xmath122 .
priors of 20% on each are sufficient to preserve performance nearly identical to that of the previous section in which calibration and evolution were not factors ; the bottom curve in each panel is very similar to the middle curve in figure [ fig - sourcedep ] . , where the priors were delta functions .
[ fig - evol],title="fig : " ] + , where the priors were delta functions .
[ fig - evol],title="fig : " ] we stress that the choice of an isothermal profile as a calculation aid here is not fundamental to the method . in practice , surveys would measure the average cluster profile and construct an optimal integrated shear estimator before commencing the redshift analysis .
the results of this section indicate that profile calibration and evolution uncertainties at the 20 - 50% level do not substantially degrade the performance of this method .
we have outlined a new method which uses gravitational lensing by clusters and knowledge of the cosmology to constrain source redshift distributions .
although this is an unorthodox use of the lensing - redshift information , it has some strengths and some relevant applications . the method s strengths are : * it is independent of other methods .
it relies on no assumptions about galaxy spectral properties or biasing .
* it is applicable to any set of galaxies .
one can determine the redshift distribution of a set of galaxies chosen in any number of ways , such as photometric redshift cut , pure color selection , or even x - ray or radio selection .
multiwavelength observations are not necessary as they are for photometric redshifts . * in the context of a large imaging survey , it requires little additional data . where lens spectroscopic redshifts are not available , photometric lens redshifts will do , and optical richness provides a sufficient lens - mass prior for the high - redshift lenses
however , pinning down the low - redshift end of the source distribution will require strong mass priors on low - redshift lenses .
our fisher matrix forecast involves some approximations and caveats : * lens morphology : we have assumed axisymmetric isothermal lenses .
although we have argued that other profiles would not yield substantially different forecasts , a precise experiment would include some sort of marginalization over profiles .
this marginalization might include allowances for non - axisymmetry .
* foreground / background structure : we have assumed that shear from foreground / background structure is negligible in the `` footprint '' of each lens , and totally dominant outside . in reality , unmodeled shear will be present in the footprint as well , and contribute to slightly looser constraints .
however , unmodeled shear is not an irreducible source of noise .
one can choose to model previously unmodeled structure and thus add information to the fisher matrix .
this entails a cost - benefit analysis rather than a hard limit . *
magnification : the observed source population behind lenses will be slightly different than in other locations , due to lensing magnification .
if unmodeled , this may result in a nontrivial systematic error .
however , because the lenses are being modeled , the model magnification at any given point provides an easily available basis for correction .
deep fields which establish the source counts fainter than the main survey flux limit will be very useful for this purpose . as with the foreground / background structure
, it may be possible to add information to the fisher matrix with improved modeling of the magnification .
* cluster contamination : the areas around clusters differ from the general area of a survey in another respect , the presence of cluster members .
care must be taken so that the source set whose shear is being measured is not contaminated by these members .
* cosmology dependence of @xmath34 : appendix [ sec - cosmo ] shows that marginalizing over uncertainty in cosmographic parameters will not degrade the results presented here . but using the cosmology as input here implies that the redshift estimates from this method should not feed in to some other method which estimates the cosmology unless the covariances are properly treated . *
wide source redshift bins : given a lens redshift , @xmath34 varies over the width of a source redshift bin . accurately modeling this effect
will require some assumptions about the source redshift distribution inside the bins .
it may be possible to find a parameterization which reduces this difficulty
. the source redshift distributions obtained with this method will be most useful in applications where the background cosmology would have been assumed anyway , such as galaxy evolution , and especially in applications which are very sensitive to photometric redshift outliers .
for example , measurements of the bright end of the galaxy luminosity function at high photometric redshift are extremely sensitive to catastrophic outliers , as a small fraction of low - redshift galaxies mistakenly put at high redshift results in many more `` high - luminosity '' galaxies at that redshift .
surveys can use this method to independently derive the full redshift distribution of the galaxies in each photometric redshift bin , thus characterizing the number and distribution of outliers .
indeed , the use of multiwavelength photometry to infer both source redshift _ and _ source spectral / photometric properties can be problematic , and this method offers a possible workaround .
other methods such as cross - correlation with a spectroscopic sample ( newman 2008 , matthews & newman 2010 ) should be able to achieve 1% accuracy without assuming a cosmology or requiring expensive lens mass priors .
this method may therefore not be competitive statistically , but as it is based on different physics it may offer value in having different systematic errors .
redshift distributions derived using this method can and should be used in consistency checks . given the best - fit cosmology from weak lensing tomography ,
this method should produce source redshift distributions which are consistent with those used in the weak lensing tomography .
a prime application for this method may therefore be as an internal consistency check for surveys .
we thank j. a. tyson for suggesting that we pursue the idea of using lensing to constrain source redshift distributions , and the anonymous referee for suggestions which led to improvements in the paper .
this appendix justifies the assumption in the main body of the paper that uncertainties in the distance ratios contribute negligible uncertainty to the parameters of interest , the fraction of sources in each redshift bin .
uncertainty in the distance ratios comes from uncertainty in cosmographic parameters , including hubble s constant @xmath123 and the energy densities in matter ( @xmath124 ) , curvature ( @xmath125 ) , and in the cosmological constant ( @xmath126 ) .
of course , more general dark energy models can introduce further parameters , but the cosmological constant has a greater effect on distances than a dark energy model with equation of state parameter @xmath127 .
therefore , if the assumption is justified for a cosmological constant model , it is justified for models with @xmath127 .
the remaining parameters are more difficult to treat analytically , so we make a numerical estimate . the derivatives of @xmath34 with respect to @xmath124 and @xmath126 ( with @xmath129 )
depend on the lens and source redshifts , but are typically of order 0.02 .
the worst case is for uncertainty in @xmath124 when considering high - redshift sources and lenses , where @xmath130 .
we take a conservative approach by setting @xmath131 for all sources and lenses , so that the partial derivative of equation [ eqn - gammadef ] with respect to @xmath124 is 0.15 times @xmath132 .
we include @xmath124 as a parameter in the fisher matrix using these partial derivatives .
we rerun the forecast with a gaussian prior of @xmath133 on @xmath124 , which corresponds to the current wmap7+bao+h0 uncertainty for the `` olcdm '' model in which curvature is allowed to vary .
if the results were plotted on figure [ fig - masspriors ] or figure [ fig - sourcedep ] , they would be nearly indistinguishable from the existing plots .
the biggest difference would be the point in the lower right corner of figure [ fig - sourcedep ] increasing from 0.052 to 0.055 .
the effect of varying @xmath126 is even smaller because @xmath34 is less sensitive to it . | we introduce a new method for constraining the redshift distribution of a set of galaxies , using weak gravitational lensing shear . instead of using observed shears and redshifts to constrain cosmological parameters ,
we ask how well the shears around clusters can constrain the redshifts , assuming fixed cosmological parameters .
this provides a check on photometric redshifts , independent of source spectral energy distribution properties and therefore free of confounding factors such as misidentification of spectral breaks .
we find that @xmath0 massive ( @xmath1 km s@xmath2 ) cluster lenses are sufficient to determine the fraction of sources in each of six coarse redshift bins to @xmath311% , given weak ( 20% ) priors on the masses of the highest - redshift lenses , tight ( 5% ) priors on the masses of the lowest - redshift lenses , and only modest ( 20 - 50% ) priors on calibration and evolution effects .
additional massive lenses drive down uncertainties as @xmath4 , but the improvement slows as one is forced to use lenses further down the mass function .
future large surveys contain enough clusters to reach 1% precision in the bin fractions if the tight lens mass priors can be maintained for large samples of lenses . in practice
this will be difficult to achieve , but the method may be valuable as a complement to other more precise methods because it is based on different physics and therefore has different systematic errors . |
_ extreme events _ ( also called critical transitions , disasters , catastrophes and crises ) are a most important yet least understood feature of many natural and human - made processes . among examples are destructive earthquakes , el - nios , economic depressions , stock - market crashes , and major terrorist acts .
extreme events are relatively rare , and at the same time they inflict a lion s share of the damage to population , economy , and environment . accordingly , studying the extreme events is pivotal both for fundamental predictive understanding of complex systems and for disaster preparedness ( see @xcite and references therein ) . in this paper
we work within a framework that emphasizes mechanisms underlying formation of extreme events .
prominent among such mechanisms is _ direct cascading _ or _
fragmentation_. among other applications , this mechanism is at the heart of the study of 3d turbulence @xcite . a statistical model of direct cascade is conveniently given by the branching processes ; they describe populations in which each individual can produce descendants ( offsprings ) according to some probability distribution .
a branching process may incorporate _ spatial _ dynamics , several types of particles ( multi - type processes ) , age - dependence ( random lifetimes of particles ) , and immigration due to external driving forces @xcite .
in many real - world systems , observations are only possible within a specific domain of the phase space of a system .
accordingly , we consider here a system with an _ unobservable _ source of external driving ultimately responsible for extreme events .
we assume that observations can only be made on a _
subspace _ of the phase space .
the direct cascade ( branching ) within a system starts with injection of the largest particles into the source .
these particles are divided into smaller and smaller ones , while spreading away from the source and eventually reaching the subspace of observations .
an important observer s goal is to locate the external driving source .
the distance between the observation subspace and the source thus becomes a natural control parameter .
an extreme event in this system can be defined as emergence of a large particle in the observation subspace .
clearly , as the source approaches the subspace of observation , the total number of observed particles increases , the bigger particles become relatively more frequent , and the probability of an extreme event increases . in this paper
, we give a complete quantitative description of this phenomenon for an age - dependent multi - type branching diffusion process with immigration in @xmath0 .
it turns out that our model closely reproduces the major premonitory patterns of extreme events observed in hierarchical complex systems .
extreme events in such systems are preceded by transformation of size distribution in the permanent background activity ( see _ e.g. , _
@xcite ) . in particular , general activity increases , in favor of relatively strong although sub - extreme events . that was established first by analysis of multiple
fracturing and seismicity @xcite , and later generalized to socio - economic processes @xcite .
our results suggest a simple universal mechanism of such premonitory patterns .
the system consists of particles indexed by their _ generation _ @xmath1 .
particles of zero generation ( _ immigrants _ ) are injected into the system by an external forcing .
particles of any generation @xmath2 are produced as a result of splitting of particles of generation @xmath3 .
immigrants ( @xmath4 ) are born at the origin @xmath5 according to a homogeneous poisson process with intensity @xmath6 .
each particle lives for some random time @xmath7 and then transforms ( splits ) into a random number @xmath8 of particles of the next generation . the probability laws of the lifetime @xmath7 and branching @xmath8 are rank- , time- , and space - independent .
new particles are born at the location of their parent at the moment of splitting .
the lifetime distribution is exponential : @xmath9 @xmath10 .
the conditional probability that a particle transforms into @xmath11 new particles ( 0 means that it disappears ) given that the transformation took place is denoted by @xmath12 .
the probability generating function for the number @xmath8 of new particles is thus [ branching_pgf ] h(s ) = _
the expected number of offsprings ( also called the _ branching number _ ) is @xmath13 ( see _ e.g. _ , @xcite ) .
each particle diffuses in @xmath0 independently of other particles .
this means that the density @xmath14 of a particle that was born at instant @xmath15 at point @xmath16 solves the equation = d(_i ) p d _ p [ diffusion ] with the initial condition @xmath17 .
the solution of ( [ diffusion ] ) is given by p(*x , y*,t)= ( 4dt)^-n/2 \{- } , [ sol ] where @xmath18 .
it is convenient to introduce particle _ rank _
@xmath19 for an arbitrary integer @xmath20 and thus consider particles of ranks @xmath21 .
this reflects our focus on direct cascading , which often assumes that particles with larger size ( _ e.g. _ , length , volume , mass , energy , momentum , _ etc .
_ ) split into smaller ones according to an appropriate conservation law .
figure [ fig_example ] illustrates the model dynamics . to the origin .
model parameters are @xmath22 , @xmath23 , @xmath24 .
circle size is proportional to the particle rank .
different shades correspond to populations from different immigrants , the descendants of earlier immigrants have lighter shade .
the clustering of particles is explained by the splitting histories .
note that , as the origin approaches , the particle activity significantly changes , indicating the increased probability of an extreme event .
, title="fig : " ] to the origin .
model parameters are @xmath22 , @xmath23 , @xmath24 .
circle size is proportional to the particle rank .
different shades correspond to populations from different immigrants , the descendants of earlier immigrants have lighter shade .
the clustering of particles is explained by the splitting histories .
note that , as the origin approaches , the particle activity significantly changes , indicating the increased probability of an extreme event .
, title="fig : " ] to the origin .
model parameters are @xmath22 , @xmath23 , @xmath24 .
circle size is proportional to the particle rank .
different shades correspond to populations from different immigrants , the descendants of earlier immigrants have lighter shade .
the clustering of particles is explained by the splitting histories .
note that , as the origin approaches , the particle activity significantly changes , indicating the increased probability of an extreme event .
, title="fig : " ] to the origin .
model parameters are @xmath22 , @xmath23 , @xmath24 .
circle size is proportional to the particle rank .
different shades correspond to populations from different immigrants , the descendants of earlier immigrants have lighter shade .
the clustering of particles is explained by the splitting histories .
note that , as the origin approaches , the particle activity significantly changes , indicating the increased probability of an extreme event .
, title="fig : " ]
the model decsribed above is a superposition of independent branching processes generated by individual immigrants .
we consider first the case of a single immigrant ; then we expand these results to the case of multiple immigrants .
finally , we analyze the rank distribution of particles .
proofs of all statements will be published in a forthcoming paper .
let @xmath25 be the conditional probability that at time @xmath26 there exist @xmath27 particles of generation @xmath28 within spatial region @xmath29 given that at time 0 a single immigrant was injected at point @xmath16 .
the corresponding generating function is f_k(g,*y*,t;s ) = _ ip_k , i(g,*y*,t)s^i .
the generating functions @xmath30 solve the following recursive system of non - linear partial differential equations : f_k = -d_*y * f_k - f_k+h(f_k-1),k1 , [ pgf ] with the initial conditions @xmath31 , @xmath32 , and f_0(g,*y*,t;s ) = ( 1-p ) + ps , [ pgf_ini ] where @xmath33 [ prop_f ] next , consider the expected number @xmath34 of generation @xmath35 particles at instant @xmath36 within the region @xmath37 produced by a single immigrant injected at point @xmath16 at time @xmath38 .
it is given by the following partial derivative ( see _ e.g. _ , @xcite ) @xmath39 that satisfies , for any @xmath29 , |a_k(g,*y*,t)=_g a_k(*x*,*y*,t)d*x*. [ gs ] the expectation densities @xmath40 solve the following recursive system of linear partial differential equations : = d_*x * a_k - a_k+ba_k-1,k1 , [ exp_ave ] with the initial conditions @xmath41 @xmath42 the solution to this system is given by @xmath43 [ col ] the system has a transparent intuitive meaning .
the rate of change of the expectation density @xmath39 is affected by the three processes : diffusion of the existing particles of generation @xmath35 in @xmath0 ( first term in the rhs of ) , splitting of the existing particles of generation @xmath35 at the rate @xmath44 ( second term ) , and splitting of generation @xmath3 particles that produce on average @xmath45 new particles of generation @xmath35 ( third term ) . here
we expand the results of the previous section to the case of multiple immigrants that appear at the origin according to a homogeneous poisson process with intensity @xmath6 .
the expectation @xmath46 of the number of particles of generation @xmath35 is given , according to the properties of expectations , by @xmath47 the steady - state spatial distribution @xmath48 corresponds to the limit @xmath49 and is given by _
k(z)=()^k ( ) ^-n/2 z^k_(z ) .
[ az ] here @xmath50 , @xmath51 and @xmath52 is the modified bessel function of the second kind .
recall that the particle rank is defined as @xmath53 .
the spatially averaged steady - state rank distribution is a pure exponential law with index @xmath45 : @xmath54 to analyze deviations from the pure exponent , we consider the ratio @xmath55 between the number of particles of two consecutive generations : _ k():=. [ gamma ] for the purely exponential rank distribution , @xmath56 , the value of @xmath57 is independent of @xmath35 and @xmath58 ; while deviations from the pure exponent will cause @xmath59 to vary as a function of @xmath35 and/or @xmath58 . combining and we find _
k()= , [ gamma1 ] where , as before , @xmath60 and @xmath51 .
the asymptotic behavior of the function @xmath61 is given by @xmath62 [ gammalim ] proposition [ gammalim ] allows one to describe all deviations of the particle rank distribution from the pure exponential law .
figure [ fig_ak ] illustrates our findings .
. implies that at any spatial point , the distribution asymptotically approaches the exponential form as rank @xmath63 decreases ( generation @xmath35 increases ) .
thus the deviations can only be observed at the largest ranks ( small generation numbers ) .
analysis of the large - rank distribution is done using eqs . , .
near the origin , where the immigrants enter the system , eq .
implies that @xmath64 for @xmath65 .
hence , one observes the _ upward deviations _ from the pure exponent : for the same number of rank @xmath63 particles , the number of rank @xmath66 particles is larger than predicted by .
the same behavior is in fact observed for @xmath67 ( the details will be published elsewhere ) .
in addition , for @xmath67 the ratios @xmath61 do not merely deviate from @xmath68 , but diverge to infinity at the origin .
away from the origin , according to eq . , we have @xmath69 , which implies _ downward deviations _ from the pure exponent : for the same number of rank @xmath63 particles , the number of rank @xmath66 particles is smaller than predicted by . of generation @xmath35 particles at distance @xmath70 from the origin ( cf .
proposition [ gammalim ] ) .
the distance @xmath70 is increasing ( from top to bottom line in each panel ) as @xmath71 .
model dimension is @xmath72 ( panel a ) , @xmath73 ( panel b ) , @xmath74 ( panel c ) , and @xmath75 ( panel d ) .
other model parameters : @xmath22 , @xmath23 , @xmath24 , @xmath76 .
one can clearly see the transition from downward to upward deviation of the rank distributions from the pure exponential form as we approach the origin . ]
motivation for this work is the problem of prediction of extreme events in complex systems .
our point of departure is a classical model of spatially distributed population of particles of different ranks governed by direct cascade of branching and external driving . in the probability theory this model
is known as the age - dependent multi - type branching diffusion process with immigration @xcite .
we introduce here a new approach to the study of this process .
we assume that observations are only possible on a subspace of the system phase space while the source of external driving remains unobservable .
the natural question under this approach is the dependence of size - distributions of particles on the distance to the source .
the complete analytical solution to this problem is given by the proposition [ prop_f ] .
it is natural to consider rank as a logarithmic measure of the particle size .
if we assume a size - conservation law in the model , the exponential rank distriburtion derived in corresponds to a self - similar , power - law distribution of particle sizes , characteristic for many complex systems .
thus , the proposition [ gammalim ] describes space - dependent deviations from the self - similarity ( see also fig .
[ fig_ak ] ) ; in particular , deviations premonitory to an extreme event .
the numerical experiments ( that will be published elsewhere ) confirm the validity of our analytical results and asymptotics in a finite model .
the model studied here exhibits very rich and intriguing premonitory behavior .
figure [ fig_example ] shows several 2d snapshots of a 3d model at different distances from the source .
one can see that , as the source approaches , the following changes in the background activity emerge : a ) the intensity ( total number of particles ) increases ; b ) particles of larger size become relatively more numerous ; c ) particle clustering becomes more prominent ; d ) the correlation radius increases ; e ) coherent structures emerge .
in other words , the model exhibits a broad set of premonitory phenomena previously observed heuristically in real and modeled systems : multiple fracturing @xcite , seismicity @xcite , socio - economics @xcite , percolation @xcite , hydrodynamics , hierarchical models of extreme event development @xcite .
these phenomena are at the heart of earthquake prediction algorithms well validated during 20 years of forward world - wide tests ( see _ e.g. , _
@xcite ) . in this paper
we analyse only the first - moment properties of the system ; such properties can explain the premonitory intensity increase ( item a above ) and transformation of the particle rank distribution ( item b ) . at the same time , the framework developed here allows one to quantitatively analyze other premonitory phenomena ; this can be readily done by considering the higher - moment properties .
v. i. keilis - borok and a. a. soloviev , a. a. ( eds ) , _ nonlinear dynamics of the lithosphere and earthquake prediction .
_ ( springer , heidelberg , 2003 ) .
d. sornette , _ critical phenomena in natural sciences .
_ 2-nd ed .
( springer - verlag , heidelberg , 2004 ) .
s. albeverio , v. jentsch , and h. kantz ( eds ) , _ extreme events in nature and society _ ( springer , heidelberg , 2005 ) .
p. embrechts , c. kluppelberg , and t. mikosch , _ modelling extremal events for insurance and finance _
( springer , 2004 ) .
u. frisch , _ turbulence : the legacy of a. n. kolmogorov _
( cambridge university press , 1996 ) . v.
i. keilis - borok , proc .
usa , * 93 * , 3748 - 3755 ( 1996 ) . v.
i. keilis - borok , ann .
earth planet .
, * 30 * , 1 - 33 ( 2002 ) .
j. rundle , d. turcotte , and w. klein ( eds ) , _ geocomplexity and the physics of earthquakes .
_ ( agu , washington dc , 2000 ) .
d. l. turcotte , ann .
earth planet .
, * 19 * , 263 - 281 ( 1991 ) .
s. c. jaume and l. r. sykes , pure appl .
geophys . ,
* 155 * ( 2 - 4 ) : 279 - 305 ( 1999 ) . | we propose a framework for studying predictability of extreme events in complex systems .
major conceptual elements _ direct cascading _ or _ fragmentation _ , _ spatial dynamics _ , and _ external driving _
are combined in a classical age - dependent multi - type branching diffusion process with immigration .
a complete analytic description of the size- and space - dependent distributions of particles is derived .
we then formulate an extreme event prediction problem and determine characteristic patterns of the system behavior as an extreme event approaches . in particlular ,
our results imply specific premonitory deviations from self - similarity , which have been heuristically observed in real - world and modeled complex systems .
our results suggest a simple universal mechanism of such premonitory patterns and natural framework for their analytic study . |
the atomistic quantum transport simulation results have been obtained from the self consistent solution of 3d - poisson equation and non - equilibrium green s functions ( negf ) method using the nanoelectronics modeling tool nemo5 @xcite .
the poisson equation provides the potential for negf method and takes the free charge in return .
the tight - binding hamiltonian of phosphorene used in negf calculations employs a 10 bands @xmath44 model .
phosphorene is a material with anisotropic dielectric properties .
the details of the poisson equation with anisotropic dielectric tensor and negf equations can be found in our previous works @xcite .
this work was supported in part by the center for low energy systems technology ( least ) , one of six centers of starnet , a semiconductor research corporation program sponsored by marco and darpa .
1 a. m. ionescu , and h. riel , _
`` tunnel field - effect transistors as energy - efficient electronic switches , '' _ nature , vol .
479 , pp . 329337
( 2011 ) .
k. bernstein , r. k. cavin , w. porod , a. c. seabaugh , and j. welser _ `` device and architectures outlook for beyond cmos switches , '' _ proc .
ieee 98 , 21692184 ( 2010 ) .
j. wang , and m. lundstrom .
_ `` does source - to - drain tunneling limit the ultimate scaling of mosfets ? '' _ ieee , international electron devices meeting , iedm ( 2002 ) .
j. appenzeller , y .- m .
lin , j. knoch , and ph .
avouris , _
`` band - to - band tunneling in carbon nanotube field - effect transistors , '' _ phys .
19 , pp . 196805
( 2004 ) .
j. appenzeller , y .- m .
lin , j. knoch , z. chen , and ph .
avouris , _ `` comparing carbon nanotube transistors - the ideal choice : a novel tunneling device design , '' _ ieee trans . on electron dev .
52 , 2568 - 2576 ( 2005 ) .
u. e. avci , and i. young , _ `` heterojunction tfet scaling and resonant - tfet for steep subthreshold slope at sub-9 nm gate - length , '' _ in ieee international electron devices meeting ( iedm ) , pp . 4 - 3 .
( 2013 ) .
h. ilatikhameneh , y. tan , b. novakovic , g. klimeck , r. rahman , j. appenzeller , _
`` tunnel field - effect transistors in 2d transition metal dichalcogenide materials , '' _ ieee exploratory solid - state computational devices and circuits , vol . 1 , no . 1 , pp .
12 - 18 ( 2015 ) .
g. fiori , f. bonaccorso , g. iannaccone , t. palacios , d. neumaier , a. seabaugh , s. k. banerjee , and l. colombo , _
`` electronics based on two - dimensional materials , '' _ nature nanotechnology , vol . 9 , no . 10 , pp . 768 - 779 ( 2014 ) .
h. liu , a. t. neal , z. zhu , z. luo , x. xu , d. tomanek , and p. d. ye , _
`` phosphorene : an unexplored 2d semiconductor with a high hole mobility , '' _ acs nano , vol . 8 , no .
4 , pp . 40334041 , ( 2014 ) .
j. qiao , x. kong , z. x. hu , f. yang , and w. ji , _ `` high - mobility transport anisotropy and linear dichroism in few - layer black phosphorus , '' _ nature communications , 5 ( 2014 ) .
y. cai , g. zhang , and y. w. zhang , _ `` layer - dependent band alignment and work function of few - layer phosphorene , '' _ scientific reports , 4 ( 2014 ) .
s. das , w. zhang , m. demarteau , a. hoffmann , m. dubey , and a. roelofs , _
`` tunable transport gap in phosphorene , '' _ nano letters , 14(10 ) , 5733 - 5739 ( 2014 ) .
h. ilatikhameneh , g. klimeck , and r. rahman _ `` can homojunction tunnel fets scale below 10 nm ? '' _ ieee edl , vol .
115 - 118 ( 2016 ) .
t. a. ameen , h. ilatikhameneh , g. klimeck , and r. rahman _ `` few - layer phosphorene : an ideal 2d material for tunnel transistors , '' _ [ online . ] arxiv:1512.05021 ( 2015 ) .
f. w. chen , h. ilatikhameneh , g. klimeck , z. chen , and r. rahman , _
`` configurable electrostatically doped high performance bilayer graphene tunnel fet , '' _ ieee journal of the electron devices society ( 2016 ) .
h. lu , and a. seabaugh , _
`` tunnel field - effect transistors : state - of - the - art , '' _ ieee electron devices society , vol .
44 - 49 ( 2014 ) .
h. ilatikhameneh , t. ameen , g. klimeck , j. appenzeller , and r. rahman , _ `` dielectric engineered tunnel field - effect transistor , '' _ ieee electron device letters ( 2015 ) , 10.1109/led.2015.2474147 .
w. li , s. sharmin , h. ilatikhameneh , r. rahman , y. lu , j. wang , x. yan , a. seabaugh , g. klimeck , d. jena , p. fay , _ `` polarization - engineered iii - nitride heterojunction tunnel field - effect transistors , '' _ ieee exploratory solid - state computational devices and circuits , vol . 1 , no . 1 , pp
. 28 - 34 ( 2015 ) , doi:10.1109/jxcdc.2015.2426433 .
m. salmani - jelodar , s. mehrotra , h. ilatikhameneh , and g. klimeck , _
`` design guidelines for sub-12 nm nanowire mosfets , '' _ ieee trans .
on nanotechnology , vol . 14 , no .
210 - 213 ( 2015 ) .
r. b. salazar , h. ilatikhameneh , r. rahman , g. klimeck , j. appenzeller , _
`` a predictive analytic model for high - performance tunneling field - effect transistors approaching non - equilibrium green s function simulations , '' _ journal of applied physics , 118 , 164305 ( 2015 ) , doi : http://dx.doi.org/10.1063/1.4934682 j. e. fonseca , t. kubis , m. povolotskyi , b. novakovic , a. ajoy , g. hegde , h. ilatikhameneh , z. jiang , p. sengupta , y. tan , g. klimeck , _ `` efficient and realistic device modeling from atomic detail to the nanoscale , '' _ journal of computational electronics , vol .
592 - 600 ( 2013 ) .
s. steiger , m. povolotskyi , h. h. park , t. kubis , and g. klimeck , _
`` nemo5 : a parallel multiscale nanoelectronics modeling tool , '' _ ieee transaction on nanotechnology , vol . 10 , no . 6 , pp .
1464 - 1474 ( 2011 ) .
h. ilatikhameneh , r. b. salazar , g. klimeck , r. rahman , j. appenzeller , _ from fowler - nordheim to non - equilibrium green s function modeling of tunneling _ , ( 2015 ) [ online . ] http://arxiv.org/abs/1509.08170 e. o. kane , _
`` zener tunneling in semiconductors , '' _ journal of physics and chemistry of solids , vol .
181 - 188 ( 1960 ) .
t. krishnamohan , d. kim , s. raghunathan , and k. saraswat .
_ `` double - gate strained - ge heterostructure tunneling fet ( tfet ) with record high drive currents and subthreshold slope < < 60mv / dec , '' _ in ieee international electron devices meeting iedm , pp . 1 - 3 ( 2008 )
. j. knoch , and j. appenzeller .
_ `` modeling of high - performance p - type iii v heterojunction tunnel fets , '' _ ieee electron device letters , vol .
305 - 307 ( 2010 ) .
a.k.a lu et al .
_ `` origin of the performances degradation of two - dimensional - based metal - oxide - semiconductor field effect transistors in the sub-10 nm regime : a first - principles study , '' _ applied physics letters 108.4 , pp .
043504 ( 2016 )
h. i. came up with the idea of the l - gate tfet .
h. i. , t. a. , and b. n. worked on the atomistic simulations and analyzed the data .
y. t. provided the tight binding model for phosphorene .
g. k. and r.r .
supervised the work .
all authors contributed to writing the manuscript .
the authors declare no competing financial interests . | scaling transistors dimensions has been the thrust for the semiconductor industry in the last 4 decades . however , scaling channel lengths beyond 10 nm has become exceptionally challenging due to the direct tunneling between source and drain which degrades gate control , switching functionality , and worsens power dissipation .
fortunately , the emergence of novel classes of materials with exotic properties in recent times has opened up new avenues in device design . here
, we show that by using channel materials with an anisotropic effective mass , the channel can be scaled down to 1 nm and still provide an excellent switching performance in both mosfets and tfets . in the case of tfets ,
a novel design has been proposed to take advantage of anisotropic mass in both on- and off - state of the tfets .
full - band atomistic quantum transport simulations of phosphorene nanoribbon mosfets and tfets based on the new design have been performed as a proof .
[ 1]*_keywords _ * # 1 shrinking the size of metal oxide semiconductor field effect transistors ( mosfets ) has improved the functionality , speed , and cost of microprocessors over the last four decades . however , the advantages of scaling are quickly fading away @xcite .
for example , the operational frequency of cpus has stopped improving since 2003 due to power consumption of cpus reaching their cooling limit ( @xmath0100w/@xmath1 ) @xcite . moreover , scaling down @xmath2 towards the few nanometer regime is becoming more challenging due to source - to - drain ( sd ) leakage current @xcite ; the gate controlled potential barrier becomes more transparent as channel becomes shorter and direct sd tunneling increases .
another challenge in miniaturizing mosfets is scaling down the supply voltage @xmath3 @xcite .
a smaller @xmath3 can be achieved in a switch with sharper on to off transition .
however , the steepness of conventional mosfets have a fundamental limit due to thermionic injection of carriers over the channel barrier ( 60 @xmath4 at room temperature ) .
accordingly , @xmath3 in mosfets does not scale very well . on the other hand , tunnel fets ( tfets )
can , in principle , provide steeper switching @xcite .
nevertheless , scaling tfets is even trickier than mosfets , since scaling affects both on- and off - states of the tfets @xcite .
hence , the tremendous improvement in processing power of transistors every few years linked to the dimension scaling and empirically described by moore s law has reached a dead end .
fortunately , it is shown here that 2d materials with anisotropic effective mass ( @xmath5 ) can be used to solve these problems and save moore s law .
0.6 .,title="fig : " ] [ fig : same_eot ] first , we discuss the source - to - drain tunneling challenge of the ultra scaled mosfets .
reducing the channel size makes the potential barrier more transparent . to visualize this ,
the transmission is shown in colormap on a logarithm scale and with an overlayed band diagram of mosfets in fig .
[ fig : fig1 ] .
the band diagram and transmission profile of a 12 nm and 5 nm long channel gaas mosfet are compared respectively in fig .
[ fig : fig1]a and [ fig : fig1]b . 5
nm long channel gaas mosfet suffers significantly from sd leakage which reduces the gate control .
equation ( [ eq : tun1 ] ) shows the dependence of tunneling current through barrier on @xmath5 of the channel material . according to equ .
( [ eq : tun1 ] ) , an apparent solution to the high transparency of channel barriers in short channel regime is a channel material with higher effective mass . @xmath6
although high @xmath5 channel materials block sd tunneling effectively , they have a set of drawbacks too .
quantum capacitance ( @xmath7 ) of channel material increases as a result of larger density of states ( dos ) and @xmath5 .
accordingly , the gate capacitance ( @xmath8 ) which is the net series capacitance of @xmath7 and oxide capacitance ( @xmath9 ) increases .
hence , a larger @xmath5 translates into a larger switching delay ( @xmath10 ) .
@xmath11 anisotropic effective mass can provide a solution to this problem with reducing @xmath7 by a factor of @xmath12 .
this reduction of @xmath7 is the result of the decreased density of states ( @xmath13 ) in anisotropic materials : @xmath14 where @xmath15 and @xmath16 are low and high effective masses of the channel material along its two main axes .
if high @xmath5 axis of channel is aligned with transport direction and low @xmath5 axis is aligned with the confinement direction , both low transparency and small switching delay can be achieved .
note that high @xmath5 along the channel increases the carriers decay rate through barrier exponentially , whereas low confinement @xmath5 reduces dos and @xmath7 .
hence , a 2d material such as phosphorene @xcite with anisotropic @xmath5 @xcite can provide an excellent switching performance in mosfets ensuring the continuation of moore s law to atomic dimensions .
0.46 0.36 here , we discuss the scaling challenge of tfets .
although tfets were intended to reduce the power consumption of transistors @xcite , scaling tfets below 10 nm is even more challenging than mosfets @xcite .
the on - state and off - state tunneling currents ( i@xmath17 and i@xmath18 ) depend on the same device parameters @xcite . thus decreasing
i@xmath18 would reduce i@xmath17 .
roughly , the on / off ratio of tfets depends on @xcite : @xmath19 where @xmath20 and @xmath2 are the tunneling distances in the on- and off - state respectively . @xmath21 and @xmath22 ( @xmath23 and @xmath24 ) are the reduced effective mass and the bandgap of the channel material ( source - to - channel junction ) , respectively
. shrinking the channel length to few nanometers brings @xmath25 close to 1 and reduces i@xmath17/i@xmath18 significantly .
one apparent solution can be a heterostructure channel where the term @xmath26 is much smaller than @xmath27 due to different materials used in the source and channel regions @xcite .
however , heterostructure tfets suffer from interface states which deteriorate their off - state performance @xcite .
although homojunction tfets do not have the interface states , it is challenging to provide high on / off ratio especially below 6 nm @xcite .
anisotropic effective mass can also provide a solution for this challenge by setting source - channel junction along low @xmath5 axis of channel material and the channel barrier along high @xmath5 axis .
although , many novel materials and designs have been proposed to enhance the performance of tfets such as 2d material tfets @xcite , nitride heterostructures @xcite , dielectric engineering @xcite , there are not many proposals for solving the scaling challenge of tfets @xcite . in this work
, a new tfet design is proposed to overcome the scaling challenge and enable downsizing to 2 nm channel lengths . fig .
[ fig : fig2]a shows a novel tfet device structure to take advantage of anisotropic effective mass .
notice that the gate is l - shaped .
fig .
[ fig : fig2]b depicts that the tunneling in the on - state occurs along the low @xmath5 axis of the channel enhancing the i@xmath17 .
however , the tunneling in the off - state occurs along the high @xmath5 axis and results in a very low i@xmath18 .
hence , this new tfet design can revive moore s law for sub-10 nm tfets . in this work ,
phosphorene nanoribbon has been chosen as the channel material since it has a large effective mass anisotropy in zigzag and armchair directions .
moreover , multi - layer phosphorene provides a range of bandgap ( @xmath281.45 to 0.4ev @xcite ) suitable for transistor applications .
@xmath29 of monolayer ( 1l- ) and bilayer ( 2l- ) phosphorene is about 1.45ev and 0.8ev , respectively .
since mosfets require larger @xmath29 for a smaller source - to - drain leakage , a monolayer phosphorene has been used here .
the situation is more tricky in tfets which need optimized @xmath29 .
it was shown previously that 2l - phosphorene has an optimum @xmath29 , and hence 2l has been chosen for tfets .
hfo@xmath30 is used as the gate dielectric with an equivalent oxide thickness ( eot ) of 0.5 nm in both mosfets and tfets and i@xmath18 is set to @xmath31 .
fig .
[ fig : fig3]a compares @xmath32-@xmath33 of a conventional 2l - phosphorene nanoribbon along zigzag and armchair transport directions with l - shaped gate ( l - gate ) tfet calculated from full - band atomistic quantum transport simulations using nemo5 @xcite . not only does the l - gate tfet have i@xmath17 close to that of the armchair ribbon ( low @xmath5 ) , but it also has i@xmath18 similar to that of the zigzag ribbon ( high @xmath5 ) .
hence , the l - gate design has the advantages of both low and high @xmath5 devices simultaneously : high i@xmath17 and low i@xmath18 .
the performance of the l - gate tfet depends on the length @xmath34 ( see fig . [
fig : fig3]b ) which determines the width for on - current . in conventional tfets
, @xmath34 equals 0 .
fig .
[ fig : fig3]b shows i@xmath17 of l - gate tfet as a function of @xmath34 for a fixed i@xmath18 of 10@xmath35 .
increasing @xmath34 enhances i@xmath17 significantly , however it reduces the source extension by @xmath36 .
accordingly , there is a limit on @xmath34 according to the footage requirements in the design .
nevertheless , a @xmath34 of about 2.5 nm can improve the performance of tfet approximately by 2 orders of magnitude .
0.25 -@xmath33 of conventional 2l - phosphorene nanoribbons along zigzag and armchair directions with that of the l - gate tfet .
b ) on - current of l - gate tfet as a function of @xmath34 .
c ) impact of channel length scaling on @xmath32-@xmath33 of l - gate tfets .
d ) i@xmath17/i@xmath18 ratio of the l - gate tfet as a function of @xmath2 . , title="fig : " ] 0.25 -@xmath33 of conventional 2l - phosphorene nanoribbons along zigzag and armchair directions with that of the l - gate tfet .
b ) on - current of l - gate tfet as a function of @xmath34 .
c ) impact of channel length scaling on @xmath32-@xmath33 of l - gate tfets .
d ) i@xmath17/i@xmath18 ratio of the l - gate tfet as a function of @xmath2 .
, title="fig : " ] 0.25 -@xmath33 of conventional 2l - phosphorene nanoribbons along zigzag and armchair directions with that of the l - gate tfet .
b ) on - current of l - gate tfet as a function of @xmath34 .
c ) impact of channel length scaling on @xmath32-@xmath33 of l - gate tfets .
d ) i@xmath17/i@xmath18 ratio of the l - gate tfet as a function of @xmath2 . ,
title="fig : " ] 0.25 -@xmath33 of conventional 2l - phosphorene nanoribbons along zigzag and armchair directions with that of the l - gate tfet .
b ) on - current of l - gate tfet as a function of @xmath34 .
c ) impact of channel length scaling on @xmath32-@xmath33 of l - gate tfets .
d ) i@xmath17/i@xmath18 ratio of the l - gate tfet as a function of @xmath2 . , title="fig : " ] fig .
[ fig : fig3]c shows ultra - scaled l - gate tfets with channel lengths from 9 nm down to 2.3 nm with a @xmath3 of 0.2v . in ultra - scaled tfets
, @xmath3 can not scale below @xmath37 since the maximum tunneling energy window is limited by @xmath3 .
the l - gate tfets with @xmath2 above 2 nm provide i@xmath17/i@xmath38 and satisfy the minimal itrs requirement for i@xmath17/i@xmath18 ratio .
although the l - gate design has improved the performance of tfets significantly , the on - state performance of tfet decreases for devices with @xmath2 and @xmath3 below 2 nm and 0.2v , respectively .
fig .
[ fig : fig3]d shows i@xmath17/i@xmath18 ratio of l - gate tfets as a function of @xmath2 .
ultra - scaled channel lengths put a limit on @xmath34 .
hence , @xmath34 shrinks down from 3.5 nm to 1 nm when the channel length scales down from 9 nm to 2.3 nm .
l - gate tfets with channel lengths down to 2 nm provide i@xmath17/i@xmath18 ratio larger than 10@xmath39 ( required by itrs as minimum amount of i@xmath17/i@xmath18 ratio ) .
this result proves that l - gate tfets with a channel material of anisotropic @xmath5 enable successful scaling of tfets down to the ultimate limit ; a channel with a few atoms .
[ sec : opt ] 0.4 -@xmath33 of phosphorene nanoribbon mosfets with transport direction along high
@xmath5 ( zigzag : blue ) and low @xmath5 ( armchair : red ) axes .
c ) impact of @xmath2 scaling on @xmath32-@xmath33 of phosphorene mosfets .
d ) i@xmath17/i@xmath18 ratio of mosfets as a function of @xmath2 along zigzag and armchair transport directions.,title="fig : " ] 0.27 -@xmath33 of phosphorene nanoribbon mosfets with transport direction along high @xmath5 ( zigzag : blue ) and low @xmath5 ( armchair : red ) axes .
c ) impact of @xmath2 scaling on @xmath32-@xmath33 of phosphorene mosfets .
d ) i@xmath17/i@xmath18 ratio of mosfets as a function of @xmath2 along zigzag and armchair transport directions.,title="fig : " ] 0.27 -@xmath33 of phosphorene nanoribbon mosfets with transport direction along high @xmath5 ( zigzag : blue ) and low @xmath5 ( armchair : red ) axes .
c ) impact of @xmath2 scaling on @xmath32-@xmath33 of phosphorene mosfets .
d ) i@xmath17/i@xmath18 ratio of mosfets as a function of @xmath2 along zigzag and armchair transport directions.,title="fig : " ] 0.27 -@xmath33 of phosphorene nanoribbon mosfets with transport direction along high @xmath5 ( zigzag : blue ) and low @xmath5 ( armchair : red ) axes .
c ) impact of @xmath2 scaling on @xmath32-@xmath33 of phosphorene mosfets .
d ) i@xmath17/i@xmath18 ratio of mosfets as a function of @xmath2 along zigzag and armchair transport directions.,title="fig : " ] as mentioned before , ultra - scaled mosfets require large @xmath5 and @xmath29 to block source - to - drain tunneling . hence , 1l - phosphorene nanoribbon has been chosen here which has the highest @xmath5 and @xmath29 compared to multilayer phosphorene .
the schematic of the 1l - phosphorene mosfet has been shown in fig .
[ fig : fig4]a .
the supply voltage is fixed to 0.5v , much higher than @xmath3 of tfets , since the boltzmann limit of subthreshold swing in mosfets ( i.e. 60 mv / decade in room temperature ) does not allow the scaling of @xmath3 .
the transfer characteristics of a short channel 1l - phosphorene ( @xmath2=3 nm ) with transport direction along low @xmath5 ( armchair ) and high @xmath5 ( zigzag ) are compared in fig .
[ fig : fig4]b .
as expected , the gate efficiency of a phosphorene mosfet is much better when the high @xmath5 ( zigzag ) axis is along the transport direction .
this better gate efficiency improves the subthreshold slope of mosfet significantly .
fig .
[ fig : fig4]c shows @xmath32-@xmath33 of zigzag scaled phosphorene mosfets with channel lengths from 12 nm to 1.6 nm . notice that for phosphorene mosfets with @xmath40 an @xmath41 larger than @xmath42 and an i@xmath17/i@xmath18 ratio larger than @xmath43 have been achieved .
1l - phosphorene mosfets show a significant advantage over other 2d materials whose performances are diminished below 5 nm channel lengths @xcite .
mosfets with long channels do not suffer from source - to - drain tunneling .
accordingly , a high transport @xmath5 is not required for blocking this leakage current .
actually , in long channel regime , a low transport @xmath5 can be beneficial and enhance the on - state performance of the transistor since it leads to a higher carrier injection velocity .
fig .
[ fig : fig4]d shows i@xmath17/i@xmath18 ratio of phosphorene nanoribbon mosfets as a function of @xmath2 along zigzag and armchair transport directions . although zigzag nanoribbon mosfets significantly outperform the armchair ones in short channels due to lower source - to - drain tunneling , armchair nanoribbon mosfets show a better performance in longer channels due to higher injection velocity .
there is a critical channel length ( i.e. 6 nm in 1l - phosphorene ) in mosfets below which having a low @xmath5 becomes critical and above which a high @xmath5 is beneficial . in summary ,
the channel materials with anisotropic effective mass can be used to design transistors scalable to 1 - 2 nm channel lengths . in mosfets , the high effective mass along transport direction blocks the direct source to drain tunneling and low effective mass reduces the quantum capacitance and switching delay . on the other hand in tfets , a novel l - shaped gate design is proposed which can provide advantage of high tunneling rate in the on - state and low tunneling rate in off - state by engineering the tunneling paths along low and high effective mass directions . in summary , anisotropic effective mass can be used in an l - gate design to obtain large on / off ratio in an ultra - scaled homojunction tfet . |
the very existence of a massive cluster at z @xmath0 0.5 is problematic for standard cdm theories of hierarchical structure formation ( evrard 1989 ; peebles 1989 ) .
while this problem has been recognized for some time , it has not been taken too seriously because of the lack of conclusive evidence of the existence of such massive systems at high redshift .
naturally the statistics on the abundance of high z clusters are poor because the number of such clusters known at present is small .
furthermore , one might argue that many of these clusters may not be as massive as they appear due to inflation of the measured velocity dispersion from field galaxy contamination and projection effects ( see among others frenk 1990 ) .
however , the number of distant clusters is steadily increasing due to the combined use of telescopes such as hst ( hubble space telescope ) and keck , as we have heard from many contributors at this meeting .
x - ray observations of distant ( z @xmath0 0.7 ) optically selected clusters have shown that only a few clusters are detected ( see among others castander 1994 ) , and those few are relatively modest x - ray emitters ( l@xmath2 erg s@xmath3 ) , with l@xmath4 much lower than that of present day clusters . while this fact would seem to imply that extremely rich systems with large amounts of x - ray emitting gas do not exist at these high redshifts , i will show several examples that indicate that this is not the case . in table 1 , five examples of emss ( einstein medium sensitivity survey ; gioia 1990 , stocke 1991 ) clusters with average z = 0.66 , and z@xmath5 = 0.826 are listed .
in addition to their high x - ray luminosity , there is other compelling evidence that these clusters are genuinely massive .
three of the [email protected] - 0.7 clusters contain large lensed arcs which allow a crude estimate of the projected mass in the cluster cores . for four of them
weak lensing studies have been performed .
the letter `` s '' in the third column of table 1 indicates the presence of giant arcs in the cluster core , while the letter `` w '' indicates that weak lensing analyses have been performed .
l@xmath4 is given in units of 10@xmath7 erg s@xmath3 in the [email protected] kev band , t@xmath4 is measured in the 2@xmath810 kev band and n@xmath9 is the central richness of the cluster ( defined in bahcall 1981 ; notice that in the same units coma has n@xmath9=28 ) .
.emss clusters with z @xmath0 0.5 [ cols= " < , > , > , > , > , < , < " , ]
since its discovery ms0015.9@xmath116 ( alias cl0016@xmath116 , koo 1981 ) has been the archetypal rich , x - ray luminous , distant cluster .
it is part of the emss since it was `` rediscovered '' in an observation pointed at another target .
the cluster is very rich and has a linear structure elongated in the ne - sw direction .
smail ( 1994 ) reported detection of weak shear , which has been reconfirmed by the analysis performed by luppino ( 1996 ) using different deep images with a larger field of view and a different technique ( kaiser and squires , 1993 ; squires and kaiser , 1996 ) .
two hours of r band keck data are in hand but the images have not been assembled yet ( clowe , private communication ) .
ms0451@xmath803 with an l@xmath4 of almost 2@xmath1010@xmath11erg s@xmath3 is the most x - ray luminous cluster in the emss and among the brightest clusters known .
asca data by donahue ( 1996 ) show a hot cluster at 10.4 kev with iron abundance of 15% solar and a total mass ( within 1h@xmath12mpc ) of 9.7@xmath13 m. the shear signal detection , performed by luppino ( 1996 ) on a 7200s r band image taken with the uh 2.2 m telescope is at @xmath05@xmath14 .
two hours of r band keck data are in hand also for this cluster but the analysis has not been performed yet .
ms1054@xmath803 is the most distant cluster of the emss and among the most distant x - ray selected clusters known .
the cluster has a filamentary morphology with the x - ray coming from the center and elongated in the same e@xmath8w direction as the optical galaxies ( donahue 1997 ) .
ms1054@xmath803 is extremely rich and quite hot at 14.7@xmath15 kev as obtained by asca ( donahue 1997 ) .
a strong shear signal at 6@xmath14 level is detected ( luppino & kaiser , 1997 ) .
the total mass ( within 1 mpc ) from x - rays and from weak lensing are consistent ( 2 - 6@xmath16 mvs 3 - 30@xmath16 m ) .
deep imaging of ms1137@xmath166 with keck and with the uh 2.2 m have been collected . from the reduced images a large arc has been discovered close to the cluster center ( clowe 1997 ) .
differently from ms1054@xmath803 and ms0015.9 @xmath116 this cluster is compact and concentrated with no filamentary structure . a weak lensing analysis , performed by clowe on a 8700 second r - band exposure of keck , using the i band ( 7500 s ) 2.2 m data as a color selection to remove cluster galaxies , finds a nice centrally concentrated mass peak falling exactly on the brightest cluster galaxy . the mass from weak lensing comes out to be 2.9@xmath17 m at 500 @xmath18 kpc ( assuming the background galaxies lie at z=2 ) .
also for ms2053 - 04 two hours of r band keck data are in hand but the images have not been assembled .
a recent observation of this cluster has been performed with the italian - dutch bepposax satellite by scaramella ( in preparation ) .
the data have not been fully reduced yet .
a preliminary analysis shows this cluster to be much cooler than ms1054@xmath803 or ms0451@xmath803 .
the north ecliptic pole ( nep ) region of the rosat all - sky survey ( rass ; trmper 1991 ) has the largest exposure time ( approaching 10 ks ) of the all rass .
the nep region covers a 9@xmath19 field , and contains a total of 465 x - ray sources detected at @xmath0 4@xmath14 in the [email protected] kev ( mullis 1998 ) .
we are identifying all sources in the field .
the principal derivative is a statistically complete sample of galaxy clusters appropriate for a better characterization of the x - ray luminosity function evolution .
we have discovered a very distant cluster in the nep at z= 0.81 .
rxj1716@xmath166 ( henry 1997 ) is among the most distant x - ray selected clusters together with ms1054@xmath803 and the x - ray clusters detected by rosati ( 1997 ) in the rdcs ( rosat deep cluster survey ) or ebeling ( 1997 ) in the warps ( wide angle rosat pointed survey ) . as with ms1054@xmath803
, it is not likely that rxj1716@xmath166 is in virial equilibrium .
the galaxies in rxj1716@xmath166 are in an inverted s - shaped filament running northeast to southwest ( fig .
cluster members extend all along the s. the distance from the top to the bottom of the s is about 1.5 mpc ( h@xmath2050 km s@xmath3 mpc@xmath3 , q@xmath21=0.5 ) .
x - ray follow - up observations with asca ( 100 ks ) and with the rosat hri ( 171 ks ) provide a temperature in 2@xmath810 kev of kt = 6.7 @xmath22 ( 68% confidence ) and a flux in [email protected] kev of [email protected] @xmath10 10@xmath24 erg @xmath25s@xmath3 .
it is intriguing that the morphology of rxj1716@xmath166 and ms1054@xmath803 is filamentary with the x - rays coming from the center .
we note this since the initial formation of protoclusters is often described as matter flowing along filaments ( bond 1996 ) with the x - rays generated at the impact point of the two colliding streams of matter ( henry 1997 ) .
11.5truecm fig .
1-a 1024@xmath101024 subarray image of rxj1716@xmath166 extracted from the center of a 4500s exposure in the i - band taken by g. luppino and m. metzger with the uh 8k@xmath108k ccd mosaic camera on the cfht prime focus in sept 1995 .
north is up and east to the left .
this image spans [email protected] ( 0.9h@xmath3 mpc at @xmath260.81 ) at a scale of 0.21/pixel .
both cfht 8k@xmath108k mosaic ccd deep images and keck r - band ( 7500s ) images were taken .
the weak lensing analysis was performed on the r - band keck image , using 2.2 m i - band data ( 26,100s ) for color terms ( clowe 1997 ) .
the mass peak is just east of the bcg , and an arm of mass to the ne , neatly following the line of galaxies ne of the cluster core is detected .
doug clowe analysis gives a very strong signal of 4@xmath27 m(background galaxies at z=1.5 ) .
the m / l@xmath28 is equal to 210@xmath29 .
both the optical lightmap and weak lensing massmap have two spatially distinct massive sub - clusters , as well as a long filamentary structure .
there is clear evidence from the existing data that large , dense mass concentrations existed at an early epoch .
another possible search strategy to find distant x - ray clusters is to look for clusters around powerful radio galaxies .
high z massive structures have been found around several 3c radio galaxies ( i.e. : 3c184 , at [email protected] , deltorn 1997 ; 3c324 , at [email protected] , smail & dickinson , 1995 , just to quote a few ) .
cluster abundances has been cited as one of the strongest evidence against the standard cdm model as normalized to reproduce the microwave background anisotropies seen by cobe satellite .
the number of high - z massive clusters predicted by standard cdm , or other mixed dark matter models , is too low with respect to the number of clusters observed . in low - density universes
there is less evolution and nearly all 10@xmath31 mclusters formed by [email protected] . in a high - density universe , instead
, only 5% of the present day 10@xmath31 mclusters have formed by the same redshift .
thus massive clusters should be much rarer at epochs earlier than 0.5 if @xmath32=1 , contrary to the observations .
alternatives have been suggested with low @xmath32 models ( viana & liddle , 1996 ; eke 1996 ) .
the open cdm ( @xmath32=0.3 ) and @xmath33-dominated cdm models are preferred because they are compatible with the data .
it may be possible to estimate @xmath32 from forthcoming observations of intermediate - distant clusters ( up to z@xmath61 ) .
i am grateful to p. henry , c. mullis , n. kaiser , g. luppino , m. donahue , d. clowe for stimulation , help and advice .
this work has received partial finacial support from nasa - stsci grant go-5402.01 - 93a and go-05987.02 - 94a , from nsf ast95 - 00515 , from nasa grant nag5 - 2594 and asi grants ars-94 - 10 and ars-96 - 13 .
bahcall , n. , 1981 , , 247 , 787 .
bond , j.r . ,
kofman , l. & pogosyan , d. , 1996 , nature , 380 , 603 .
carlberg , r. , yee , h. & ellingson , e. , 1994 , , 437 , 63 castander , f. , ellis , r. , frenk , c. , dressler , a. & gunn , j. , 1994 , , 424 , l79 .
clowe , d.i . ,
luppino , g.a . ,
kaiser , n. , henry , j.p .
& gioia , i.m . , 1997 , , submitted .
deltorn , j - m , le fevre , o. , crampton , d. & dickinson , m. , 1997 , , 483 , l21 .
donahue , m. , 1996 , , 468 , 79 .
donahue , m. , gioia , i.m .
, luppino , g. , hughes , j.p . & stocke , j.t . , 1997 , , submitted .
dressler , a. & gunn , j. , 1992 , , 78 , 1 .
ebeling , 1997 , , submitted .
eke , v.r . , cole , s. & frenk , c.s . , 1996 , , 282 , 263 .
evrard , a. , 1989 , , 341 , l71 .
frenk , c.s . , white , s.d . ,
efstathiou , g. & davis , m. , 1990 , , 351 , 10 .
furuzawa , a. , yamashita , k. , tawara , y. , tanaka , y. & sonobe , t. , 1994 , in `` new horizon of x - ray astronomy '' , eds f. makino & t. ohashi , ( universal academy press : tokyo ) , pag 541 .
gioia , i.m . , 1990 , , 72 , 567 .
gioia , i.m . & luppino , g.a . , 1994 , , 94 , 583 .
henry , j.p . , 1997 , , 114 , 1293 .
kaiser , n. & squires , g. , 1993 , , 404 , 441 .
koo , d.c .
, 1981 , , 251 , l75 .
luppino , g.a . ,
kaiser , n. , clowe , d.i . ,
gioia , i.m . , & metzger , m.r .
, 1996 , nuclear phisics , 51b , 107 .
luppino , g. a. & gioia , i.m . , 1995 , , 445 , l77 .
luppino , g. a. & kaiser , n. 1997 , , 475 , 20 .
maccacaro , t. , 1994 , astro .
lett . & communications , gordon and breach pub . , 29 , 267 .
mullis , c.r . ,
gioia , i.m .
& henry , j.p . , 1998 , to appear in the iau symposium 188 `` the hot universe '' , kyoto , aug 26 - 30 , 1997 .
peebles , j. , daly , r. & juszkiewicz , r. , 1989 , , 347 , 563 .
rosati , p. , della ceca , r. , norman , c. & giacconi , r. , 1997 , astro - ph/9710308 .
smail , i. , ellis , r. , fitchett , m. & edge , a. , 1994 , , 270 , 245 .
smail , i. , dickinson , m. , 1995 , , 455 , l99 .
stocke , j.t .
, 1991 , , 76 , 813 .
squires , g. & kaiser , n. , 1996 , , 473 , 65 .
trmper 1991 , adv .
, 2 , 241 .
viana , p.t.p . & liddle , a.r .
, 1996 , , 281 , 323 . | as the largest gravitationally bound structures known , clusters provide clear constraints on the formation of structure and on the composition of the universe . despite their extreme importance for cosmology
the number of clusters at high redshift ( [email protected] ) is rather small .
there are only a few x - ray emitting examples reported and a handful of optically - selected ones .
these clusters can provide stringent constraints on theories of large scale structure formation , if they are massive enough .
i will review the status of these distant x - ray selected clusters .
these objects are of special importance because their x - ray emission and presence of gravitational arcs imply that they are massive , comparable to low redshift examples , and their existence is problematic for some theories of structure formation . 04ms0440@xmath10204 |
the study of @xmath0 in three dimensions during the past years has revealed very interesting physics in this theory @xcite-@xcite . at finite temperature , a version of @xmath0 , i.e.
@xmath1 , has been proposed as a model of high - temperature superconductivity @xcite .
the behaviour of the theory with increasing temperature @xmath2 and number of fermion flavours @xmath3 was recently studied numerically in the schwinger - dyson formalism in a bare - vertex approximation @xcite .
this approximation corresponds to a particular truncation of the schwinger - dyson hierarchy , and its effects , along with wave - function renormalization effects , could be important @xcite , especially in connection with the behaviour of the theory when the number of fermion flavours is varied .
this has led to several interesting studies at zero - temperature trying to include these effects @xcite-@xcite , in order to solve the controversy on whether there is a critical number of fermion flavours beyond which there is no mass generation . at finite temperature ,
similar studies have so far used several approximations that we will relax in this study .
the first purpose of this work is to provide an accurate value for the ratio @xmath4 @xcite , where @xmath5 is the fermion mass function at zero momentum and energy and @xmath6 is the critical temperature beyond which there is no mass generation .
apart from being theoretically very interesting , this quantity is of direct physical relevance since it can be compared to corresponding values measured in certain high - temperature superconductors .
the second purpose is to provide a reliable phase diagram of the theory with respect to @xmath2 and @xmath3 .
we continue to consider momentum- and energy - dependent self - energies , but we include wave - function renormalization effects and a non - bare vertex .
this complicates the study considerably and constitutes a non - trivial step forward , since instead of having only one gap equation , a system of two integral equations has to be solved .
we continue to neglect the imaginary parts of the photon polarization functions and of the fermion self - energy for simplicity , as other studies have done so far @xcite-@xcite .
their inclusion , even though in principle necessary , would double the number of coupled equations to be solved , which would render the numerical study too complicated for our algorithm and the computer power at hand .
we will employ the schwinger - dyson formalism in order to gain information on the momentum - dependent fermion self - energy and the non - perturbative physics behind it .
the schwinger - dyson equation for the fermion self - energy is given by @xmath7 where @xmath8 , @xmath9 is the dimensionful gauge coupling of the theory which we will take to be constant throughout this study , @xmath10 is the photon propagator with @xmath11 , @xmath12 is the full photon - fermion vertex , @xmath13 is a four - dimensional representation of the @xmath14-matrices , @xmath15 is the bare fermion propagator , and the finite - temperature fermion propagator in the real - time formalism is given by @xmath16 where @xmath17 , @xmath18 is the wave - function renormalization function , @xmath19 is the usual dirac function and we have made a rotation to euclidean space .
note that we avoid the matrix form that the propagator has in this formalism , since the schwinger - dyson equation that we have written down involves only a one - loop diagram directly , so complications due to the field - doubling problem do not arise @xcite .
however , it has to be noted that a more careful treatment involving the imaginary parts of the self - energies would involve the full matrix propagators @xcite .
the resulting equations would in that case be unfortunately too complicated to be solved even numerically , without serious truncations .
moreover , due to the broken lorenz invariance at finite temperature , the wave function renormalization could in principle affect differently the @xmath20 and @xmath21 propagator components , i.e. we should replace @xmath22 by @xmath23 with @xmath24 . for simplicity we will restrict ourselves to situations where @xmath25 , which correspond to a zero chemical potential , and we will work in the approximation where @xmath26 also for non - zero temperatures as done in similar studies @xcite .
for the vertex @xmath27 we use the ansatz @xmath28 , where @xmath29 is the same wave - function renormalization function appearing in the fermion propagator .
it is unfortunately not symmetric in the vertex momenta , but its use simplifies the numerical algorithm considerably .
even though this vertex does not satisfy _ a priori _ the ward - takahashi identities , it is expected to incorporate the basic qualitative features of a non - perturbative vertex at zero temperature when used in a schwinger - dyson context @xcite .
it has also been used in a finite - temperature case @xcite , supported by the qualitative agreement of the results of this ansatz with the ones obtained by more elaborate treatments @xcite .
furthermore , in the problem of the present study it gives results similar to a symmetrized vertex @xcite .
moreover , the photon propagator in the landau gauge is given by @xcite @xmath30 where @xmath31 with @xmath32 , and where we neglect its temperature - dependent delta - function part since it is expected to give a vanishingly small contribution @xcite , @xcite , @xcite .
the longitudinal and transverse photon polarization functions @xmath33 and @xmath34 are given explicitly in @xcite and taken from @xcite , where they are calculated in a massless - fermion approximation and where wave - function renormalization and vertex effects cancel only for specific vertex choices @xcite .
one should in principle couple the expressions for @xmath35 in our system of integral equations in a self - consistent manner , but unfortunately this would render our numerical algorithm too complicated . identifying the parts of this equation with the same spinor structure , we reduce the problem to that of a system of two three - dimensional integral equations involving two functions varying independently with @xmath20 and @xmath21 .
the equations take the following form : @xmath36 where @xmath37 , and it is more convenient to work with the mass function @xmath38 .
we also sum over the photon polarizations @xmath39 and over the two roots of the delta function by introducing @xmath40 .
the quantity @xmath41 is approximated by the relation @xmath42 @xcite , where use of the delta - function property @xmath43 has been made .
furthermore , the functions @xmath44 and @xmath45 are given by @xmath46 one can easily check that @xmath47 , which would reproduce the result of @xcite if one takes @xmath48 and switches to imaginary - time formalism . after inspecting the equations
, we note that the vertex ansatz we chose makes the integral giving the function @xmath49 depend only on the function @xmath50 and independent of @xmath51 .
on the other hand , the equation for @xmath52 has to be solved self - consistently , and it actually always accepts , apart from the solutions we will seek , the trivial solution as well .
an analytical study of such a system would not be possible without severe approximations , so we proceed to the numerical solution of the equations given above .
the numerical method used to solve the system of equations is the same as the one presented in @xcite .
the physical quantities of interest do not vary substantially with our ultra - violet ( uv ) cut - off , since at the effective uv cut - off @xmath53 of the theory they are already negligibly small @xcite .
the solution at @xmath54 and @xmath55 for the functions @xmath56 and @xmath57 is given in figs . 1 and 2 respectively .
the general variation of these functions with momentum does not change with increasing temperature or varying of @xmath3 , even though the overall scale of @xmath58 drops fast with @xmath3 .
the function @xmath56 falls as expected with increasing momentum , and is of the same form as the mass function @xmath52 .
the function @xmath49 is always in the range between -1 and 0 as required @xcite , it is tending to zero for increasing momenta , and it is of the same form and magnitude as the approximate form used in @xcite . for a given number of fermion flavours @xmath3 , when the temperature exceeds some critical value there is no solution for the fermion mass function but the trivial one . the value of the ratio @xmath59 is concentrated approximately around @xmath60 , which is comparable to the values obtained in @xcite which neglected @xmath49 .
the ratio @xmath59 found is also comparable to typical values obtained in ref .
@xcite , which includes wave - function renormalization effects but uses several approximations which we were able to by - pass in this study .
we confirm therefore that adding these effects does not influence the behaviour of the theory in a significant way .
we have to note moreover that our @xmath59 values are somewhat larger than the value @xmath61 measured for some high - temperature superconductors @xcite .
however , we could be overestimating this ratio because of a possibly poor convergence of the algorithm for temperatures close to the critical one . at zero - temperature
, this theory is known to exhibit also an interesting behaviour with the number of fermions @xmath3 . in fig .
3 we plot the zero - momentum and zero - temperature fermion mass function with respect to @xmath3 .
we fit the data with the exponential curve @xmath62 . at @xmath63 ,
the mass function is still roughly four times larger than the cut - off .
when @xmath64 , our algorithm does not converge and the mass function tends fast below the ir cut - off .
this behaviour could indicate that @xmath63 is some critical point beyond which dynamical mass generation is impossible .
the value of @xmath3 we find is remarkably close to the one quoted in the numerical study of @xcite , and it is also close to our previous result @xcite . a similar study @xcite , which includes a calculation of the fermion - field anomalous dimension to second - order in @xmath65 , predicts a critical value @xmath66 , which is only slightly larger than the theoretical prediction neglecting wave - function renormalization which gives @xmath67 @xcite , and quite close to the value we find numerically . in fig .
4 we plot the phase diagram of the theory with respect to @xmath3 and @xmath68 .
it separates two regions of the parameter space which either allow or do not allow dynamical mass generation .
the choice of an exponential fitting curve was only made to describe phenomenologically " the general tendency of the data and to provide a measure for a @xmath59-ratio independently of @xmath3 , and is reminiscent of the results in ref . @xcite but with a somewhat steeper slope .
however , there are also studies that predict a non - analytic behaviour of @xmath58 for @xmath3 near its critical value @xcite . lack of convergence of the algorithm and fall of the mass function below the ir cut - off
however do not allow us to test the precise behaviour of the theory near @xmath63 , and it is also clear that the fitting curves should not be extrapolated for @xmath3 larger than this value .
we were able to solve a system of two coupled integral equations for the fermion mass function and wave - function renormalization for a finite - temperature version of three - dimensional @xmath0 , by applying a numerical relaxation technique .
one main result is a @xmath59-ratio of about 10.6 , which is close to previous numerical studies , confirming that including wave - function renormalization , in conjunction with a particular non - bare fermion - photon vertex , does not affect the theory in a significant way .
the other important result is the existence of a possibly critical fermion flavour number of roughly 3.35 , which is also consistent with some theoretical expectations and other numerical results .
it is the first time that such a study includes the momentum and energy dependence of the fermion and photon self - energies and of wave - function renormalization in a gap equation with a non - bare vertex , and this allows a more reliable description of the behaviour of the theory .
we estimate the numerical uncertainty for the values quoted , which comes mainly from the convergence criteria imposed , at about @xmath69 . the next step for future studies should be the inclusion of the imaginary parts of the self - energies in the equations and the relaxing of the approximation @xmath70 , which could in principle influence the results of this investigation .
pisarski , _ phys .
* d 29 * , 2423 ( 1984 ) ; t. appelquist , m. bowick , d. karabali and l.c.r .
wijewardhana , _ phys .
* d 33 * , 3704 ( 1986 ) ; e. dagotto , j.b .
kogut and a. kocic , _ phys .
lett . _ * 62 * , 1083 ( 1989 ) ; e. dagotto , a. kocic and j.b .
kogut , _ nucl .
* b 334 * , 279 ( 1990 ) ; d. atkinson , p.w .
johnson and p. maris , _ phys .
* d 42 * , 602 ( 1990 ) .
t. appelquist , d. nash and l.c.r .
wijewardhana , _ phys .
lett . _ * 60 * , 2575 ( 1988 ) .
gusynin , v.a .
miranskii and a.v .
shpagin , hep - th/9802136 .
n. dorey and n.e .
mavromatos , _ nucl
* b 386 * , 614 ( 1992 ) .
g. triantaphyllou , _ phys .
* d 58 * , 065006 ( 1998 ) .
m. r. pennington and d. walsh , _ phys .
* b 253 * , 246 ( 1991 ) ; d.c .
curtis , m.r .
pennington and d. walsh , _ phys .
* b 295 * , 313 ( 1992 ) .
d. nash , _ phys .
lett . _ * 62 * , 3024 ( 1989 ) .
gusynin , a.h .
hams and m. reenders , _ phys .
* d 53 * , 2227 ( 1996 ) .
p. maris , _ phys .
* d 54 * , 4049 ( 1996 ) .
n. dorey and n.e .
mavromatos , _ phys
* b 256 * , 163 ( 1991 ) ; i.j.r .
aitchison , n. dorey , m. klein - kreisler and n.e .
mavromatos , _ phys
* b 294 * , 91 ( 1992 ) .
aitchison , _ z. phys .
_ , * c 67 * , 303 ( 1995 ) .
aitchison and m. klein - kreisler , _ phys .
rev . _ * d 50 * , 1068 ( 1994 ) .
n. p. landsman , c.g .
van weert , _ phys .
_ * 145 * , 145 ( 1987 ) ; r. kobes , g.w .
semenoff , _ nucl .
phys . _ * b 260 * , 714 ( 1985 ) and * b 272 * , 329 ( 1988 ) ; l. dolan and r. jackiw , _ phys .
rev . _ * d 9 * , 3320 ( 1974 ) . for interesting reviews ,
smilga , _ phys .
rept . _ * 291 * , 1 ( 1997 ) ; p.a .
henning , _ phys .
_ * 253 * , 235 ( 1995 ) .
m. r. pennington and s.p .
webb , bnl report no .
bnl-40886 , 1988 ( unpublished ) ; d. atkinson , p. w. johnson and m.r .
pennington , bnl report no .
bnl-41615 , 1988 ( unpublished ) .
g. triantaphyllou , _ technische universitt mnchen _ preprint no tum - hep-310/98 , august 1998 .
a. kocic , _ phys .
* b 189 * , 449 ( 1987 ) ; t.s .
evans , r.j .
rivers , _ z. phys .
_ * c 40 * , 293 ( 1988 ) .
see for instance c. itzykson and j .- b .
zuber , _ quantum field theory _ , mcgraw - hill 1985 .
z. schlesinger , r.t .
collins , d.l .
kaiser and f. holtzberg , _ phys .
* 59 * , 1958 ( 1987 ) . | the gap equation for fermions in a version of thermal @xmath0 in three dimensions is studied numerically in the schwinger - dyson formalism .
the interest in this theory has been recently revived since it has been proposed as a model of high - temperature superconductors .
we include wave - function renormalization in our equations and use a non - bare vertex .
we then have to solve a system of two integral equations by a relaxation algorithm .
fermion and photon self - energies varying independently with energy and momentum are used , which should produce more accurate results than in the previous literature .
the behaviour of the theory with increasing temperature and number of fermion flavours is then carefully analyzed . |
it has been suggested@xcite that the use of collective systems like a magnet can reduce the intrinsic switching energy ( that is dissipated throughout switching ) significantly compared to that required for individual spins .
there is also a lot of experimental effort@xcite at this time to implement switching circuits based on magnets .
there has been some work @xcite on modeling magnetic circuits like mqca s in the atomic scale using quantum density matrix equation but most of the work @xcite is in the classical regime using the well known micromagnetic simulators ( oommf ) based on the landau - lifshitz - gilbert ( llg ) @xcite equation .
this paper too is based on the llg equation , but our focus is not on obtaining the energy requirement of any specific device in a particular simulation .
rather it is to obtain generic results that can guide the design of magnet based switching circuits as well as providing a basis for comparison with alternative technologies .
+ the results we present are obtained by analyzing the cascadable switching scheme illustrated in fig.1 where the magnet to be switched ( magnet 2 ) is first placed along its hard axis by a magnetic pulse ( see ` mid state ' in fig.1 ) .
on removing the pulse , it falls back into one of its low energy states ( up or down ) determined by the ` bias ' provided by magnet 1 .
what makes this scheme specifically suited for logic operations is that it puts magnet 2 into a state determined by magnet 1 ( thereby transferring information ) , but the energy needed to switch magnet 2 comes largely from the external pulse _ and not from magnet 1_. this is similar to conventional electronic circuits where the energy needed to charge a capacitor comes from the power supply , although the information comes from the previous capacitors .
this feature seems to be an essential ingredient needed to _ cascade logic units_. to our knowledge , the switching scheme shown in fig.1 was first discussed by bennett @xcite and is very similar to the schemes described in many recent publications ( see e.g likharev et.al @xcite , kummamuru et.al@xcite and csaba et.al @xcite ) .
+ ) where a small bias field due to magnet 1 can tilt it upwards or downwards thereby dictating its final state on removing the pulse.,title="fig:",width=226,height=226 ] + this paper uses the llg equation to establish two central results .
one is that the switching energy drops significantly as the ramp time @xmath0 of the magnetic pulse exceeds a critical time @xmath1 given by equation .
this is similar to the drop in the switching energy of an rc circuit when @xmath2 .
but the analogy is only approximate since the switching energy for magnets drops far more abruptly with increasing @xmath0 .
the significance of @xmath1 is that it tells us how slow a pulse needs to be in order to qualify as `` adiabatic '' and thereby reduce dissipation significantly .
considering typical magnets used in the magnetic storage industry , and using ramp times of a few @xmath1 , intrinsic switching frequency of 100 mhz to 1 ghz can easily be in the adiabatic regime of switching where dissipation is very small .
+ interestingly , we find that the switching energy for the trapezoidal pulses investigated in this paper in both the ` fast ' and ` slow ' limits can be described by a single equation which is the other central result of this paper . later in this paper ( [ s6 ] ) we will discuss how equations and can be used to guide scaling and increase switching speeds . furthermore these equations can be used to compare magnet based switching circuits with alternative technologies .
+ it has to be emphasized that dissipation of the external circuitry also has to be evaluated for any new technology .
a careful evaluation would require a consideration of actual circuitry to be used ( see e.g. @xcite,@xcite ) and is beyond the scope of this paper .
however following nikonov et.al .
@xcite , if a wire coil is used to produce the pulse , we can estimate the energy dissipated in creating the field @xmath3 as @xmath4 in cgs system of units .
@xmath5 is the quality factor of the circuit and @xmath6 is the volume over which the field extends .
depending on q , v and @xmath3 the dissipated energy can be much larger , comparable to or much smaller than @xmath7 which sets the energy scale for the effects considered here in this paper .
+ _ overview of the paper _ : as mentioned before our results are based on direct numerical simulation of the llg equation .
however we find that in two limiting cases , it is possible to calculate switching energy simply using the energetics of magnetization and these limiting results are described in sections [ s3 ] ( dissipation with fast pulse ) and [ s5 ] ( dissipation with adiabatic pulse ) which are related to equation . in [ s6 ]
we use the llg equation to show that the switching energy drops sharply for ramp times larger than the critical time given by equation . in section
[ s7 ] using coupled llg equations we analyze a chain of inverters to show that the total dissipation increases linearly with the number of nanomagnets thus making it reasonable to use the one - magnet results in our paper to evaluate complex circuits , at least approximately . finally in section [ s8 ] practical issues such as dissipation versus speed , increasing the switching speed and scaling are qualitatively discussed in the light of these results . +
before we get into the discussion of switching energy , let us briefly review the energetics of a magnet .
the energy of a magnet with an effective second order uniaxial anisotropy can be described by @xmath9 where @xmath10 measures the deflection from the easy axis which we take as the @xmath11 axis .
all isotropic terms have been omitted because they do not affect dynamics and hence dissipation of the magnet @xcite .
there are two magnetic fields that control the switching ( see fig.[front ] ) : the external pulse @xmath3 and the bias field @xmath12 due to the neighboring magnet . including the internal energy and the interaction energy of magnetic moment with external fields
, the energy equation reads @xmath13 @xmath14 is the saturation magnetization .
if the unit volume is magnetized to saturation , @xmath14 is equivalent to the magnetic moment per unit volume .
@xmath15 is a unit vector in the direction of magnetization .
v is the volume of the magnet and @xmath16 is the second order anisotropy constant with dimensions of energy per unit volume .
the applied field @xmath3 is along the hard axis @xmath17 , the bias field @xmath12 is along the easy axis @xmath18 so the energy equation becomes @xmath19 where @xmath20 is defined as in a standard spherical coordinate system .
using equation [ energy ] we will show that dissipation with a fast pulse ( small ramp time ) can be written as [ disspulsenohdc ] @xmath21 for reasons to be explained , under the condition of equation [ lower ] , logic device will not work .
nevertheless it is useful for determining dissipation in the adiabatic limit . in the equations above
, @xmath22 is the minimum field necessary to put the magnet along its hard axis .
notice that the bias field @xmath12 is a dc field coming from the neighboring magnet . in practice ,
whether the bias field is a dc field or not , its magnitude has to be bigger than noise such that when the magnet is put along its hard axis as in fig.[front ] , the bias field can deterministically tilt the magnet towards its direction
. we will show in
[ s34 ] that for @xmath23 , dissipation can still be calculated using equation [ disspulsenohdc ] .
+ to derive equations [ disspulsenohdc ] we find the initial and final state energies under various conditions and evaluate the difference .
we have to emphasize that all these states essentially pertain to the energy minima ( equilibrium states ) i.e. they are either the minimum of energy ; or they represent a non - equilibrium state instantaneously after the equilibrium state ( minimum of energy ) has changed . since all the fields considered here are in the @xmath24 plane and no out - of - plane field is considered , the equilibrium states ( the energy minima ) will always lie in the @xmath24 plane for which @xmath25 . , dissipation is equal to the barrier height ( magnet relaxes from point 1 ( or 4 ) to point 2 ) . when the field is turned off fast , magnet relaxes from point 3 to point 4 or 1 depending on any infinitesimal bias again dissipating an amount equal to the barrier height.,title="fig:",width=302,height=226 ] + fig.[energy landscape with no dc field ] is plotted using equation [ energy ] with @xmath25 and @xmath26 which is the first case to be discussed .
the different contours correspond to different values of @xmath3 .
+ _ derivation of equation [ 2kv ] _ : let s start with equation [ 2kv ] which is the most important and also easiest .
dissipation occurs both during turn - on and turn - off of the pulse and the overall switching energy is sum of the two in general .
the dashed contour in fig.[energy landscape with no dc field ] corresponds to @xmath27 which is the minimum value needed to make @xmath28 ( point 2 ) the energy minimum . for a pulse with fast ( @xmath29 ) _ turn - on _
, dissipation can be calculated using equation [ energy ] as the difference between the initial and the final energies which are given by point 1 ( or 4 ) and point 2 on the dashed contour .
this value is @xmath30 for a pulse with fast ( @xmath29 ) _ turn - off _ , the energy contour immediately changes from the dashed one to the uppermost one in fig.[energy landscape with no dc field ] . under any infinitesimal bias ,
magnetization falls down the barrier to the left ( relaxing to point 1 ) or to the right ( relaxing to point 4 ) giving a dissipation of @xmath31 equal to the turn - on dissipation . the switching energy ( _ total dissipation
is sum of the values for turn - on and turn - off which gives us equation [ 2kv ] .
+ _ derivation of equation [ higher ] _ : this is the case with @xmath32 . the bottom most energy contour in fig.[energy landscape with no dc field ] shows such a situation as an example .
the minimum of energy is still at @xmath28 ( point 5 ) however now the energy well is deeper . for a pulse with fast ( @xmath29 ) _ turn - on _ , dissipation is the difference between the initial and final state energies @xmath33 ( where @xmath34 is used as a generic notation for the bottom of any well with @xmath35 ) . for a pulse with fast ( @xmath29 ) _ turn - off _
, the energy contour immediately changes from the bottom most curve to the uppermost curve in fig.[energy landscape with no dc field ] . depending on any infinitesimal bias magnet
will relax from point 3 to either point 1 or 4 dissipating the difference @xmath36 the switching energy is sum of the values for turn - on and turn - off which with straightforward algebra gives us equation [ higher ] .
+ _ derivation of equation [ lower ] _ : with @xmath37 , magnetization will not align along its hard axis ( @xmath28 ) .
this can be seen in fig.[energy landscape with no dc field ] where for a pulse lower than @xmath38 there are two minima of energy not located along the hard axis .
the logic device will not work in this regime because it needs to be close to its hard axis so that the field of another magnet can tilt it towards one minima deterministically .
nevertheless we derive dissipation for these pulses because we use the results in section
[ s51 ] to show switching energy in the adiabatic limit . for a pulse with fast ( @xmath29 )
_ turn - on _ , dissipation is the difference between the initial and final state energies @xmath39 for a pulse with fast ( @xmath29 ) _ turn - off _ , the energy contour suddenly becomes the uppermost one in fig.[energy landscape with no dc field ] . at that moment
magnetization is still at the same @xmath10 ( point 7 ) .
it follows down the barrier with the dissipation given by @xmath40 the _ total dissipation _ is sum of the values for turn - on and turn - off which gives us equation [ lower ] . in this section
we show that for @xmath27 , so long as @xmath23 switching energy can be calculated fairly accurately using equation [ 2kv ] considering only the effect of @xmath3 . for @xmath35
the effect of @xmath12 is even less pronounced as compared to @xmath3 and equation [ higher ] can be used to calculate dissipation .
again we are interested in initial and final state energies which can be calculated using equation [ energy ] with @xmath42 . in the @xmath43 direction for two values of the pulse : 0 and @xmath44 . upon _ turn - on _ , if magnetization starts from @xmath45 ( _ case 1 _ ) , it drops from point 1 ( @xmath46 ) to point 2 dissipating the difference .
if it starts from @xmath47 , it drops from point 4 ( @xmath48 ) to point 2 dissipating the difference . upon _ turn - off
_ , both cases 1 and 2 drop from point 3 to point 4 dissipating the difference.,title="fig:",width=302,height=226 ] + @xmath12 can be positive ( along @xmath11 ) or negative ( along @xmath43 ) . fig.[energy landscape dc ] shows the energy landscape with an @xmath12 in the @xmath43 direction .
if @xmath41 then the up and down states ( points 1 and 4 ) of the magnet have different initial energies which result in two different cases to be analyzed .
_ case 1 _ designates the situation where initial magnetization ( point 1 ) and @xmath12 are in the _ opposite _ direction .
_ case 2 _ designates the situation where initial magnetization ( point 4 ) and @xmath12 are in the _ same _ direction .
+ for a pulse with fast ( @xmath29 ) _ turn - on _ , _ case 1 _ dissipates the difference between points 1 and 2 and _ case 2 _ dissipates the difference between points 4 and 2 .
when the pulse is suddenly turned off , in both cases magnetization finds itself at point 3 , drops down to point 4 and dissipates the difference .
it is not possible to give an exact closed form expression for the value of dissipation with non - zero bias .
instead based on numerical calculations , we show figures that provide useful insight to conclude that for pulses with fast ramp time the effect of bias on switching energy is negligible .
+ the energy of point 2 ( and subsequently point 3 ) depicted in fig.[energy landscape dc ] changes as the relative magnitude of @xmath12 and @xmath38 are changed .
we like to know how dissipation changes as a function of the ratio @xmath49 .
the numerical results are plotted in fig.[dbi ] using equation [ energy ] .
+ fig.[dbi]a shows that for a pulse with fast _ turn - on _ and small values of @xmath49 , both cases dissipate about @xmath7 . as this ratio
is increased , the energy separation between points 1 and 2 ( see fig.[energy landscape dc ] ) increases and that of points 4 and 2 decreases which results in higher dissipation of _ case 1 _ and lower dissipation of _ case 2_. fig.[dbi]b shows the dissipation for a pulse with fast _ turn - off _ which is less than the barrier height @xmath7 and is expected because under the presence of @xmath12 , after turn - on , magnetization ends up closer to the final state ( see fig.[energy landscape dc ] ) as compared to the case where @xmath26 ( see fig.[energy landscape with no dc field ] ) .
the switching energy is sum of the dissipation values for turn - on and turn - off plotted in fig.[dbi]c .
for @xmath50 the bias field @xmath12 alone can switch the magnet and it is completely an unwanted situation @xcite .
note that for practical purposes , values of @xmath12 are small compared to @xmath38 ( for instance @xmath51 ) and the switching energy is more or less about @xmath52 which gives us equation [ 2kv ] . for @xmath53 the effect of bias
is even less pronounced and switching energy can be calculated using equation [ higher ] .
we have seen in section [ s3 ] that for pulses with fast ramp times , the effect of bias ( @xmath12 ) is negligible for @xmath51 and switching energy is obtained fairly accurately even if we set @xmath26 .
by contrast for pulses with slow ramp time , switching energy can be made arbitrarily small for @xmath26 and the actual switching energy is determined entirely by the @xmath12 that is used . in this section we will first show why the switching energy can be arbitrarily small for @xmath26 and then show that for @xmath41 it will saturate in _ case 1 _ but can be made arbitrarily small in _ _ case 2__@xcite .
two points are in order .
first , the analysis presented here is exact in the absence of noise .
if thermal noise is present the analysis may not be true in general and needs to be modified accordingly .
second , if in the process of switching , a bit of information is destroyed as in two inputs and one output gates ( e.g. and/or ) , then there will be a finite switching energy even for adiabatic switching .
@xmath55 gradual _ turn - on _ of the pulse corresponds to increasing the pulse in many small steps .
fig.[adiabatic progression]a shows the energy landscape . as the field is gradually turned - on the energy contours change little by little from top to bottom .
the minimum of energy gradually shifts from point 1 ( or 4 ) to point 2 .
magnetization hops from one minimum of energy to the other .
but why is it that gradual turn - on of the pulse dissipates less than sudden turn - on ? +
if the external pulse is turned on to @xmath38 in @xmath56 equal steps , we show that there is equal amount of dissipation at each step .
then total dissipation is @xmath56 times that of each step .
we show that dissipation of each step is proportional to @xmath57 ; hence as the number of steps increases , dissipation decreases as @xmath58 and in the limit of @xmath59 , @xmath60 ( this is not unlike a similar argument that has been given for charging up a capacitor adiabatically@xcite ) . at each step
when the pulse is increased by @xmath61 , the dissipated energy is the difference between initial and final state energies .
+ such a situation is illustrated in fig.[adiabatic progression]a where @xmath62 denotes a minimum on an energy contour corresponding to @xmath63 ( magnitude of the pulse after @xmath64 steps ) . when the pulse is stepped up to @xmath65 ,
magnetization suddenly finds itself at point @xmath66 ( initial state ) and falls down to @xmath67 ( final state ) .
note that dissipation is @xmath68 and not @xmath69 .
this is because when the field suddenly changes from @xmath63 to @xmath65 , magnet has not had time to relax and dissipate energy .
here we use @xmath70 and @xmath71 as generic notations for initial and final energy of any step .
@xmath72 can be found by finding the @xmath10 which corresponds to point @xmath62 ( the minimum of energy with @xmath73 ) and substituting it in equation [ energy ] with @xmath74 . with straightforward algebra
we get @xmath75 .
equation [ low field dissipation on ] can be used to calculate @xmath76 . using the identities @xmath77 and @xmath78 , the dissipated energy per step
is obtained as @xmath79 for gradual _ turn - off _ consider points @xmath67,@xmath80 and @xmath62 .
when @xmath74 , magnetization is at @xmath67 and after the pulse is decreased by one step to @xmath63 , it finds itself at @xmath80 , falls down to @xmath62 dissipating the difference @xmath81 .
@xmath82 can be found by finding the @xmath10 which corresponds to point @xmath67 ( the minimum of energy with @xmath74 ) and substituting it in equation [ energy ] with @xmath73 .
we get @xmath83 .
again equation [ low field dissipation on ] can be used to give @xmath84 . using the identities @xmath77 and @xmath85
, we obtain for the dissipated energy per step @xmath86 the switching energy is sum of the dissipation values for _ turn - on _ : @xmath87 and _ turn - off _ :
@xmath87 which in the limit of @xmath59 , tends to 0 ( @xmath60 ) . for _ turn - on _
let s consider _ case 1 _ first where initial magnetization and @xmath12 are in opposite directions ( point @xmath88 in fig.[adiabatic progression]b ) . as the field is gradually turned - on
, magnetization starts from point @xmath88 and hops from one minimum of energy to the next .
increasing the number of steps brings the minima closer to each other so that magnetization stays in its ground state while being switched .
however when magnetization gets to point @xmath89 , situation changes . at that point
the energy barrier which formerly separated the two minima on the two sides disappears .
magnetization falls down from point a to b and dissipates the energy difference .
this sudden change in the minimum of energy occurs no matter how slow the pulse is turned on and causes the switching energy to saturate so long as @xmath41 .
quantitatively this can be seen by plotting @xmath90 vs. @xmath3 ( fig.[adiabatic progression]c ) using equation [ energy ] . when the left solid curve is traced from @xmath91 ,
it is evident that there is a discontinuous jump in the @xmath90 values which minimize energy when the pulse is increases from @xmath92 to @xmath38 in infinitesimal steps .
this discontinuity goes away only when @xmath26 ( right solid curve ) . in _ case 2
_ , magnetization starts from point @xmath93 , i.e. @xmath94 ( see fig.[adiabatic progression]b and c ) , gets to point b at which there is _ no _ sudden change of minimum and as the pulse is increased further to @xmath38 , it gradually moves to point @xmath95 . during _ turn - off _ in both cases 1 and 2 , magnetization gradually moves from ( see fig.[adiabatic progression]c ) point @xmath95 to b and then finally to point @xmath93 all along staying in its minimum of energy with no discontinuity .
dissipation tends to zero as the pulse is turned off in infinitesimal steps .
+ is less than the barrier height @xmath7 .
the dashed line is plotted using equation [ adis].,title="fig:",width=302,height=226 ] + in the slow limit the entire dissipation is determined by the energy difference between points @xmath89 and @xmath96 , @xmath97 in fig.[adiabatic progression]b . for a given @xmath12 , one has to find that particular value of @xmath3 for which the local energy maximum in the middle disappears which means that the second derivative of energy with respect to @xmath10 must be zero ( no curvature ) .
since magnetization has been in the minimum of energy while getting to point @xmath89 , first derivative of energy with respect to @xmath10 must also be equal to zero . under these conditions
, the value of @xmath10 at @xmath89 and subsequently @xmath98 can be found using equation [ energy ] .
@xmath99 can be found as the true minimum of energy from equation [ energy ] where the first derivative of energy with respect to @xmath10 is zero but the second derivative is not .
what affects @xmath97 is the relative magnitude of @xmath12 and @xmath38 .
it is not possible to give an analytical closed form expression for this saturating value of dissipation . instead we ve
numerically plotted dissipation versus @xmath49 ( solid curve in fig.[adiabatic dissipation ] ) . for small values of @xmath49
, dissipation can be written as @xmath100 where the value of @xmath101 is obtained by an almost perfect fit to the solid curve for @xmath102 .
the dashed curve is plotted using equation [ adis ] . as is evident from fig.[adiabatic dissipation ] , this equation is fairly accurate .
there is some digression from the actual value of dissipation for large values of @xmath49 which are not of practical interest especially @xmath103 for which @xmath12 alone can switch the magnet and is completely an unwanted situation@xcite .
+ it is important to note that the switching energy in the adiabatic limit is case dependent . for _ case 1 _
, it is given by equation [ adis ] and it is not zero as it might have been expected for dissipation in the adiabatic limit . interestingly if @xmath101 was equal to 1 , the dissipation would be equal to the energy difference between initial and final states ( see points @xmath88 and @xmath93 in fig.[adiabatic progression]b ) .
however the actual value is significantly smaller .
+ dissipation in both the fast and slow limits can be casted into a single equation @xmath104 in the fast limit , @xmath105 is the magnitude of the pulse while in the slow limit , @xmath105 is related to the magnitude of the small bias field as states above .
@xmath7 is the height of the anisotropy energy barrier separating the two stable states of the magnet , and has to be large enough so that the magnet retains its state while computation is performed without thermal fluctuations being able to flip it .
the retention time for a given @xmath7 can be calculated using@xcite @xmath106 where @xmath107 is the attempt frequency with the range @xmath108 @xcite which depends in a nontrivial fashion on variables like anisotropy , magnetization and damping . +
thus far we ve shown switching energy in the two limiting cases of @xmath29 and @xmath109 . to understand how switching energy changes in between and also how fast it decreases we need to start from the llg equation which in the gilbert form reads : @xmath110 and in the standard form
reads : @xmath111 @xmath112 is the gyromagnetic ratio of electron and its magnitude is equal to @xmath113 in si and @xmath114 in cgs system of units .
@xmath115 is the phenomenological dimensionless gilbert damping constant .
@xmath116 is the magnetization . here
@xmath117 where @xmath118 . in general @xmath119 can be derived as the overall effective field : @xmath120 .
+ the following expressions are all equivalent statements of dissipated power @xcite : @xmath121 the dissipated power has to be integrated over time to give the total dissipation . in general ,
llg can be solved numerically using the method . to obtain generic results that are the same for various parameters
, we recast llg and the dissipation rate into a dimensionless form .
this will also show the significance of @xmath1 and demonstrate why for ramp times exceeding @xmath122 , there is a significant drop in dissipation .
+ using scaled variables @xmath123 and @xmath124 equation [ llg2 ] in dimensionless form can be written as + @xmath125 where @xmath126 with @xmath1 given by equation [ tau_c importance ] .
the energy dissipation normalized to @xmath7 can be written as @xmath127 to estimate the time constant involved in switching a magnet it is instructive to plot the integrand @xmath128 appearing above in equation [ dissipated energy ] assuming a step function for @xmath3 and obtaining the corresponding @xmath129 from equation [ llgn ] .
note that the integrands die out exponentially for a wide range of @xmath115 s from @xmath130 to @xmath131 .
in other words , all the curves ( ignoring the oscillations ) can be approximately described by @xmath132 thus suggesting that the approximate time constant is @xmath1 @xmath133 + this is more evident from fig.[ramp time ] where we show the energy dissipation for pulses with different ramp times .
the dissipated energy drops when @xmath0 exceeds @xmath1 as we might expect , but the drop is sharper than an rc circuit . needless to say ,
the dissipation values calculated from llg equation for the two limits of fast pulse @xmath8 and adiabatic pulse @xmath54 are consistent with the values calculated using energetics previously .
fig.[ramp time]a shows the _ turn - on _ dissipation where _ case 1 _ has saturated and _ case 2 _ goes down as ramp time is increased .
the curve in the middle is the case with infinitesimal bias @xmath26 and it is just provided for reference .
fig.[ramp time]b shows the _ turn - off _ dissipation where both cases 1 and 2 dissipate arbitrarily small amounts as the ramp time is increased . with slow pulses , overall switching energy of _ case
2 _ is very small and the entire switching energy of _ case 1 _ essentially occurs during _ turn - on _ which is illustrated in fig.[ramp time]c .
this dissipation was discussed in section [ s52 ] ; and it is associated with the sudden fall down from point a to b ( see fig.[adiabatic progression]b , c ) .
it has a saturating nature and will never become zero . as @xmath3 is applied more and more gradually ,
the dissipated power in fig.[ramp time]c becomes narrower and taller . in the true adiabatic limit
it will become a delta function occurring for one particular value of @xmath3 .
fig.[invchaindiss]a shows an array of spherical nanomagnets ( mqca ) that interact with each other via dipole - dipole coupling@xcite .
the objective is to determine the switching energy if we are to switch magnet 2 according to the state of magnet 1@xcite . in section [ s71 ] we will show a clocking scheme under which propagation of information can be achieved and basically shows how magnets can be used as _ cascadable logic _ building blocks . in section [ s72 ] , we briefly go over the method and equations used to simulate the dynamics and dissipation of the coupled magnets . in section [ s73 ]
we analyze the dissipation of the chain of inverters where we show that after cascading the magnetic bits , dissipation changes linearly with the number of magnets that the pulse is exerted on .
this shows that the switching energy of larger more complicated circuits can be calculated using the one - magnet results presented in this paper at least approximately .
+ + in the introduction we mentioned that in the clocking scheme the role of the clock field is to provide energy whereas field of another magnet acts as a guiding input . using a clock we can operate an array of exactly similar magnets as a chain of inverters .
fig.[invchaindiss]a shows a 3 phase inverter chain where the unit cell is composed of 3 magnets .
each magnet has two stable states showed as _ up _ and _ down _ in the figure .
we want to switch magnet 2 according to the state of magnet 1 .
first consider only magnets 1 and 2 .
we ve already explained ( see section [ s1 ] ) how magnet 1 can determine the final state of magnet 2 .
but what happens if more magnets are present ?
+ consider magnets 1 , 2 and 3 .
just like magnet 1 , magnet 3 also exerts a field on magnet 2 and if it is in the opposite direction can cancel out the field of magnet 1 . to overcome this
, we apply the pulse to magnet 3 as well thereby diminishing the exerted @xmath11 field of magnet 3 on magnet 2 so that magnet 1 becomes the sole decider of the final state of magnet 2 . in the process
the data in magnet 3 has been destroyed ( it will end up wherever magnet 4 decides ) .
it takes 3 pulses to transfer the bit ( in an inverted manner ) in magnet 1 to magnet 4 .
magnet 4 has been included because it affects the dissipation of magnet 3 through affecting its dynamics .
inclusion of more magnets to the right or left of the array will not change the quantitative or qualitative results of this paper .
next we ll briefly go over the method used to simulate the chain of inverters .
equations [ llgn ] ( with @xmath134 ) and [ dissipated energy ] are used to simulate the dynamics and dissipation of each magnet respectively .
the overall scaled ( divided by @xmath38 ) magnetic field @xmath135 of equation [ llgn ] for each magnet at each instant of time is modified to @xmath136 composed of the applied pulse : @xmath137 the anisotropy ( internal ) field of each magnet : @xmath138 and exerted dipolar fields of other magnets which in general in cgs system of units reads @xmath139 all field values are time dependent . here
@xmath140 denotes any one magnet and @xmath141 runs over magnetic moments of the other magnets . though this equation can be simplified for an array of magnets along the same line , in this form it can be used for more complicated arrangement of magnets .
fig.[invchaindiss]b shows the llg simulations of the chain of inverters where magnet 2 is switched solely according to the state of magnet 1 irrespective of its history or the state of magnets 3 and 4 .
fig.[invchaindiss]c shows dissipation of the entire array after one application of the pulse as a function of ramp time .
the pulse is exerted on magnets 2 and 3 which accounts for the @xmath142 value in the fast limit .
this essentially points out that after cascading these logic building blocks , dissipation changes linearly with the number of magnets .
+ in the slow limit , depending on the initial configuration , dissipation will be affected .
the 4 magnet array can initially be in any of its 16 possible states .
some configurations saturate and some do nt . here
the field of magnet 1 plays the role of the bias field @xmath12 for magnet 2 and the field of magnet 4 is like another bias field on magnet 3 which accounts for the 3 groups of curves in fig.[invchaindiss]c .
the upper curves correspond to the situation where initial magnetization of both magnets 2 and 3 are opposite to the fields exerted from magnets 1 and 4 respectively .
the middle curves correspond to only one of magnets 2 or 3 initially being opposite to the exerted fields of magnet 1 or 4 respectively .
the lower curves correspond to both magnets 1 and 3 initially being in the same direction as the exerted fields from magnets 2 and 4 respectively . + an added complication is the field of the other neighbor ( magnet 3 ) which is diminished in the @xmath11 direction but has a non - negligible @xmath143 component exerted on magnet 2 .
all this @xmath143 directed field does is to wash away a tiny bit the effect of the field of magnet 1 which has little bearing on the qualitative or quantitative results as illustrated in fig.[invchaindiss]c .
the speed of switching can be increased by increasing the magnitude of the external pulse @xmath3 above @xmath38 .
larger fields will dissipate more energy but have the advantage of aligning the magnet faster during the turn - on segment but are of no use for increasing the speed of the turn - off segment because the magnet relaxes to its stable state under its own internal field . if @xmath144 can be altered , then it is a better idea to increase @xmath38 and always set @xmath27 . this way
the speed of switching is increased by shortening the time of both turn - on and turn - off segments .
+ consider equation ( [ tau_c importance ] ) .
increasing @xmath115 shortens the switching time constant ( note that @xmath115 is usually less than 1 ) ; however this parameter is not very controllable in experiments .
@xmath145 is a physical constant and can not be altered .
so to increase the switching speed , one has to increase @xmath144 .
thermal stability of a magnet requires @xmath7 to be larger than a certain amount for the desired retention time .
for instance with an attempt frequency of about 1ghz ( see the discussion at the end of [ s5 ] ) and @xmath7 of about 0.5 ev , magnet is stable for about 0.5 seconds which is large enough because switching takes place in the nano - second scale .
a higher retention time requires higher @xmath7 .
once @xmath7 is set because of stability requirements , the only way to increase @xmath38 is to decrease @xmath146 .
assuming that volume is magnetized to saturation , @xmath147 is the magnetic moment of the magnet .
@xmath148 is the number of spins giving rise to the magnetization and @xmath149 is bohr magneton .
so decreasing @xmath146 translates to making the magnet smaller or decreasing its saturation magnetization .
+ the discussion just presented is similar to the theory of scaling in cmos technology where decreasing the capacitance causes an increase in the switching speed by decreasing the @xmath150 time constant . with the same operating voltage , smaller capacitance results in lower number of charges stored on the capacitor . in the case of cmos , as @xmath151 decreases , energy dissipated i.e. @xmath152 also decreases . in the case of magnet however , energy dissipation is fixed around @xmath52 so for a lower @xmath146 , dissipation of the ferro - magnetic logic element ( already very small ) is _ not _ altered ; however one might be able to reduce the dissipated energy in the external circuitry since it needs to provide the energy for a shorter period of time .
again we should emphasize that a thorough analysis of external dissipation also has to be done .
this has to do with generating the external source of energy for switching . in the case of mqca circuits
this is done by running currents through wires and generating magnetic fields . in principle , spin transfer torque phenomena or electrically controlled multi - ferroicity could also be used to provide the source of energy .
these methods would also have energy dissipation associated with them .
a complete lay - out circuit is necessary to properly evaluate the integration density of logic circuits made of magnets .
for example fringing fields and unwanted cross talks have to be taken into account .
external circuitry will take up space .
efficient methods have to be developed to porperly address these issues .
one component of the lay - out is the magnetic logic bit itself which we discuss here .
the barrier height @xmath153 between the stable states of a magnet can be engineered by adjusting @xmath16 ( anisotropy constant ) and @xmath6 ( volume ) . increasing the anisotropy constant is of great interest for the magnetic storage industry because it allows stable magnets of smaller volume that translates to higher densities .
many experiments report @xmath16 values on the order of a few @xmath154@xcite .
this results in stable magnets with volumes of only 10s of @xmath155 ; which means that stable magnets can be made as small as a few @xmath156 in each dimension . even though a complete lay - out is necessary , nevertheless these numbers are very promising and could potentially result in very high integration densities .
in this paper we analyzed the switching energy of single domain nanomagnets used as cascadable logic building blocks .
a magnetic pulse was used to provide the energy for switching and a bias field was used as an input to guide the switching .
the following conclusions can be drawn from this study .
+ ( 1 ) through analyzing the complete dependence of the switching energy on ramp time of the pulse , it was concluded that there is a significant and sharp drop in dissipation for ramp times that exceed a critical time given by equation [ tau_c importance ] whose significance is separating the energy dissipation characteristic of a fast pulse ( small ramp time ) and energy dissipation characteristic of a slow pulse ( big ramp time ) .
+ ( 2 ) the switching energy can be described by a single equation ( equation [ general ] ) in both fast and slow limits for trapezoidal pulses analyzed in this paper . in the fast limit
the effect of the bias field or equivalently the field of neighboring magnet in mqca systems is negligible so long as the bias field is less than 10th of the switching field of the magnet . in the slow limit
however , dissipation is largely determined by the value of the bias field .
+ ( 3 ) by evaluating switching energy of both one magnet and a chain of inverters for mqca systems , it was shown that the switching energy increases linearly with the number of magnets so that the one magnet results provided in this paper can be used to calculate the switching energy of larger more complicated circuits , at least approximately .
+ ( 4 ) practical issues such as dissipation versus speed , increasing the switching speed and scaling were discussed qualitatively .
it was concluded that by proper designing , ferromagnetic logic bits can have scaling laws similar to the cmos technology . + noise was not directly included in the models ; however we took it into account indirectly : thermal noise is the limiting factor on the anisotropy energy @xmath7 ( that determines the magnet s thermal stability ) of a magnet which we discussed thoroughly .
thermal noise also limits the lowest possible magnitude of the bias field ( or equivalently coupling between magnets in mqca systems ) .
we ve provided the results for a wide range of bias values .
more thorough discussions of dissipation in the external circuitry can be found in references @xcite,@xcite .
100 s. salahuddin and s. datta,``interacting systems for self correcting low power systems '' , appl .
lett . , vol.90 , pp.093503.1 - 093503.3 , feb .
cowburn and m.e .
welland,``room temperature magnetic quantum cellular automata '' , science , vol .
287 , pp.1466 - 1468 , feb . 2000 .
cowburn , a.o .
adeyeye and m.e .
welland,``controlling magnetic ordering in coupled nanomagnet arrays '' , new j. of phys . 1 , pp.16.1 - 16.9 , nov .
a.imre , g. csaba , l. ji , a. orlove , g. h. bernstein and w. porod,``majority logic gate for magnetic quantum - dot cellular automata'',science , vol .
311 , pp.205 - 208 , jan . 2006 .
a. ney , c. pampuch , r. koch and k.h .
ploog,``programmable computing with a single magnetoresistivevelement '' , nature , vol .
425 , pp.485 - 487 , oct . 2003 .
d. a. allwood , gang xiong , m. d. cooke , c. c. faulkner , d. atkinson , n. vernier and r. p. cowburn,``submicrometer ferromagnetic not gate and shift register '' , science , vol .
296 , pp.2003 - 2004 , jun . 2002 .
allwood , g. xiong , c. c. faulkner , d. atkinson , d. petit and r. p. cowburn , `` magnetic domain - wall logic '' , science , vol .
309 , pp.1688 - 1692 , sep . 2002 .
nikonov , g.i .
bourianoff , p.a .
gargini , `` simulation of highly idealized , atomic scale mqca logic circuits '' , http://arxiv.org/ , arxiv:0711.2246v1 [ cond-mat.mes-hall ] , nov . 2007 .
g. csaba , w. porod and a. i , csurgay,``a computing architecture composed of field - coupled single domain nanomagnets clocked by magnetic field '' , int . j. circ .
appl . , vol .
31 , pp.67 - 82 , jan . 2003 .
g. csaba , a. imre , g. h. bernstein , w. porad and v. metlushko,``nanocomputing by field - coupled nanomagnets '' , ieee trans . on nanotech . ,
vol.1 , pp.209 - 213 , dec .
g. csaba , p. lugli and w.porod,``power dissipation in nanomagnetic logic devices '' , 4th ieee conference on nanotechnology , pp.346 - 348 , aug
g. csaba , p. lugli , a. csurgay and w.porod,``simulation of power gain and gissipation in field - coupled nanomagnets '' , j. comp .
, vol . 4 , pp.105 - 110 , aug .
m. niemier , m. alam , x. s. hu , g. bernstein , w. porod , m. putney and j. deangelis , `` clocking structures and power analysis for nanomagnet - based logic devices '' , proceedings of the 2007 international symposium on low power electronics and design ( islped ) , new york , acm , 2007 .
nikonov , g.i .
bourianoff and p.a .
gargini , `` power dissipation in spintronic devices out of thermodynamic equilibrium '' , j. super . novel .
magn . , vol .
19 , no.6 , pp .
497 - 513 , aug . 2006 .
l. landau and e. lifshitz,``on the theory of the dispersion of magnetic permeability in ferromagnetic bodies '' , phys .
z. sowjetunion , vol .
8 , pp.153 - 169 , 1935 .
gilbert , `` a phenomenological theory of damping in ferromagnetic materials '' , ieee trans . magn .
40 , pp.3443 - 3449 , nov . 2004 .
b. hillebrands ( editor ) and k. ounadjela ( editor),``spin dynamics in confined magnetic structures i , ii and iii '' , new york , springer , 2001 - 2007 . c.h .
bennet,``the thermodynamics of computation - a review '' , intern .
905 - 940 , dec .
likharev and a.n .
korotkov,``single - electron parametron : reversible computation in a discrete - state system '' , science , vol .
763 - 765 , aug .
kummamuru , a.o .
orlov , r. ramasubramaniam , c.s .
lent , g.h .
bernstein , and g. snider , `` operation of a quantum - dot cellular automata ( qca ) shift register and analysis of errors '' , ieee tran . elec . devi . ,
vol.50 , pp.1906 - 1913 , sep .
r. street and j.c .
woolley , `` a study of magnetic viscosity '' , proc .
, sec . a , vol .
62 , pp.562 - 572 , sep . 1949
l. neel , `` thermoremanent magnetization of fine powders '' , rev .
293 - 295 , jan .
brown,``thermal fluctuations of a single - domain particle '' , phys .
130 , pp.1677 - 1686 , jun . 1963 .
gaunt,``the frequency constant for thermal activitation of a ferromagnetic domain wall '' , j. appl .
48 , pp.3470 - 3474 , aug .
the isotropic terms , like the demagnetizing field for a sphere or the weiss field have not been taken into account because they can be expressed as @xmath157 where @xmath116 is the magnetization and @xmath158 is an scalar .
based on the llg equation , fields of this form will _ not _ alter the dymanics of magnetization and have no bearing on dissipation . in principle magnet a can switch magnet b unidirectionally with no need for an external pulse given that : @xmath159 holds .
this entails designing circuits with magnets of different parameters ( e.g. volume ) so no two magnets in the circuit can have the same parameters ; not to mention the complexities caused by the fields exerted from other neighbors .
this essentially states that it is not a matter of principle that interacting magnets must have saturating dissipation as was previously suggested @xcite .
this is also shown by analyzing the dissipation of a chain of coupled magnets where for some configurations dissipation arbitrarily goes down whereas for some other configurations it saturates .
the concepts behind such effects can easily be traced back to the two cases ( 1 and 2 ) discussed in the context of one magnet and a bias field @xmath12 .
cavin iii , v.v .
zhirnov , j. a. hutchby , and g. i. bourianoff , `` energy barriers , demons , and minimum energy operation of electronic devices ( plenary paper ) '' , proceedings of spie , vol .
5844 , pp.1 - 9 , may 2005 .
sun and x.r .
wang , `` fast magnetization switching of stoner particles : a nonlinear dynamics picture '' , phys .
174430 - 1 to 174430 - 9 , may 2005 .
note that the the term @xmath160 should not have been included in equation 4 of reference 1 .
however this will not change the results of that paper since this term was assumed to be zero anyways .
shouheng sun , c. b. murray , d. weller , l. folks , and a. moser , `` monodisperse fept nanoparticles and ferromagnetic fept nanocrystal superlattices '' , science , vol .
287 , pp.1989 - 1992 , mar .
wu , c. liu , l.li , p. jones , r.w .
chantrell and d. weller , `` nonmagneti shell in surfactant - coated fept nanoparticles '' , j. appl .
95 , no.11 , pp.6810 - 6812 , jun . 2004 .
a. perumal , h.s .
ko and s.c .
shin , `` magnetic properties of carbon - doped fept nanogranular films '' , appl .
83 , no.16 , pp.3326 - 3328 , oct . 2003 .
k. elkins , d. li , n. poudyal , v. nandwana , z. jin , k. chen and j.p .
liu , `` monodisperse face - centered tetragonal fept nanoparticles with giant coercivity '' , j. phys .
38 , pp.2306 - 2309 , jul
jackson , `` classical electrodynamics '' , new york , wiley , 1999 , pp.198 - 200 . | power dissipation in switching devices is believed to be the single most important roadblock to the continued downscaling of electronic circuits .
there is a lot of experimental effort at this time to implement switching circuits based on magnets and it is important to establish power requirements for such circuits and their dependence on various parameters .
this paper analyzes switching energy which is dissipated in the switching process of single domain ferromagnets used as _
cascadable logic _ bits .
we obtain generic results that can be used for comparison with alternative technologies or guide the design of magnet based switching circuits .
two central results are established .
one is that the switching energy drops significantly if the ramp time of an external pulse exceeds a critical time .
this drop occurs more rapidly than what is normally expected of adiabatic switching for a capacitor .
the other result is that under the switching scheme that allows for logic operations , the switching energy can be described by a single equation in both fast and slow limits .
furthermore , these generic results are used to quantitatively examine the possible operation frequencies and integration densities of these logic bits which show that nanomagnets can have scaling laws similar to cmos technology . |
in this paper we consider a class of random cantor sets .
this consists of a sample space @xmath0 and a probability measure @xmath1 .
the sample space @xmath0 contains a family of compact subsets of @xmath2^{d}$ ] , furthermore @xmath0 is a compact metric space endowed with the hausdorff metric .
we will compute their almost sure and typical dimensions .
for each object of @xmath0 , we put a natural measure on this object .
we also calculate the local dimensions of these natural measures . in the end
, we study the hitting probabilities of a special subclass of these random cantor sets .
we start by a description of these random cantor sets .
closely related random models have been considered in @xcite .
let @xmath3 and @xmath4 be sequences of integers with @xmath5 for all @xmath6 .
let @xmath7 we denote by @xmath8^d)$ ] the family of @xmath9-adic closed subcubes of @xmath2^d$ ] , @xmath10 : 0\leq i_{\ell}\leq r_n^{-1}-1 \big\}\ ] ] and let @xmath11 we divide the unit cube @xmath2^d$ ] into @xmath12 interior disjoint @xmath13-adic closed subcubes and randomly choose interior disjoint @xmath14 of these closed subcubes in the following way .
we randomly choose a cube among @xmath12 cubes uniformly which means that every cube has the same probability of being chosen , then we randomly choose an other cube among the remaining @xmath15 cubes uniformly , and continue this process until we obtain @xmath16 cubes .
note that each of the closed subcubes has the same probability ( i.e.@xmath17 ) of being chosen , and denote their union by @xmath18 .
given @xmath19 , a random collection of @xmath20 interior disjoint @xmath9 - adic closed subcubes of @xmath2^{d}.$ ] for each cube of @xmath21 , we divide it into @xmath22 interior disjoint @xmath23-adic closed subcubes and randomly choose interior disjoint @xmath24 of these closed subcubes in the same fashion as above ( i.e. we randomly choose a cube among @xmath22 cubes uniformly , then we randomly choose an other cube among the remaining @xmath25 cubes uniformly , and continue this process until we obtain @xmath24 cubes ) .
we ask that the choices are independent for different cubes of @xmath21 .
let @xmath26 be the union of the chosen closed cubes and @xmath27 be a random limit set .
let @xmath28 be our probability space which consists of all the possible outcomes of random limit sets .
for convenience we will write @xmath29 or @xmath30 in the following . our main object of study in this paper is the space @xmath0 .
figure [ figure1 ] shows an example of this construction .
let @xmath31^{d})$ ] be all the compact subsets of unit cube @xmath2^{d}$ ] .
we endow @xmath32 with the hausdorff metric .
recall that the hausdorff distance of two compact sets @xmath33 and @xmath34 of @xmath32 is defined by @xmath35 where @xmath36 .
observe that @xmath37 and @xmath0 is a closed subset of @xmath38 together with the well known fact that @xmath32 is a compact space , we obtain that @xmath0 is compact subset of @xmath32 .
recall that a subset of a metric space @xmath39 is of first category if it is a countable union of _ nowhere dense _ sets ( i.e. whose closure in @xmath39 has empty interior ) ; otherwise it is called of second category .
we say that a _ typical _
element @xmath40 has property @xmath41 , if the complement of @xmath42 is of first category . for the basic properties and various applications of baire category , we refer to @xcite .
let @xmath43^d$ ] . for any @xmath44 ,
the @xmath45-dimensional hausdorff measure is defined as @xmath46 where @xmath47 and @xmath48 is the diameter of @xmath49 .
the hausdorff dimension of @xmath33 is @xmath50 for any @xmath51 , let @xmath52 be the smallest number of sets of diameter at most @xmath53 which can cover @xmath33 .
then the lower and upper box dimensions are defined respectively as @xmath54 if @xmath55 we denote this common value by @xmath56 and call it the box dimension of @xmath33 .
the packing dimension of @xmath33 is defined as @xmath57 the assouad dimension of @xmath33 is defined as @xmath58 the basic relationships of these dimensions are @xmath59 for more details and further properties of these dimensions , we refer to @xcite and especially @xcite for the assouad dimension .
let @xmath60 be a radon measure on @xmath61 . for @xmath62 ,
the lower and upper local ( pointwise ) dimensions of @xmath60 at @xmath63 are defined respectively as @xmath64 if @xmath65 we denote this common value by @xmath66 , and call it the local dimension of @xmath60 at @xmath63 . for
further details and basic properties on the local dimensions of measures , see ( * ? ?
* chapter 10 ) .
we consider the natural random measure on the random cantor set .
let @xmath67 be a realization .
for each @xmath68 , let ( @xmath69 are from ) @xmath70 and @xmath71 where @xmath72 is the indicator function of the set @xmath34 .
note that for every @xmath73 ( we will denote this by @xmath74 in the following for convenience ) , we have @xmath75 .
it is clear that @xmath76 weakly convergence to a measure @xmath77 , see ( * ? ? ?
* chapter 1 ) .
we call this measure @xmath77 the natural measure on @xmath33
. there exists a huge literature on computing the ` almost sure ' dimensions for many other random fractal sets .
we refer to @xcite and reference therein . for the general estimations and the almost sure dimensions of these random cantor sets , we have the following result .
let @xmath78 denote @xmath79 and @xmath80 furthermore let @xmath81 note that if the @xmath82 are bounded then @xmath83 and @xmath84 figure [ figure2 ] ` explains ' why there is @xmath85 in the definition of @xmath86 .
figure [ figure3 ] ` explains ' why there is @xmath87 in the definition of @xmath88 .
subcubes ( dark cubes ) of @xmath89 ( @xmath74 ) which belongs to @xmath26 , and they uniformly distributed inside the cube @xmath89 .
thus there are nearly @xmath24 subcubes of @xmath89 with side length @xmath90 ( depends if @xmath91 is an integer or not ) which intersect @xmath33 . in the end
, we have @xmath92 interior disjoint cubes with side length @xmath90 which intersect @xmath33 .
this come out the definition of @xmath86 . ]
[ thm : dimension ] ( 1 ) for any @xmath93 , we have @xmath94 \(2 ) for any @xmath93 , we have @xmath95 \(3 ) the almost sure hausdorff dimension and lower box dimension are maximal , i , e . , almost surely @xmath96
\(4 ) the almost sure packing dimension and upper box dimension are maximal , i , e . , almost surely @xmath97 \(5 ) for any @xmath93 , we have @xmath98 provided @xmath99 is bounded .
otherwise , almost surely @xmath100 .
we can also regard the space @xmath0 as a subclass of moran sets .
the dimensional properties of moran sets have been studied extensively , we refer to @xcite and reference therein .
the results of theorem [ thm : dimension ] are similar to the dimensional results of one dimensional homogeneous cantor sets ( uniform cantor sets ) .
an interesting fact is that they have the ` same ' dimensional formulas ( for our case @xmath101 ) . for hausdorff , lower box , upper box , and packing dimensions of one dimensional homogeneous cantor sets ,
see @xcite . for assouad dimension of one dimensional homogeneous cantor sets ,
see @xcite .
the figure [ figure3 ] corresponds to the partial homogeneous cantor sets of @xcite .
the above statements ( 1 ) and ( 2 ) generalize the results of @xcite from one dimensional moran sets to our model , and the statement ( 5 ) when @xmath102 are bounded generalize the result of @xcite from homogeneous cantor sets to our model .
the proof of @xmath103 is adapted from ( * ? ? ?
* theorem 2.1 ) to our setting , while the method for the proof of @xmath104 is different from that of @xcite .
the proof of the statement ( 5 ) when @xmath102 are bounded generalize the method in @xcite to high dimension .
our main contribution of theorem [ thm : dimension ] is to determine the almost sure dimensions of these random cantor sets for the case when @xmath105 is unbounded .
our method combines geometric and probability estimates on the distribution of these random cantor sets .
recall that @xmath106 is a compact metric space . for the typical dimensions of these random cantor sets , we have the following result .
for some related results we refer to @xcite [ thm : typical ] ( 1 ) the typical hausdorff dimension and lower box dimension are minimal , i.e. , for a typical @xmath107 we have @xmath108 \(2 ) the typical packing dimension and upper box dimension are maximal , i.e. , for a typical @xmath93 , we have @xmath109 \(3 ) if @xmath99 is unbounded , then for a typical @xmath93 , we have @xmath110 note that the typical hausdorff dimension and lower box dimension are as small as possible , but the almost sure hausdorff dimension and lower box dimension are as large as possible .
furthermore the packing dimension , upper box dimension and assouad dimension are as large as possible in the sense of both almost sure dimension and typical dimension . for the local dimensions of the natural measures supported on these random cantor sets , we have the following result .
let @xmath111 [ thm : localdimension ] ( 1 ) for any @xmath112 , we have @xmath113 \(2 ) for any @xmath112 , we have @xmath114 \(3 ) for @xmath1-almost all @xmath93 , and @xmath77 almost every @xmath115 , we have @xmath116 \(4 ) for @xmath1-almost all @xmath93 , and @xmath77 almost every @xmath115 , we have @xmath117 same kind of results have been obtained for other `` random '' measures , we refer to @xcite and reference therein .
for the local dimensions of the moran measures , we refer to @xcite .
[ rem : ttt ] the dimension of a set has essential connection with the local dimension of the measure on it , we refer to ( * ? ? ? * proposition 2.3 - 2.4 ) for more details .
in fact there are some overlaps between our theorem [ thm : dimension ] and theorem [ thm : localdimension ] .
actually theorem [ thm : localdimension ] ( 3)-(4 ) combined with the propostion 2.3 of @xcite and theorem [ thm : dimension ] ( 1)-(2 ) implies theorem [ thm : localdimension ] ( 3)-(4 ) .
we present more details in the following .
theorem [ thm : localdimension ] ( 3 ) and ( * ? ? ?
* proposition 2.3 ( a ) ) implies that almost surely @xmath118 . combining this with theorem [ thm : dimension ] ( 1 ) which gives @xmath119 for any set @xmath93
, we obtain theorem [ thm : dimension ] ( 3 ) .
theorem [ thm : localdimension ] ( 4 ) and ( * ? ? ? * proposition 2.3 ( c ) ) implies that almost surely @xmath120 . combining this with theorem [ thm : dimension ] ( 2 ) which gives @xmath121 for any set @xmath93
, we obtain theorem [ thm : dimension ] ( 4 )
. since our methods for theorem [ thm : dimension ] ( 3)-(4 ) and theorem [ thm : localdimension ] ( 3)-(4 ) are different , and the methods are interesting on it s own , we present them separately . for the hitting probability of these random cantor sets ,
we consider the special case that @xmath122 and @xmath123 for all @xmath124 , and we have the following result .
note that the result is similar to the hitting probability of fractal percolation ( see ( * ? ? ?
* theorem 9.5 ) ) and random covering sets ( see @xcite ) . [ thm : hitting ] let @xmath34 be a borel subset of @xmath2^d$ ] with @xmath125 and @xmath126
. then we have \(1 ) if @xmath127 then almost surely @xmath128 is empty .
\(2 ) if @xmath129 then @xmath33 intersects @xmath34 with positive probability .
\(3 ) if @xmath129 then @xmath130 , where the norm is an essential supremum in the underlying probability space . note that under the condition @xmath131 , it will become much easier to prove theorem [ thm : dimension ] , theorem [ thm : typical ] , and theorem[thm : localdimension ] .
our main contribution of this project is to deal with the case when @xmath132 is unbounded .
the paper is organised as follows . in section [ section : lemma ]
we will show some lemmas for later use .
theorems [ thm : dimension ] , [ thm : typical ] , [ thm : localdimension ] , and [ thm : hitting ] are proved in sections [ section : dimension ] , [ section : typical ] , [ section : localdimension ] , and [ section : hitting ] respectively .
we conclude with additional results and open problems in section [ section : remarks ] .
we show some useful lemmas in this section .
* proposition 2.3)[lem : lp ] let @xmath133 be a borel set and let @xmath77 be a finite measure . if @xmath134 for all @xmath135 and @xmath136 then @xmath137 ( * ? ? ?
* corollary 3.9)[lem : bp ] let @xmath138 be a compact and such that @xmath139 for all open sets @xmath140 that intersects @xmath33
. then @xmath141 .
for convenience we put an easy fact about assouad dimension as the following lemma .
for further basic facts on assouad dimension , we refer to @xcite .
[ lem : wen ] let @xmath138 .
if there are sequences @xmath142 and @xmath143 of positive real numbers with @xmath144 as @xmath145 , such that for every @xmath146 there exists @xmath115 with @xmath147 then we have @xmath148 . the following estimate will be used in the proof of lemma [ cor : intersection ] .
[ lem : erdos ] let @xmath149 .
now we randomly choose @xmath150 numbers from @xmath151 in the same way as our construction of random cantor sets ( we randomly choose a number from @xmath151 uniformly , then we randomly choose an other number among the remaining @xmath152 numbers uniformly , and continue this process until we obtain @xmath146numbers ) .
let @xmath153 be the random chosen @xmath146 numbers , then @xmath154 note that the random set @xmath153 will intersect ( hit ) the set @xmath155 with probability one when @xmath156 in the following , we assume that @xmath157 .
let @xmath158 where @xmath159 means the @xmath160-th chosen number .
let @xmath161 be the event @xmath162 , then @xmath163 by the fact that the event @xmath164 is the complement of the event @xmath165 , we complete the proof .
the following estimate will be used in the proof of lemma [ lem : box ] . for
more details on large deviations estimates , see ( * ? ? ?
* appendix a ) .
[ lem : law of large numbers ] let @xmath166 be a sequence nonegative independent random variables with @xmath167 and @xmath168 for all @xmath169 .
then @xmath170 let @xmath171 .
we apply markov s inequality to the random variable @xmath172 .
this gives @xmath173 the last equality holds since @xmath174 is a sequence independent random variables . for any @xmath175
$ ] we have @xmath176 since @xmath177 $ ] for all @xmath169 , we have that for all @xmath178 @xmath179 and hence @xmath180 combining this with , we finish the proof .
for any @xmath93 and @xmath124 , we have @xmath181 , and hence @xmath182 for convenience , let @xmath183
. suppose @xmath184
( @xmath185 is the trivial case ) . for any @xmath186 , by the definition of @xmath88
, there exist @xmath187 such that for any @xmath188 , @xmath189 let @xmath190 and @xmath77 be the natural measure on @xmath33 .
we intend to show that @xmath191 for any ball @xmath192 with @xmath193 where @xmath194 is a constant . for @xmath195
, there exists @xmath6 such that @xmath196 .
case 1 . @xmath197 . in this case , the ball @xmath192 intersects at most @xmath198 cubes of @xmath199 , hence @xmath200 case 2 . @xmath201 . in this case , observe that there exists a constant @xmath202 such that any ball @xmath192 can intersects at most @xmath203 cubes of @xmath204 , and hence @xmath205 thus the mass distribution principle ( * ? ? ?
* chapter 4 ) implies that @xmath206 .
since this holds for any @xmath207 , we obtain that @xmath103 . for each @xmath124 , let @xmath208 . for any @xmath51 , there exists @xmath6 such that @xmath209 case 1 .
@xmath210 . in this case
we have @xmath211 , and hence @xmath212 case 2 . @xmath213 . in this case
, we have @xmath214 .
thus @xmath215 taking the upper limit of and , we obtain that @xmath216 suppose @xmath217 . for any @xmath218 there
exists a sequence of numbers @xmath219 with @xmath220 as @xmath221 , such that @xmath222 for all @xmath223 let @xmath224 then we have @xmath225 and hence @xmath226 .
since this holds for all @xmath115 , together with lemma [ lem : lp ] we have that @xmath227 . by the arbitrary choice of @xmath218
, we obtain that @xmath228 .
thus we complete the proof .
let @xmath229 be the boundary of @xmath89 , define @xmath230 it is clear that @xmath231 has zero lebesgue measure and for any @xmath232^d\backslash \widetilde{b}$ ] , @xmath233 recall that @xmath234 .
for the purpose of estimating the lower bound for hausdorff dimension , we need the following estimate . [
lem : twopoint ] for any @xmath235 there exists positive constant @xmath236 , such that @xmath237 for all @xmath238 and @xmath239^d$ ] . for any @xmath235 , by the definition of @xmath240
, there exists @xmath241 , such that @xmath242 for all @xmath243 which implies that @xmath244 we first assume that @xmath239^d\backslash \widetilde{b}$ ] ( to make sure that for any @xmath68 there exists unique @xmath245 with @xmath246 and @xmath247 )
. there is @xmath124 such that @xmath248 it follows that there exists two distinct cubes @xmath249 such that @xmath250 and @xmath251 .
therefore , for any @xmath252 we have @xmath253 now we turn to the estimate of [ eq : twopoint ] .
case 1 . @xmath254 . in this case
we have @xmath255 case 2 . @xmath256 . in this case
there will appear three subcases depending on @xmath257 .
subcase 1 . @xmath258 . in this case
, @xmath259 combining this with the estimate we obtain @xmath260 subcase 2 . @xmath261 . applying the estimate we have @xmath262 subcase 3 .
@xmath263 . applying the estimates and ,
we have @xmath264 let @xmath265 , then the estimate holds for all @xmath239^d\backslash \widetilde{b}$ ] .
note that for every point @xmath63 and @xmath68 there exist at most @xmath266 cubes of @xmath267 such that each of these cube contains @xmath63 .
it follows that for any @xmath239^d$ ] @xmath268 where @xmath269 are two cubes of @xmath267 which contain @xmath270 and @xmath271 separately .
thus there exists a larger constant such that the estimate holds for all @xmath239^d$ ] . for the convenience ,
we denote this larger constant also by @xmath194 . by theorem [ thm : dimension ]
( 1 ) , it is sufficient to prove that almost surely @xmath118 . for any @xmath272 there
exists a positive constant @xmath194 , such that lemma [ lem : twopoint ] holds .
applying lemma [ lem : twopoint ] , fatou s lemma , and fubini s theorem , we obtain @xmath273^d } \int_{[0,1]^d } d(x , y)^{-s_1 + 2\varepsilon}d(x , y)^{s_1-d-\varepsilon}dxdy
< \infty . \end{aligned}\ ] ] this implies that a.s .
@xmath274 thus by applying the energy argument ( * ? ? ?
* theorem 4.13 ) , we have that almost surely @xmath275 . by the arbitrary choice of @xmath276
, we obtain that almost surely @xmath277 note that the estimate implies that @xmath278 for @xmath77 almost all @xmath115 , for a proof see the argument in ( * ? ? ?
* theorem 4.13 ) .
together with the estimate and the arbitrary choice of @xmath276 we obtain that for @xmath1-almost all @xmath93 , and @xmath77 almost every @xmath115 , we have @xmath279 .
as we claimed before in remark [ rem : ttt ] , we will present a different proof in section [ section : localdimension ] . for every @xmath280
, we define the random set @xmath281 recall that @xmath282 means that @xmath283 and @xmath284 . in the following we are going to show that the set @xmath285 is fairly uniformly distributed ( this motivated the formula of the upper box dimension ) .
let @xmath286 where @xmath287 denotes the integer part of @xmath63 .
for every @xmath288 , we divide it into @xmath289 interior disjoint closed subcubes with side length @xmath290 and denote by @xmath291 the collection of these subcubes . for every @xmath292 ,
define @xmath293 by a volume argument , we have @xmath294 where @xmath295 denotes the cardinality of a set @xmath296 . and random variable @xmath297 figure [ figure4 ] shows the relative position of the above geometric objects .
[ cor : intersection ] let @xmath288 , then for every @xmath298 we have @xmath299 and hence @xmath300 applying lemma [ lem : erdos ] for @xmath301 @xmath302 and the estimate , we obtain @xmath303 it follows that @xmath304 thus we complete the proof .
the following proposition contains the second statement of theorem [ thm : dimension ] @xmath305 .
[ lem : box ] almost surely @xmath306 if @xmath99 is bounded then theorem [ thm : dimension ] ( 2 ) implies that @xmath307 for all @xmath93 .
furthermore [ thm : dimension ] ( 2 ) clams that @xmath121 for any @xmath93 .
thus it is sufficient to prove that almost surely @xmath308 for the case that @xmath99 is unbounded .
suppose @xmath309 . by the definition of @xmath86
( see ) , for any @xmath310 there exist a sequence @xmath311 such that @xmath312 observe that if @xmath313 is bounded , then @xmath314 for any @xmath190 .
thus we suppose that @xmath315 as @xmath316 , and @xmath317 .
it follows that for all @xmath318 , @xmath319 for each @xmath320 , by corollary [ cor : intersection ] we have @xmath321 furthermore , conditional on @xmath322 , we have that @xmath323 and @xmath324 are independent for any two distinct cubes @xmath325 . thus applying lemma [ lem : law of large numbers ]
, we obtain that @xmath326 recall that @xmath327 by elementary geometry , there exists a positive constant @xmath202 such that @xmath328 for each @xmath124 , define the event @xmath329 combining this with the estimate , we have @xmath330 it follows that @xmath331 since @xmath332 as @xmath333 , thus there exists @xmath334 such that @xmath335 for all @xmath188 , and hence @xmath336 applying the borel - cantelli lemma , we obtain @xmath337 where @xmath338 means the complement of @xmath339 .
thus we obtain that for almost any @xmath340 , there exists @xmath341 , such that @xmath342 for all @xmath343 it follows that for every @xmath344 , we have @xmath345 since @xmath346 and @xmath347 , we have @xmath348 holds for all @xmath344 , and hence by the estimate we have @xmath349 . by the arbitrary choice of @xmath276
we finish the proof .
now we intend to show that almost surely @xmath350 .
recall that @xmath351 .
let @xmath352 be a dense subset of @xmath2^d$ ] , and @xmath353 be an open ball .
for every @xmath68 , by the homogeneous structure of our random cantor sets , we obtain that almost surely on @xmath354 @xmath355 it follows that almost surely for any @xmath356 ( here the order of ` almost surely ' and ` for every @xmath68 ' is different from above ) with @xmath357 , @xmath358 observe that for any @xmath93 and any open set @xmath49 with @xmath359 , there there is a ball @xmath356 for some @xmath68 such that @xmath360 hence for almost all @xmath93 and any open set @xmath361 @xmath362 applying lemma [ lem : bp ] , we obtain that almost surely @xmath363 by the fact that @xmath364 and the proposition [ lem : box ] , we complete the proof .
assume first that @xmath99 is bounded .
let @xmath365 . by the definition of @xmath366 we have that for any @xmath367 there exists @xmath187 such that for any @xmath188 , @xmath368 let @xmath93 , for any @xmath369 , there exist @xmath370 , such that @xmath371 case 1 .
@xmath372 . for any @xmath115 we have @xmath373 case 2 .
@xmath374 . for any @xmath115 , by estimates
and we obtain @xmath375 thus we have @xmath376 . by the arbitrary choice of @xmath276 we obtain @xmath377 .
for the lower bound . for any @xmath378 ,
there exists @xmath379 as @xmath380 , such that for every @xmath160 @xmath381 and so there exists @xmath382 such that @xmath383 it follows that @xmath384 where @xmath202 is a positive constant .
applying lemma [ lem : wen ] and the estimate @xmath385 we obtain that @xmath386 . by the arbitrary choice of @xmath276 we have @xmath387 .
thus we complete the proof in the case when @xmath99 is bounded .
now suppose @xmath99 is unbounded .
since @xmath388 holds for any @xmath389^d$ ] , it is sufficient to show that almost surely @xmath390 .
let @xmath391 for every @xmath392 , define the event @xmath393 recall the random variable @xmath323 defined in .
thus @xmath394 combining this with corollary [ cor : intersection ] , we have @xmath395 for every @xmath124 , define the event @xmath396 conditional on @xmath322 , recall that the cubes form @xmath397 are chosen independently inside each cube of @xmath322 .
thus the random variables @xmath323 and @xmath324 are independent for any two distinct cubes @xmath398 of @xmath322 .
together with the estimate we have
@xmath399 it follows that for every @xmath124 , @xmath400 thus for any @xmath401 we have @xmath402 , and hence @xmath403 it follows that for almost all @xmath340 , there exists @xmath404 , such that @xmath405 for all @xmath223 combining this with lemma [ lem : wen ] , we obtain that almost surely @xmath406 .
thus we complete the proof .
for each cube @xmath89 , let @xmath407 be the nearest point of @xmath89 to zero vector . for each @xmath68 let @xmath408 theorem [ thm : dimension ] ( 1 ) claims that any element of @xmath93 has @xmath409 . in the following we intend to show that a typical set @xmath93 has @xmath410 .
for each @xmath68 , let @xmath411 .
for each @xmath412 , we choose an object @xmath413 with @xmath414 let @xmath415 be the collection of these @xmath416 .
observe that for any infinite set @xmath417 , the set @xmath418 is a countable dense subset of @xmath0 . by the definition of @xmath88
there is a subsequence @xmath419 with @xmath420 as @xmath333 such that @xmath421 let @xmath422 and @xmath423 where @xmath424 is an open set of @xmath106 with center @xmath425 and radius @xmath426 .
since @xmath427 is a countable dense subset in @xmath428 the set @xmath429 is a dense open set in @xmath0 .
it follows that the complement of @xmath430 is of first category .
let @xmath431 , then there is subsequence @xmath432 with @xmath433 as @xmath333 and @xmath434 such that @xmath435 observe that @xmath436 combining this with the definition of @xmath437 and the formula , we obtain @xmath438 thus we complete the proof .
[ re : lowlocal ] from the construction of @xmath439 , it follows that for any set @xmath440 , there exists a constant @xmath441 , such that for any @xmath442 , @xmath443 and hence @xmath444 .
theorem [ thm : dimension ] ( 2 ) claims that any element @xmath93 has @xmath445 . in the following we intend to show that a typical set @xmath93 has @xmath120 . for each @xmath68
, recall that @xmath446 . for each @xmath412
we intend to choose a set @xmath447 depending on the relative size of @xmath448 and @xmath449 .
case 1 . @xmath450 . in this case for each @xmath21 we choose a set @xmath451 with @xmath452 case 2 . @xmath453 . for each @xmath21
we choose a set @xmath454 with @xmath455 the notation @xmath291 is given at the beginning of subsection [ ll ] . in this case , we may think @xmath451 as those @xmath26 which the cubes of @xmath26 is well separated .
let @xmath415 be the collection of these @xmath456 observe that for any infinite set @xmath417 , the set @xmath418 is a countable dense subset of @xmath0 . by the definition of @xmath86
there is a subsequence @xmath419 with @xmath420 as @xmath333 such that @xmath457 let @xmath422 and @xmath458 applying the same argument as in the proof of theorem [ thm : typical ] ( 1 ) , we obtain that the complement of @xmath430 is of first category .
let @xmath431 , then there is subsequence @xmath432 with @xmath433 as @xmath333 and @xmath434 such that @xmath459 let @xmath77 be the natural measure on @xmath33 .
we are going to present that there exists a positive constant @xmath202 such that for any @xmath460 , @xmath461 for the above case 1 , we have @xmath462 now we turn to the case 2 .
since for any @xmath460 there exists at most one cube of @xmath463 intersects @xmath464 , we have @xmath465 thus we obtain the estimate . together with the formula , we have @xmath466
since this holds for any @xmath431 and @xmath467 by lemma [ lem : lp ] we obtain that any @xmath431 has @xmath120 .
thus we complete the proof .
note that the above proof also implies that a typical @xmath93 has full assouad dimension .
we show an outline for the proof .
assume @xmath468 with @xmath469 as @xmath333 .
let @xmath470 be the set in .
then the structure of @xmath471 and lemma [ lem : wen ] imply that any element of @xmath470 has full assouad dimension . thus we complete the proof .
for every @xmath135 and @xmath124 , we have @xmath472 and hence @xmath473 on the other hand , it follows immediately from the proof of theorem [ thm : dimension ] ( 1 ) that @xmath474 thus we complete the proof . for any @xmath475
, there exists @xmath6 such that @xmath476 .
observe that @xmath477 and @xmath478 therefore @xmath479 on the other hand , for any @xmath124 , @xmath480 and hence @xmath481 thus we complete the proof .
we start from the following lemma .
[ lem : massdistributionr ] for any @xmath482 , there exists a positive constant @xmath194 such that for any fixed @xmath232^{d}$ ] , @xmath483 furthermore we have @xmath484 for @xmath485 , by the definition of @xmath240 , there exists @xmath486 such that for all @xmath487 , @xmath488 . for @xmath489
, there exists @xmath6 such that @xmath490 .
case 1 . @xmath491 . in this case
we have @xmath492 the last inequality holds by @xmath493 and @xmath489 .
. there will appear three subcases depending on the size of @xmath494 .
subcase 1 . @xmath495 . in this case
, we have @xmath496 subcase 2 . @xmath497 .
in this case we have @xmath498 since this holds for any @xmath68 , we have @xmath499 subcase 3 . @xmath500 .
let @xmath501 be the collection of cubes of @xmath502 which intersects @xmath192 .
by a volume argument there exists a positive constant @xmath503 such that @xmath504 note that for @xmath505 and @xmath506 we have @xmath507 and hence @xmath508 combining these with @xmath509 , we have @xmath510 we fix a large constant @xmath194 such that all the above estimates hold .
thus we obtain the estimate . note
that for any open set @xmath511^{d}\times [ 0,1]^{d}$ ] , we have ( see ( * ? ? ?
* chapter 1 ) ) @xmath512 it follows that ( let @xmath192 be an open ball ) @xmath513 applying fatou s lemma and , we have @xmath514^d } p_n^{-1}{\mathbb{e}}\left(\mu_n \left(b \left(x , r \right ) \right ) \textbf{1}_{e_n}(x ) \right)dx \\ & = \liminf_{n\rightarrow \infty } \int_{[0,1]^d } { \mathbb{e}}\left(\mu_n \left(b \left(x , r \right ) \right ) \big\ ?
x \in e_n \right)dx\\ & \leq c r^s .
\end{aligned}\ ] ] thus we finish the proof . for the lower bound , let @xmath515 with @xmath516 .
note that for this @xmath45 , by lemma [ lem : massdistributionr ] there is constant @xmath194 such that the estimate holds .
let @xmath517 for @xmath518 .
then @xmath519 thus we obtain that a.s .
@xmath520 and hence for @xmath77-a.e .
@xmath63 @xmath521 combining this with our choice @xmath522 , we obtain @xmath523 since this holds for any @xmath524 we have a.s .
@xmath525 for @xmath77-a.e .
thus we finish the proof .
let @xmath526 . applying the similar arguments to lemma [ lem : massdistributionr ]
, we have the following result .
[ lem : massdistributiondelta ] for any @xmath527 , there exists @xmath194 and a subsequence @xmath528 , such that @xmath529 furthermore we have @xmath530 for any @xmath531 , there exists a subsequence @xmath528 such that @xmath532 for all @xmath518 . for each @xmath518 , let @xmath533 be the @xmath494 in the proof of lemma [ lem : massdistributionr ] . by the choice of @xmath534
, it is sufficient to consider subcase 2 and subcase 3 in the proof of lemma [ lem : massdistributionr ] .
moreover we use the estimate @xmath535 at the estimates and .
thus we complete the proof .
lemma [ lem : lp ] and theorem [ thm : dimension ] ( 2 ) imply that for any @xmath190 , @xmath536 holds for @xmath77-almost every @xmath115 . for the lower bound .
suppose @xmath309 .
let @xmath515 with @xmath537 .
applying lemma [ lem : massdistributiondelta ] and the same argument as in the estimate , we obtain @xmath538 by the same argument as in the proof of theorem [ thm : localdimension ] ( 3 ) , we complete the proof .
in this section , we study the hitting probabilities of random cantor sets in @xmath539 .
note that the hausdorff dimension of any @xmath93 is @xmath540 . the methods which we use in the following proof
are mainly from ( * ? ? ?
* chapter 8) , @xcite ( first - moment and second - moment methods ) and @xcite .
before we give the proof , we first show the following heuristic calculation . for @xmath541^{d},$ ]
define @xmath542 suppose @xmath543 roughly equals @xmath544 .
we simply denote it as @xmath545 .
observe that @xmath546 therefore theorem [ thm : hitting ] should follows from the relationships between @xmath547 and @xmath548 .
recall that @xmath549 and @xmath550 . applying the equivalent definition of hausdorff dimension ( ( * ? ? ?
* chapter 2.4 ) ) , we have that for any @xmath367 , there exists a sequence of interior disjoint cubes @xmath551 , such that @xmath552 and ( see ) @xmath553 recall that @xmath554 is the diameter of @xmath89 . for any @xmath555 , we have @xmath556 here we used the condition @xmath557 .
observe that @xmath558 combining this with the estimates and , we obtain @xmath559 we complete the proof by the arbitrary choice of @xmath276 .
let @xmath367 such that @xmath560 .
since @xmath561 , by ( * ? ? ?
* theorem 4.10 ) there exists a compact subset @xmath562 such that @xmath563 .
furthermore , by ( * ? ? ?
* theorem 4.13 ) there exists a probability measure @xmath564 on @xmath153 such that for all @xmath565 , @xmath566 for each @xmath238 , defining @xmath567 where @xmath568 denotes the modification of @xmath267 such that the elements of @xmath568 form a partition of @xmath2^{d}$ ] . roughly speaking
, @xmath568 denotes the collection of @xmath569 `` half close half open cubes '' cubes with side length @xmath570 such that any two distinct cubes are disjoint .
let @xmath571 ( we may consider @xmath572 as a subset of @xmath2^{d}$ ] for convenience of notation ) . for @xmath573
, define the random set @xmath574 let @xmath575 , define the random measure @xmath576 where @xmath577 is the measure @xmath564 restricted to @xmath578 .
let @xmath579 since @xmath153 is a compact set , we obtain that for any @xmath580 , @xmath581 in the following we intend to show that @xmath582 with positive probability , where @xmath583 is the weak limit measure of @xmath584 .
the random sets @xmath585 give rise to an increasing filtration of @xmath586-algebras @xmath587 . for any @xmath588 , we have @xmath589 and @xmath590 therefore @xmath591 in fact this estimates holds for any @xmath592 it follows that @xmath593 thus the sequence @xmath594 is a martingale sequence .
applying the same argument as in ( * ? ? ?
* lemma 8.7 ) , we see that almost surely @xmath584 weakly converges to a measure @xmath583 .
furthermore , applying lemma [ lem : twopoint ] and the condition we obtain @xmath595^d))^2 ) = p^{-2n}{\mathbb{e}}(\lambda(k_n)^2)\\ & = p^{-2n}{\mathbb{e}}(\int \int \textbf{1}_{k_n \times k_n}(x , y ) d\lambda(x ) d\lambda(y))\\ & \leq c \int \int d(x , y)^{s - d-\varepsilon}d\lambda(x ) d\lambda(y)<\infty . \end{aligned}\ ] ] it means that @xmath596^d)\}_{n\in{\mathbb{n}}}$ ] is an @xmath597-bounded martingale .
thus by ( * ? ? ?
* corollary 8.4 ) we obtain that @xmath598^d))={\mathbb{e}}(\nu_1([0,1]^d))=1,\ ] ] and hence @xmath599^d)>0 $ ] with positive probability .
note that for any @xmath340 , we have @xmath600 \backslash k^\omega)=0.$ ] it follows that @xmath601 with positive probability . by the inclusion
we complete the proof .
let @xmath367 such that @xmath560 and @xmath602 .
we use the same notations as in the previous proof .
applying fatou s lemma , fubini theorem , and lemma [ lem : twopoint ] , we obtain @xmath603 the last inequality holds by the choice of @xmath564 , see estimate .
recall that @xmath582 with positive probability . as before this implies that @xmath604 with positive probability . by the arbitrary choice of @xmath276
, we complete the proof .
applying the similar argument to ( * ? ? ?
* chapter 7 ) , we show a different proof from above for theorem [ thm : hitting ] ( 2 ) in the following . for any @xmath605 there
exists a compact subset @xmath606 , such that @xmath607 we choose small @xmath276 satisfies @xmath608 recalling @xmath609 . since @xmath153 is a compact set , we have @xmath610 observe that the events @xmath611 is monotone decrease , hence we have @xmath612 for each @xmath238 , defining @xmath567 where @xmath568 denotes the modification of @xmath267 such that the elements of @xmath568 form a partition of @xmath2^{d}$ ] . roughly speaking
, @xmath568 denotes the collection of @xmath569 `` half close half open cubes '' cubes with side length @xmath570 such that any two distinct cubes are disjoint .
let @xmath564 be a probability measure on @xmath153 such that for any @xmath565 , @xmath613 let @xmath614 , defining @xmath615 where @xmath616 when @xmath617 , otherwise equal zero . for any @xmath618 , we have
@xmath619 it follows that @xmath620 note that for any @xmath68 , @xmath621 observe that there exists a positive constant @xmath622 such that for any @xmath623 , and @xmath624 , we have @xmath625 note that the equality holds when @xmath63 and @xmath626 are interior point of @xmath89 and @xmath627 respectively . applying lemma [ lem : twopoint ] , the conditions and , we obtain @xmath628 here the constant @xmath629 comes from lemma [ lem : twopoint ] . by the cauchy - schwarz inequality ,
we obtain @xmath630 and hence ( paley - zygmund inequality ) @xmath631 combining this with estimates and , we obtain @xmath632 applying the estimates and , we obtain @xmath633 thus we complete the proof .
here we present some examples of exceptional sets for the almost sure type results in the case @xmath101 ( i.e. any element of @xmath0 is a subset of @xmath2 $ ] ) . for @xmath634 , we consider the space @xmath635 .
in fact our examples will always looks like @xmath636 , but the sequences @xmath637 are different in different examples .
it is clear that for any @xmath634 the classic cantor ternary set @xmath638 , and it is well known that @xmath639 for convenience , let @xmath640 .
note that for any @xmath634 , theorem [ thm : dimension ] ( 3 ) claims that almost surely @xmath644 while theorem [ thm : typical ] ( 1 ) implies that for a typical @xmath93 , @xmath645 .
however , we show an concrete example in the following for clearness . for @xmath646 , we divide @xmath2 $ ] into @xmath647 interior disjoint @xmath647-adic closed intervals and choose @xmath648 closed intervals of them from the left part of @xmath2 $ ] .
they are interior disjoint and their union is @xmath649 $ ] . let @xmath18 be the collection of these @xmath648 intervals . given @xmath199 ,
the collection of @xmath650 closed intervals with the same length @xmath651 . for every interval @xmath652
, we divide it into @xmath653 interior disjoint @xmath654-adic closed intervals and choose @xmath655 closed intervals of them from the left part of @xmath656 ( see figure [ figure5 ] ) , and let @xmath204 be the union of the chosen closed intervals .
let @xmath657 .
note that @xmath641 implies that @xmath658 for every @xmath318 , we have @xmath659 and hence @xmath660 it follows that @xmath661 .
thus we complete the proof .
it is clear that the cantor ternary set does not contain any normal numbers , but things are different when we add randomness .
we have the following result for our random cantor sets under the natural measure @xmath77 . for the definition of normal numbers and further results , see @xcite .
this follows by borel s normal numbers theorem and the following lemma .
recall that borel s normal number theorem claims that almost every ( with respect to lebesgue measure ) real numbers are normal .
the following lemma ( observation ) is due to pablo shmerkin .
let @xmath367 , then there is an open set @xmath671 with @xmath672 .
note that @xmath673 , see ( * ? ? ?
* theorem 1.24 ) .
applying fubini s theorem we obtain @xmath674 combining these with fatou s lemma , we have @xmath675 by the arbitrary choice of @xmath676 we finish the proof .
a set @xmath677 is called tube null if for any @xmath678 , there exist countable many tubes @xmath679 covering @xmath33 and @xmath680 . here a tube @xmath681 with width @xmath682 is the @xmath683- neighborhood of some line in @xmath684 .
we refer to @xcite for the background and more details on tube null sets . in @xcite ,
the following result is proved .
it is natural to ask that how about the case @xmath687 .
another interesting question is that what will happen if there is no randomness .
for instance , what happens for the self - similar sets of @xmath539 , that is the elements of @xmath539 we take the same position for the chosen subcubes in every step during our construction .
for the self - similar sets , see ( * ? ? ?
* chapter 9 ) .
note that the classical marstrand - mattila projection theorem ( see e.g @xcite ) implies that any set @xmath689 with @xmath690 is tube null , see ( * ? ? ?
* proposition 7 ) .
thus it is sufficient to consider the self - similar set of @xmath539 with hausdorff dimension larger or equal @xmath686 for above question .
we can also consider which kind of self - similar set or self - affine sets are tube null . in @xcite
, the author proved that the koch snowflake curve is tube null .
in fact we can apply the similar arguments to @xcite to obtain that the sierpiski triangle is tube null also , we omit the details here . for self - affine sets and bedford - mcmullen carpets , see ( * ? ? ?
* chapter 9 ) .
the _ lower dimension _ can be considered as the dual of assouad dimension .
it is defined as follows : @xmath691 the lower dimension was introduced by larman , see @xcite . for the recent works on the lower dimension , we refer to @xcite and references therein . for our random cantor sets ,
if @xmath99 is bounded then we have the dual result for the lower dimension . for the case
@xmath693 is unbound .
observe that any set @xmath93 has lower dimension zero .
thus it is sufficient to show that the formula also give the zero value .
this follows from the fact that for any @xmath694 @xmath695 thus we complete the proof .
let @xmath93 .
for any @xmath367 , there exists @xmath486 such that @xmath700 implies @xmath701 note that there exists @xmath441 which depends on @xmath702 only such that for any @xmath115 , @xmath703 by the condition that @xmath697 , we obtain that @xmath704 , and hence @xmath705 by the arbitrary choice of @xmath276 .
let @xmath93 .
note that there exist positive constants @xmath708 such that for any @xmath709 , @xmath710 hence there exists @xmath711 such that for any @xmath712 , @xmath713 thus the claim follows by the fact that any set of @xmath714 has lower dimension less or equal than @xmath702 .
probability on trees : an introductory climb . in lectures on probability theory and statistics ( saint - flour , 1997 ) , volume 1717 of lecture notes in math . , pages 193 - 280 .
springer , berlin , 1999 . | in this paper we study a class of random cantor sets .
we determine their almost sure hausdorff , packing , box , and assouad dimensions . from a topological point of view
, we also compute their typical dimensions in the sense of baire category .
for the natural random measures on these random cantor sets , we consider their almost sure lower and upper local dimensions . in the end
we study the hitting probabilities of a special subclass of these random cantor sets . |
an optical lattice is a regular periodic potential for neutral cold atoms @xcite which enables the controlled experimental exploration of paradigmatic ideas and models from condensed - matter physics .
this is because cold atomic gases generally allow for a great deal of experimental tunability .
for example , feshbach scattering resonances allow for the interaction strength to be experimentally varied over a considerable range @xcite .
other quantities that may be altered include temperature , density , and strength and shape of the trapping potential .
in particular , an optical - lattice potential plays the role of the ion - lattice potential encountered in electronic solid - state physics .
the energy bands resulting from this periodic potential lead to a quenching of the kinetic energy of the atoms with respect to their interaction energy , enabling the exploration of strongly - correlated phases that play a significant role in condensed - matter physics .
an important model that can be studied experimentally with cold atoms is the single - band hubbard model , which consists of interacting fermions in the tight - binding approximation .
the hubbard hamiltonian is realized by cold atoms in an optical lattice when the potential is strong enough so that only the lowest - energy band is populated @xcite . for bosonic atoms one
then commonly refers to this model as the bose - hubbard model .
the theoretically predicted mott - insulator - to - superfluid phase transition @xcite for this model has indeed been observed experimentally @xcite .
the fermionic hubbard model , referred to simply as the hubbard model , is important in the context of high - temperature superconductivity @xcite and has also been realized with cold atoms @xcite . at half
filling , corresponding to one particle per lattice site , the ground state of this model is antiferromagnetic , i.e. , a nel - ordered state , for strong enough on - site interactions . as the filling factor
is reduced by doping , the system is conjectured to undergo a quantum phase transition to a @xmath0-wave superconducting state @xcite .
a theoretical proof of the existence of @xmath0-wave superconductivity in the hubbard model is still lacking and would be a major step towards understanding the superconducting state of the cuprates . with the recent experimental advances in the field of ultracold atoms ,
an experimental exploration of this issue is within reach . in view of this motivation ,
a significant problem is determining how the nel state of the hubbard model can be reached experimentally . in this paper
, we study theoretically the process of adiabatically turning on the optical lattice @xcite , with the goal of determining the conditions required for an initially trapped balanced two - component fermi gas with repulsive interactions to reach the nel state in the lattice .
experimentally , the presence of antiferromagnetic order in this cold - atom experiment can be subsequently detected from shot - noise correlations in the density distribution @xcite .
( @xmath1 is the recoil energy ) from single - site mean - field theory ( solid curve ) and with fluctuations ( dash - dotted curve ) , where @xmath2 is the fermi temperature in the trap .
the horizontal dotted lines illustrate cooling and heating into the nel state at constant entropy by starting in the harmonic trap and adiabatically turning on the lattice.,title="fig : " ] * * our results are summarized in fig .
[ fig1 ] . for initial temperatures lower than @xmath2 , the fermi temperature in the trap
, the entropy per particle in the trap depends linearly on temperature as is shown by the dashed line .
the optical lattice is then turned on adiabatically and to determine the final temperature of the gas we need the entropy per atom in the lattice . for a sufficiently smooth trapping potential such that the tunneling does not become site - dependent , the only effect of the trap
is to place a restriction on the total number of particles which we discuss later and , other than this , we may neglect the trap for calculations in the lattice .
since we consider balanced gases here , we will at sufficiently low temperatures first enter the mott phase with one particle per site , and the subsequent evolution of the gas is then described by the heisenberg model for the spins alone . the result from the usual mean - field theory
is shown for a lattice depth of @xmath3 ( where @xmath1 is the recoil energy ) by the black curve , and is equal to @xmath4 everywhere above the critical temperature @xmath5 . since entropy is conserved in adiabatic processes ,
the final temperature is simply the temperature at which the final entropy in the lattice equals the initial entropy in the trap .
two such processes are shown by the dotted lines for different initial temperatures demonstrating that the gas is sometimes heated and not cooled by the lattice .
nevertheless , mean - field theory leads to the intuitive result that as long as the entropy per particle in the initial state is less than @xmath6 , which is the maximum entropy of the heisenberg model , the nel state is always reached by adiabatically turning on the optical lattice .
the inclusion of fluctuations leads however to a more restrictive condition . to probe the effect of fluctuations , we present an improved mean - field theory which produces a temperature - dependent entropy above @xmath5 , as seen from the inset of fig . [ fig2 ] .
although this approach is exact at high temperatures , it fails to account for spin waves present at low temperatures and for critical phenomena near @xmath5 . by further extending the improved mean - field theory to reproduce the correct critical and low temperature behavior due to fluctuations
, we are able to determine the entropy in the lattice for all temperatures ( red curve in fig . [ fig1 ] ) .
in particular , we find that fluctuations lower the entropy of the atoms in the square lattice at @xmath5 as @xmath7 where @xmath8 is the critical exponent of the correlation length @xmath9 . for the case of three dimensions , @xmath10 @xcite .
as a result , the initial temperature required to reach the nel state is more than 20% lower than that found from the usual mean - field theory , but fortunately remains experimentally accessible .
for example , with @xmath11k atoms and a final lattice depth of @xmath12 the nel state is achieved when the final temperature in the lattice is @xmath13 , which can be obtained with an initial temperature of @xmath14 .
the hamiltonian for the hubbard model is given by @xmath15 in terms of fermionic creation and annihilation operators , denoted by @xmath16 and @xmath17 , respectively , where @xmath18 labels the two hyperfine spin states @xmath19 or @xmath20 of the atoms .
in the first term of this expression , the sum over lattice sites labeled by indices @xmath21 and @xmath22 is over nearest neighbors only and proportional to the hopping amplitude given by @xmath23 here , @xmath24 is the depth of the optical lattice potential defined by @xmath25 , % v({\bf x})=v_0\sum_{\alpha } \cos^2\left(\frac{2\pi x_{\alpha}}{\lambda}\right),\ ] ] where @xmath26 is the wavelength of the lattice lasers .
the second term in the hamiltonian corresponds to an on - site interaction of the strength given in the harmonic approximation by @xmath27 where @xmath28 is the @xmath29-wave scattering length which is equal to @xmath30 for @xmath31 .
it is well - known @xcite that at half filling and in the limit that @xmath32 the ground state of the hubbard model is antiferromagnetic and that , for @xmath33 , its low - lying excitations are described by the effective heisenberg hamiltonian @xmath34 with @xmath35 being one half times the vector of pauli matrices .
the exchange constant @xmath36 arises from the superexchange mechanism .
that is , the system can lower its energy by virtual nearest - neighbor hops only when there is antiferromagnetic ordering . within the usual mean - field analysis of the effective hamiltonian in eq .
, the total entropy for @xmath37 atoms in the optical lattice is given by @xmath38 where @xmath39 is the landau free energy , @xmath40 \right\}~,\ ] ] in terms of the staggered , or nel , order parameter @xmath41 for the phase transition to the antiferromagnetic state .
in the expression for the free energy , @xmath42 is the number of nearest neighbors for a three - dimensional simple square lattice on which we focus here , @xmath43 is the thermal energy , and @xmath44 is the equilibrium value of the order parameter determined from @xmath45 it is nonzero below a critical temperature @xmath46 . after solving eq .
the entropy is determined using eq . .
the results for @xmath47 and @xmath48 obtained in this way are plotted as solid black curves in figs .
[ fig1 ] and [ fig2 ] .
the entropy @xmath49 of the initial normal state before ramping up the optical lattice is the entropy of a trapped ideal fermi gas .
it is most conveniently determined from the grand potential @xmath50~,\ ] ] where @xmath51 is the chemical potential , and the effect of the harmonic trapping potential with the effectively isotropic frequency @xmath52 is incorporated via the density of states @xmath53 of the atoms .
the entropy at fixed total particle number @xmath54 is then given by @xmath55 . at temperatures
much lower than the fermi temperature in the trap , given by @xmath56 , we find in this manner that @xcite @xmath57 .
now , by equating the final and initial entropies we calculate the temperature of the heisenberg spin system that results after adiabatically turning on the optical lattice , in terms of the initial temperature of the trapped fermi gas . from the expression for the free energy , eq .
we immediately see that @xmath58 for all temperatures @xmath59 , as was shown in fig .
although this is the correct high - temperature limit of the entropy , temperature dependence will lower the entropy at @xmath5 and therefore lower the initial temperature required to achieve the nel state . to obtain the temperature dependence above @xmath5
, we must thus go beyond single - site mean - field theory to include fluctuations .
the simplest such model described below incorporates the interaction of a given site with one of its neighbors exactly and treats interactions with the rest of the neighbors within mean - field theory .
the two - site hamiltonian for neighboring sites labeled `` 1 '' and `` 2 '' is given by @xmath60 where the last term is a correction to avoid double counting of mean - field effects .
diagonalizing this hamiltonian we obtain the free energy @xmath61
f_{\rm l } ( { \bf n } ) = & n\biggl\lb\half(z-1)j|\bn|^2 - \frac{1}{2\beta}\ln \biggl[2 e^{-\beta j/4 } \\ & + 2 e^{\beta j/4 } \cosh \ ( \frac{\beta j}{2 }
\sqrt{1 + 4(z-1)^2|{\bf n}|^2}\)\biggr]\biggr\rb~ , \end{aligned}\ ] ] and find the entropy from eq . with the condition eq .
as in the single - site model .
the results are plotted in fig .
[ fig2 ] , where we see that fluctuations lower the critical temperature and also bring about a 2% depletion of the order parameter which is now less than 0.5 near @xmath62 . of the single ( solid curves ) and two - site ( dashed curves ) mean - field theories , the latter of which shows depletion at @xmath62 and a lowering in @xmath5 .
the entropy of both theories is plotted in the inset . above @xmath5
we see that the entropy of the two - site theory is temperature dependent.,title="fig : " ] * * the two - site result carries the exact @xmath63 dependence of the entropy of the heisenberg model at high temperatures . near @xmath62 , however , the entropy is still exponentially suppressed reflecting the energy cost of flipping a spin .
this exponential suppression is an artefact of the mean - field approximation that ignores the goldstone modes which are present in the symmetry - broken phase .
furthermore , critical behavior can not be properly accounted for by a one- , two- or higher - site model since , near the onset of nel order , critical fluctuations extend throughout the entire lattice so one would in principle need to include all sites exactly . to overcome these shortcomings ,
we extend our two - site model below to all temperatures .
the two - site mean - field theory produces the correct normal - state entropy behavior in the high - temperature limit , @xmath64~.\ ] ] in the low - temperature regime , the entropy is determined from spin - wave fluctuations prevalent near @xmath62 which give a black - body - like entropy , @xmath65 the continuous interpolation between these two regimes has the additional constraint that , near @xmath5 , we should obtain the correct critical behavior of the antiferromagnet , namely , the correct universal ratio of the amplitudes above and below the phase transition @xmath66 and correct critical exponent @xmath67 where @xmath68 this follows from the fact that the singular part of the free energy density behaves as @xmath69 , where the correlation length diverges like @xmath70 as @xmath71 .
explicit expressions for the entropy embodying the correct behavior in the low- , high- and critical temperature regimes are presented in the appendix of this paper , and plotted as the red curve in fig .
[ fig1 ] for @xmath72 using @xmath73 and @xmath10 @xcite , and the nel temperature of @xmath74 @xcite .
their value at @xmath5 leads to the central result of this paper , namely , eq .
( [ eq : criticals ] ) which specifies the initial entropy required to reach the nel state .
as briefly mentioned earlier , there is a limit on the total number of atoms in the trap , beyond which at low temperatures it is energetically more favorable to doubly occupy sites in the center of the trap , thereby destroying the antiferromagnetic state , rather than singly occupying outlying sites where the trap potential is larger than @xmath75 .
thus , insisting that the system end up in the mott - insulator state with one particle per site entails the upper bound , @xmath76 , where @xmath77 is the mass of the atoms and @xmath26 is the wavelength of the lattice lasers .
for @xmath11k atoms in a lattice with a wavelength @xmath78 nm and depth @xmath12 , and with a harmonic trap frequency @xmath79 hz , @xmath80 which is well above the typical number of atoms in experiments .
we have also attempted to determine the effect that fluctuations have on the entropy in a more microscopic manner by studying gaussian fluctuations about the mean - field @xmath48 for the single - site mean - field theory in the low - temperature regime .
but such a random - phase approximation has severe complications related to the fact that @xmath48 enters in the magnon dispersion as @xmath81 .
hence , as can be already seen from eq . , the contribution of the magnons to the entropy diverges when @xmath82 near @xmath5 .
one way to potentially resolve this issue is to start from the hubbard hamiltonian eq .
directly but such an analysis is involved @xcite and has yet to be carried out . in the above ,
we have focused on the @xmath72 case . whilst our results can easily be extended to the @xmath83 case , a more pertinent way to reach the two - dimensional antiferromagnet most relevant to high - temperature superconductors ,
would be to adiabatically prepare a three - dimensional nel state , as explained in this paper , and then decrease the tunneling in one direction by changing the intensity of one of the lattice lasers . in this way ,
the three - dimensional system is changed into a stack of pancakes of atoms in the two - dimensional nel state .
furthermore , studying doped optical lattices made by introducing a small imbalance in the initial state may shed some light on the physics of high - temperature superconductors and would be an exciting direction for future research .
we would like to thank gerard barkema , dries van oosten and randy hulet for helpful discussions .
this work is supported by the stichting voor fundamenteel onderzoek der materie ( fom ) , the nederlandse organisatie voor wetenschaplijk onderzoek ( nwo ) , and the deutsche forschungsgemeinschaft ( dfg ) .
for temperatures above @xmath5 we use @xmath84+\ln(2)~,%\mkern-30mu\ ] ] with @xmath85 and @xmath86 @xcite .
the first term embodies the correct critical behavior whereas the remaining terms are present to recover the correct high - temperature limit . below @xmath5 , however , we have @xmath87 \frac{s(t\leq t_c)}{nk_{\rm b } } = -\alpha_2 & \bigg[\(\frac{t_c - t}{t_c}\)^\kappa - 1 + \kappa \frac{t}{t_c } \\&- \frac{\kappa(\kappa - 1)}{2 } \frac{t^2}{t_c^2}\bigg ] + \beta_0 \frac{t^3}{t_c^3 } + \beta_1 \frac{t^4}{t_c^4}~,%\mkern-30mu \end{aligned}\ ] ] where @xmath88 \alpha_2 = & \begin{aligned}[t ] \frac{6}{(\kappa-1)(\kappa-2)(\kappa-3 ) } & \bigg(\frac{4\pi^2 k_{\rm b}^3 t_c^3}{135\sqrt{3}j^3 } \\ & - \alpha_1(\kappa-1 ) + \beta_1 - \ln(2)\bigg)~;%\mkern-30mu \end{aligned}\\ \beta_0 = & \frac{\kappa}{(\kappa-3 ) } \bigg(\frac{4\pi^2 k_{\rm b}^3 t_c^3}{45\sqrt{3}\kappa j^3 } + \alpha_1 ( \kappa-1 ) - \beta_1 + \ln(2)\bigg)~;%\mkern-30mu \\
\beta_1 = & \ln 2 - j^2\frac{6(a^+/a^- + 1 ) + \kappa(\kappa-5)}{64\kappa k_{\rm b}^2 t_c^2 a^+/a^- } - \frac{4\pi^2 k_{\rm b}^3 t_c^3}{135\sqrt{3}j^3}~.%\mkern-30mu \end{aligned}\ ] ] the first and last terms in @xmath89 embody the critical phenomena and allow for the continuous interpolation with @xmath90 respectively , whereas the remaining terms are included to retrieve the correct low - temperature behavior .
99 p. s. jessen and i. h. deutsch , adv . at .
. phys . * 37 * , 95 ( 1996 ) .
w. c. stwalley , , 1628 ( 1976 ) .
e. tiesinga , b. j. verhaar , and h. t. c. stoof , phys .
a * 47 * , 4114 ( 1993 ) .
d. jaksch , c. bruder , j. i. cirac , c. w. gardiner and p. zoller , , 3108 ( 1998 ) . m. p. a. fisher , p. b. weichman , g. grinstein , and d. s. fisher , , 546 ( 1989 ) .
m. greiner , o. mandel , t. esslinger , t. w. hnsch , and i. bloch , nature * 415 * , 39 ( 2002 ) .
j. g. bednorz and k. a. mller , z. phys .
b * 64 * , 189 ( 1986 ) .
w. hofstetter , j. i. cirac , p. zoller , e. demler and m. d. lukin , , 220407 ( 2002 ) .
m. khl , h. moritz , t. stferle , k. gnter , and t. esslinger , , 080403 ( 2005 ) .
p. a. lee , n. nagaosa , and x .-
wen , , 17 ( 2006 ) .
f. werner , o. parcollet , a. georges , and s. r. hassan , , 056401 ( 2005 ) .
p. b. blakie , a. bezett , p. buonsante , phys .
a * 75 * , 063609 ( 2007 ) .
e. altman , e. demler and m. d. lukin , phys .
a. * 70 * , 013603 ( 2004 ) .
t. rom , th .
best , d. van oosten , u. schneider , s. foelling , b. paredes , and i. bloch , nature * 444 * , 733 ( 2006 ) .
see , for example , a. auerbach , _ interacting electrons and quantum magnetism _
( springer - verlag , new york , 1994 ) .
l. d. carr , g. v. shlyapnikov , and y. castin , phys .
* 92 , * 150404 ( 2004 ) .
j. zinn - justin , _ quantum field theory and critical phenomena _ ,
4th edition ( oxford , 2002 ) .
r. staudt , m. dzierzawa , and a. muramatsu , eur .
j. b * 17 * , 411 ( 2000 ) .
k. borejsza and n. dupuis , phys . rev .
b * 69 * , 085119 ( 2004 ) . | we theoretically study the possibility of reaching the antiferromagnetic phase of the hubbard model by starting from a normal gas of trapped fermionic atoms and adiabatically ramping up an optical lattice .
requirements on the initial temperature and the number of atoms are determined for a three dimensional square lattice by evaluating the nel state entropy , taking into account fluctuations around the mean - field solution .
we find that these fluctuations place important limitations on adiabatically reaching the nel state . |
observations of co line emission from high redshift ( @xmath0 ) galaxies has become an important diagnostic tool in the study of galaxy formation .
such studies include observations of ir - luminous starburst galaxies selected at ir and ( sub)mm wavelengths ( brown & van den bout 1992 ; solomon , downes & radford 1992 ; frayer et al . 1998 ; ivison et al .
2001 ) , and active galaxies ( qsos and radio galaxies ) selected at optical wavelengths ( omont et al . 1996a , b , 2001 ; ohta et al . 1996 ; barvainis et al
1998 ; guilloteau et al .
1997 , 1999 ; papadopoulos et al .
2000 , 2001 ; carilli , menten & yun 1999 ; carilli et al .
due to the sensitivity limitations of current mm - wave telescopes , such studies have been limited to extreme high mass , high luminosity galaxies , corresponding to galaxies with far - ir ( fir ) luminosities : @xmath2 l@xmath3 , and molecular gas masses : @xmath4 m@xmath3 . while such systems are rare in the nearby universe , @xmath5 mpc@xmath6 , or 10@xmath7 times lower than @xmath8 galaxies , ( sub)mm surveys suggest that the comoving number density of such objects increases by up to three orders of magnitude by @xmath9 to 3 ( smail , ivison & blain 1997 ; blain et al .
1999 ; barger et al .
another serious limitation to current instruments is the narrow bandwidth of the spectrometers .
current spectrometers have maximum bandwidths of 0.5 to 1 ghz , corresponding to 1500 to 3000 km s@xmath10 at 100 ghz .
the limited sensitivity and bandwidth of existing instruments effectively preclude large - volume searches for high-@xmath0 galaxies via their molecular line emission .
the next generation of mm- and cm - wave telescopes will have sensitivities and bandwidths at least an order of magnitude better than existing instruments . with such capabilities
it becomes feasible to consider searches for high-@xmath0 galaxies at mm and cm wavelengths .
this problem has already been addressed in detail for ( sub)mm - wave observations using the atacama large millimeter array ( alma ) , the next - generation mm - wave telescope ( blain et al .
2000 paper i ; blain 1996 , 2001 ; van der werf & israel 1996 ; silk & spaans 1997 ; combes , maoli , & omont 1999 ; gnedin , silk & spaans 2001 ) .
this paper dramatically expands the discussion in paper i of the capabilities of future telescopes to study the evolution of dusty star forming galaxies via their low - excitation molecular line emission . in paper
i , formalism was developed to investigate optimal molecular line surveys which maximize the detection rate of galaxies for a given amount of observing time , and the analysis presented herein relies on the same formalism .
the importance of such surveys is delineated in detail in paper i. to summarize , measuring molecular lines , and especially the low - excitation lines that trace the bulk of gas in the interstellar medium ( ism ) , provides a direct probe of the massive gas reservoirs required to fuel star formation in nascent galaxies .
in addition , the velocity width of the lines provides an estimate of the dynamical mass of the systems .
observations of the infrared background , and source counts at ( sub)mm wavelengths , suggest that a substantial fraction of the star formation that occurs in the universe is invisible at optical wavelengths due to dust obscuration ( blain et al.1999 ) .
future molecular line surveys may be the only way of obtaining a complete census of the redshift distribution of this cosmic star formation .
paper i emphasized the capabilities of future mm - wave telescopes , such as the alma , for discovering high redshift galaxies via their co emission .
the focus of the analysis presented herein is on future cm - wave telescopes , such as the square kilometer array ( ska ) and the expanded very large array ( evla ) , for observing low - excitation co molecular line emission from high-@xmath0 galaxies .
the important difference between mm- and cm - wave observations is that mm - wave telescopes are limited to studying higher order co transitions from high-@xmath0 galaxies
co(3 - 2 ) or higher at 100ghz for @xmath11 while cm - wave telescopes probe the lower order transitions , co(2 - 1 ) and co(1 - 0 ) .
clearly such complementarity is critical for determining excitation conditions in high-@xmath0 molecular gas clouds .
in particular , mm - wave surveys of high order transitions will be biased toward high - density , high - temperature gas , with @xmath12cm@xmath6 and @xmath14k ( papadopoulos et al .
2001 ; papadopoulos & ivison 2001 ) .
an important advantage of cm - wave surveys is the large fractional bandwidth and primary beam area . at high redshifts
cm - wave spectroscopy will be able to map out the 3-dimensional spatial structure of line - emitting galaxies , and thus to trace directly the build up of large - scale structure in the universe .
both the thickness in redshift , and the projected area on the sky of the surveyed volume , will be much greater in the cm waveband than at mm wavelengths .
it is also likely that low - excitation lines will map out a larger fraction of the volume of the ism in these galaxies , and thus provide the opportunity to study in detail the spatially resolved kinematic structure of most of the gas in the ism in high - redshift galaxies , which resides in the cold phase .
these observations would provide information about both the evolution of the masses of galaxy disks and the disruption of disks by galaxy interactions at large redshifts .
this will allow valuable new tests of understanding of the assembly and formation of galaxies .
we use the latest galaxy evolution models from paper i , modified to take account of additional data and revised cosmology .
while these models represent perhaps the best current estimates of the evolution of dusty star forming galaxies , the observational constraints are such that substantial revisions , especially of the source redshift distribution , are certainly possible .
likewise , the design specifications of the next - generation cm - wave telescopes are still being considered .
hence , this paper is not meant as a definitive and final prediction of the results of future molecular line surveys with cm - wave telescopes , but merely to advocate the possibilities .
a particular concrete question we hope to address is : what choices in the design of such telescopes ( eg .
high frequency limit , maximum baseline ) will facilitate the study of molecular line emission from high-@xmath0 galaxies ?
we assume h@xmath15 = 65 km s@xmath10 mpc@xmath10 , @xmath16 , and @xmath17
the models we employ for predicting co source counts are updated from paper i. these models employ an analytic description of pure luminosity evolution of the low-@xmath0 _ iras _
60-@xmath18 m luminosity function ( saunders et al . 1990 ) .
the evolution function has the form , @xmath19 \ , { \rm cosh}^2 c.\ ] ] at very low and very high redshifts the function can be approximated by @xmath20 , with @xmath21 and @xmath22 respectively . with the current cosmology , and taking into account all available far - ir and ( sub)-mm background , count and redshift distribution data , the values @xmath23 and @xmath24 are required , if a typical dust temperature of 37k is assumed .
the fitting procedures and details of the information used are explained in blain et al .
a plot of this function is shown in fig.1 of blain ( 2002 ) .
the evolution peaks at @xmath25 , at which the bolometric luminosity density of infrared - luminous galaxies is about 40 times greater than at @xmath26 .
the strength of the emission from co lines is calculated as described in paper i. excitation conditions , and thus line ratios , are derived from a standard large velocity gradient model ( frayer & brown 1997 ) with a kinetic temperature of about 50k and a density of 10@xmath27cm@xmath6 .
this model provides a reasonable description of the co emission from two well - studied fir - luminous galaxies at @xmath28 , and their properties are used to normalize the co emission line strengths to the evolving bolometric luminosity function of galaxies described above .
table1 lists the assumed parameters for telescope capabilities for the ska , evla , and alma .
columns 2 and 3 give the antenna diameter and the instantaneous frequency range covered , respectively .
column 4 gives the effective aperture ( total collecting area @xmath29 aperture efficiency ) divided by the system temperature , and column 5 gives the field of view ( fov ) of the primary elements at the given frequencies .
column 6 gives the rms sensitivity in 1 hour for a fwhm spectral channel of 300kms@xmath10 .
the design for the ska , including the primary stations for the array , the correlator , and the frequency range covered , is still under investigation . for the analysis below we adopt the united states ska consortium design concept of a square kilometer of collecting area comprised of small ( 7 m ) diameter antennas , arranged in 1000 stations of 26 antennas per station ( cordes , preston , & tarter 2001 ) .
we assume the elements in each station are closely packed ( @xmath30 1 m separation ) , implying an effective station diameter of about 40 m .
beam - forming correlators generate ` effective beams ' at each station corresponding roughly to the diffraction limited beam of a 40 m diameter antenna .
signals from these effective beams are then cross correlated with the corresponding beams of other stations .
the current specifications for the ska require imaging of the full primary beam of the 7 m elements , thereby requiring roughly @xmath31 = 33 beams per station .
the number of cross correlations ( per polarization and lag channel ) required for this type of a system is then the sum of the number of cross correlations between stations @xmath32 , plus the number of cross correlations per station ( ie .
the beam - forming correlator ) @xmath33 , for a total of @xmath34 cross correlations .
this value is an order of magnitude smaller than the @xmath35 cross correlations required in the case of full cross correlation of the 26000 7 m elements .
the trade - offs between full cross correlation and beam forming instruments are in imaging fidelity , which is not a critical issue for the co searches discussed herein , and in calibration flexibility , ie .
the difficulty and complexity is shifted from the construction of the large correlator required for full cross correlation to ensuring the stability of the many beam - forming correlators .
the analysis discussed below is not critically dependent on the specific design of the array stations , except with respect to the imaging fov . for a fixed collecting area
a large fov is clearly preferable for surveys , as the source detection rate increases linearly with fov .
an important point is the high - frequency limit of the ska .
the current straw - man design has a maximum frequency of 22ghz .
however , weinreb & bagri ( 2001 ) have shown that going to higher frequencies may be feasible , both in terms of the antennas and the receivers . in this paper
we consider co line searches in a 43-ghz band in addition to the standard 22-ghz band .
one of the main motivations of this paper is to gauge the scientific advantages of going to higher frequency with future large area cm telescopes .
such information is required for proper understanding of the trade - offs between scientific capability and cost .
we assume a 4-ghz instantaneous total bandwidth , and a 22-ghz aperture efficiency that is a factor two less than that at 5ghz , ie . a ` half - ska ' .
the evla is included in the analysis since it represents the nearer - term capabilities for cm - wave high-@xmath0 co surveys .
the evla represents a major step forward in the study of low - order high-@xmath0 molecular lines in a number of ways .
first , through improved receivers and antenna structures the evla will be about twice as sensitive as the current vla at 43ghz for spectral line observations .
second , the current vla correlator is limited to only a 50-mhz bandwidth with 7 spectral channels , which corresponds to only 350kms@xmath10 at 43ghz . such a narrow band
both precludes searches for high-@xmath0 co emission and provides very limited spectral information ( carilli et al .
the evla will have an 8-ghz bandwidth in two polarizations with 16000 spectral channels .
third , the current vla high - frequency bands are limited to 20.525 ghz and 4049ghz .
the evla will have continuous frequency coverage from 1 to 50ghz .
the alma is included as a standard of comparison for future ( sub)mm telescopes ( see paper i for details ) .
for both the evla and the alma we consider the optimum bands for co searches ( 40 and 230ghz , respectively ) . note that the sensitivity of the evla at 43 ghz is comparable to that expected for alma at 43 ghz the larger collecting area of the evla is off - set by the higher aperture efficiency and better observing site for alma .
the inclusion of a 43 ghz system for alma is currently under debate . if such a system is included , then the sensitivity for the evla in table 1 is essentially that expected for the alma at this frequency , while the alma will have a factor four larger fov
. arrays of heterodyne receivers on large single dish telescopes offer a potentially competitive method for performing wide field molecular line surveys . in order to sample the same area of the sky with a single dish relative to an interferometer
the number of independent single dish beams that need to be sampled is given by the ratio of the fov of the interferometer to that of the single dish .
for example , the 100-m green bank telescope and the evla have similar collecting areas , so a 16 element receiver array on the gbt will survey a similar area of the sky as the evla with comparable sensitivity . a similar size receiver array on the 50-m large millimeter telescope ( lmt ) will sample the same fov as the alma , with about 25@xmath1 of the sensitivity . of course , a wide - band autocorrelation spectrometer will be required for each element of the receiver array .
the telescope optics in both cases ( the gbt and the lmt ) will support such receiver arrays , since both telescopes have been designed to support large format ( @xmath36 ) bolometer cameras .
as a metric of the capabilities of the different telescopes we use the detection rate for an ` optimal survey ' . the optimal survey balances area surveyed versus sensitivity to sample the source counts at the ` knee ' of the distribution , corresponding to the regime where @xmath37 , where @xmath38 is the flux in the line ( see paper i for details ) .
this criterion maximizes the number of sources detected in a given observing time . in fig.1
we show the predicted 22- and 43-ghz line source counts for a 4-ghz bandwidth using the formalism for infrared - selected galaxy evolution developed in paper i. the line counts increase approximately linearly with bandwidth .
table 2 lists the frequency ranges and optimal depths for the different telescopes , as well as the number of pointings per hour for the optimal depth and the detection rate of high-@xmath0 co - emitting galaxies . for the baseline specification of a 22 ghz high frequency upper limit , the ska galaxy detection rate will be a factor two larger than that of the alma .
if the high frequency limit is increased to 43 ghz , then the ska becomes the pre - eminent instrument for discovering high redshift galaxies via their co emission , detecting 100 s of galaxies per hour .
the evla is clearly much slower than the ska in terms of discovering high redshift galaxies due to its lower sensitivity and smaller fov .
it is important to keep in mind that the line surveys at different frequencies are sampling different co transitions at different redshifts . in fig.2
we delineate the detection rates in terms of the different transitions and redshifts .
surveys with the ska at 22ghz to the optimal depth of @xmath39wm@xmath40 detect almost exclusively the co(1 - 0 ) line from galaxies with @xmath41l@xmath3 at @xmath42 . at 43ghz at the optimal depth of @xmath43wm@xmath40 , about 92% of the detections are co(1 - 0 ) lines from galaxies with @xmath44wm@xmath40 at @xmath25 , with about 7.5@xmath1 being co(2 - 1 ) lines from galaxies with @xmath45 at @xmath46 , and perhaps 0.5% of the detections being co(3 - 2 ) emission from galaxies with similar fir luminosity but at @xmath47 , if such galaxies exist .
surveys with the alma at 230ghz to the optimal depth of @xmath48wm@xmath40 are dominated by galaxies with @xmath49l@xmath3 , at @xmath50 to 2 , emitting co(4 - 3 ) to co(6 - 5 ) , with a minor contribution from galaxies with @xmath51l@xmath3 at @xmath52 emitting co(3 - 2 ) and from more - luminous galaxies with @xmath53l@xmath3 emitting co(7 - 6 ) at @xmath28 .
the alma can also potentially detect a comparable number of redshifted fine - structure lines from higher redshifts in this band ( paper i ) .
we consider briefly a few issues relating to the cm - wave study of co emission from high redshift galaxies , including : ( i ) maximum baselines for the array given the expected brightness temperatures , ( ii ) the capabilities of studying large scale structure through optimal surveys , and ( iii ) contamination of such surveys by emission from hcn the next strongest thermal molecular emission species in the relevant frequency range .
an important issue when considering detecting thermal emission from high redshift galaxies with future cm telescopes is the maximum baselines dictated by the intrinsic brightness temperature of the emission .
the expected ( rest - frame ) brightness temperature for the co emission from high-@xmath0 galaxies is likely to be @xmath54 k. this corresponds to an observed brightness temperature of @xmath55 k for a source at @xmath56 .
the peak flux density of the line for the optimal 43-ghz survey corresponds to about 0.2mjy for the ska . in order not to spatially over - resolve the sources , much of the collecting area of the array
must then be on baselines @xmath57 km , corresponding to @xmath58 resolution at 43 ghz .
the second issue is large scale structure .
the large number of galaxies discovered over a relatively narrow redshift range for sensitive molecular line surveys at cm wavelengths has the added benefit of facilitating three - dimensional studies of large - scale structure at moderate and high redshifts .
operating at 2024ghz , the optimal ska survey with a station diameter of 7 m detects co(1 - 0 ) emission from about 360 galaxies in 24hours .
these galaxies are in the redshift range @xmath59 to 4.75 over an area of 3.5deg@xmath60 , corresponding to a comoving volume of 0.044gpc@xmath61 . at the higher frequency of 40 - 44ghz , the co(1 - 0 ) emission from about 4200 galaxies would be detected in 24hours from @xmath62 to 1.88 over an area of 23deg@xmath60 and a comoving volume of 0.085gpc@xmath61 .
a volume of about 0.1gpc@xmath61 corresponds to a representative volume of the universe , and so these surveys will provide excellent probes of large - scale structure .
the geometries of both of these daily survey volumes are only about five times deeper than they are wide .
this should ensure that three - dimensional large - scale structure is sampled reliably .
the space density of co - emitting galaxies is expected to be about 100 times lower than for optically - selected , spectroscopically - confirmed lyman - break galaxies ( steidel et al .
1999 ) , and about 10 times less than submm - selected galaxies ( blain et al .
however , the geometry of the three - dimensional redshift surveys should make low - excitation co surveys an ideal tracer of large - scale structure .
in particular , as the co line surveys sample higher luminosity galaxies , their clustering strength relative to the dark matter at high redshift might be stronger than for lower luminosity galaxies in standard hierarchical galaxy formation models ( davis et al . 1985 ) .
whether this is true will provide important new information about bias and the galaxy formation process , especially as mass measurements will be available for comparison from co line widths and excitation ratios .
finally there is the issue of hcn emission .
solomon et al . (
1997 ) have shown a non - linear relationship between @xmath63(co(1 - 0 ) ) ( in k km s@xmath10 pc@xmath60 ) and @xmath64 for galaxies in the sense that high @xmath64 galaxies have lower values of @xmath63(co(1 - 0 ) ) than would be expected based on a linear relationship .
considering @xmath64 to be a measure of star formation rate , and @xmath63(co(1 - 0 ) ) to be a measure of molecular gas mass , this would imply an increase in star formation efficiency ( @xmath65 ) with increasing luminosity , in particular for galaxies with @xmath64 @xmath66 l@xmath3 ( solomon et al . 1997 ) .
for dense nuclear starbursts a number of groups ( solomon et al . 1997 ; mao et al .
2000 ; weiss et al .
2001 ) have shown that the densities are such that the entire interstellar medium in the starburst regions may be molecular , and that the co(1 - 0 ) emission may be dominated by this molecular inter - cloud medium , as opposed to being from the denser star forming clouds themselves .
in this situation it appears that higher density molecular tracers , such as hcn , which has a critical density of order @xmath67 cm@xmath6 , may be a better probe of the star forming clouds themselves .
gao & solomon ( 2001 ) show that the relationship between @xmath63(hcn(1 - 0 ) ) and @xmath64 remains linear , with @xmath68 over the range : @xmath69 l@xmath3 to @xmath70 l@xmath3 ( see solomon 2001 ) .
given this linear relationship , and the non - linear relationship between @xmath64 and @xmath63(co(1 - 0 ) ) , the ratio : @xmath71 varies from about 0.025 for galaxies with @xmath64 = 10@xmath72 l@xmath3 to 0.10 for @xmath64 = 10@xmath73 l@xmath3 . in order to estimate the contamination of optimal cm - wave co surveys by hcn emission , we have included the hcn(1 - 0 ) ( 88.6 ghz rest frequency ) emission in the galaxy formation models discussed in section 2 . at the most efficient depths for the co searches , we find that the number of hcn(1 - 0 ) detections would be about 5@xmath1 of the total number of co(1 - 0 ) lines at 22 ghz and about 1.5@xmath1 of the total number of lines at 43 ghz .
this emission would correspond to galaxies with @xmath64 @xmath74 l@xmath3 at @xmath75 for the 22 ghz survey , and similar luminosity galaxies at @xmath50 for the 43 ghz survey .
hence , hcn emission should not present a major confusion problem for such optimal cm - wave co line searches . on the other hand
, we also find that the fractional contamination by hcn is likely to increase with the depth of the survey , eg .
a 22-ghz survey to a depth of @xmath76 w m@xmath40 will contain about 20@xmath1 hcn(1 - 0 ) lines .
from table1 and fig.2 , it can been seen that future large area cm - wave telescopes operating at 22 ghz will be competitive with future mm - wave telescopes in terms of discovering high-@xmath0 star - forming galaxies through their molecular line emission .
increasing the high frequency limit to 43-ghz allows for optimal surveys which are 20 times faster than the alma in terms of discovering high-@xmath0 galaxies .
detection rates are a simplistic metric for the alma and the ska , and it is important to emphasize their complementarity .
alma surveys will be dominated by higher - order co lines from intermediate redshift ( @xmath77 ) , intermediate luminosity ( l@xmath78 l@xmath3 ) objects .
surveys with the ska at 22 ghz will be dominated by low - order transitions from higher luminosity ( l@xmath79 l@xmath3 ) galaxies at higher redshifts ( @xmath80 ) . for the evla at 43 ghz the combined small
fov and low sensitivity make it a much slower survey instrument that either the alma and ska .
however , the evla has adequate sensitivity to resolve and study at sub - arcsec angular resolution the low - order co emission from individual fir - luminous , high-@xmath0 objects selected from wide - field surveys at other wavelengths , such as surveys using ( sub)mm bolometer arrays on large single - dish telescopes . and perhaps most importantly
, the large fractional bandwidth of the evla will allow for redshift determinations via molecular line searches using pointed observations of individual objects .
the evla will thus provide the first look into the nature of low - order co emission from high-@xmath0 galaxies .
the importance of such capabilities has already been demonstrated with the current vla in a few extreme cases , although the limited bandwidth effectively precludes proper spectroscopy ( papadopoulos et al .
2001 ; carilli et al . 1999 , 2002 ) .
the co excitation conditions assumed in the models of section 2 could lead to pessimistic predictions for the low - order transitions because the velocity integrated line flux density increases roughly with the square of the frequency ( ie .
constant brightness temperature ) , at least to co(4 - 3 ) ( paper i ) .
while this appears to be roughly appropriate for infra - red selected galaxy samples , our models do not include the possibility of a population of molecular gas - rich , high redshift galaxies with lower excitation conditions .
for example , the co excitation conditions for the milky way disk inside the solar radius ( excluding the galactic center ) imply roughly equal velocity integrated flux density for co(1 - 0 ) and co(4 - 3 ) ( fixsen , bennett , & mather 1999 ) .
if such a population of galaxies exists , then the predictions from our models can be considered lower limits to the cm - wave source counts for molecular line surveys .
recent observations with the vla provide evidence that such a population may indeed exist ( papadopoulos et al.2001 ) .
limitations to the intrinsic brightness temperature of the thermal line emission from high redshift galaxies require that much of the collecting area of future large area radio telescopes be concentrated on baselines @xmath57 km . on the other hand ,
having baselines out to 10 km provides the important capability of imaging the emission on scales relevant to galaxies ( @xmath81 kpc ) , and for resolving the multiple images of the order of 1% of line - emitting galaxies expected to be gravitationally lensed by foreground galaxies .
lastly , we have found that hcn(1 - 0 ) emission will not be a major source of confusion to optimal cm - wave co line searches , comprising about 5@xmath1 of the total number of detected galaxies at 22 ghz and 1.5@xmath1 at 43 ghz . of course , once identified as such , hcn is interesting in it s own regard as a better tracer than co of the star forming clouds in active star forming galaxies ( solomon 2001 ) . like studies of star formation in our own galaxy ,
it has become clear that a complete census of the star - formation history of the universe requires an understanding of the contribution from galaxies that are obscured by dust at optical wavelengths .
the next generation mm- and cm - wave telescopes will provide unique , and complementary , capabilities for studying the thermal and non - thermal line and continuum emission from such systems at sub - arcsecond spatial resolution .
the national radio astronomy observatory ( nrao ) is operated by associated universities , inc . under a cooperative agreement with the national science foundation .
we thank dave frayer for further discussions about the line models implemented in paper i. * figure 1 * : cumulative source counts of 300-km - s@xmath10 wide co lines at 22ghz ( solid line ) and 43ghz ( dash line ) , developed from the results of blain et al .
( 2000 , paper i ) . a 4-ghz - wide band is assumed .
* figure 2 * : the detection rate of high-@xmath0 co emission lines for ` optimal surveys ' using the alma at 230ghz and the ska at 22 and 43ghz . the histograms delineate the contribution from different co transitions at different redshifts , as listed in each segment of the histogram .
note that the quoted redshifts are the mean values for the band covered . | we consider the capabilities for detecting low order co emission lines from high - redshift ( @xmath0 ) galaxies using the next generation of radio telescopes operating at 22 and 43ghz .
low order co emission studies provide critical insight into the nature of high redshift galaxies , including : ( i ) determining molecular gas masses , ( ii ) study of large scale structure through 3-dimensional redshift surveys over cosmologically relevant volumes , ( iii ) imaging gas kinematics on kpc - scales , and , in conjunction with observations of higher order transitions using future millimeter telescopes , ( iv ) constraining the excitation conditions of the gas .
particular attention is paid to the impact on such studies of the high frequency limit for future centimeter telescopes .
we employ models for the evolution of dusty star forming galaxies based on source counts at ( sub)millimeter ( mm ) wavelengths , and on the observed mm through infrared ( ir ) backgrounds , to predict the expected detection rate of low - order co(2 - 1 ) and co(1 - 0 ) line emitting galaxies for optimal centimeter(cm)-wave surveys using future radio telescopes , such as the square kilometer array ( ska ) and the expanded very large array ( evla ) .
we then compare these results to surveys that can be done with the next - generation mm - wave telescope , the atacama large millimeter array ( alma ) .
operating at 22ghz the ska will be competitive with the alma in terms of the detection rate of lines from high-@xmath0 galaxies , and will be potentially superior by an order of magnitude if extended to 43ghz .
perhaps more importantly , cm - wave telescopes are sensitive to lower excitation gas in higher redshift galaxies , and so provide a complementary view of conditions in high redshift galaxies to mm - wave surveys .
we have also included in our models emission from hcn .
the number of hcn(1 - 0 ) detections will be about 5@xmath1 of the co detections in the ( co - optimized ) 22 ghz surveys , and about 1.5@xmath1 for 43 ghz surveys . in order not to over - resolve the sources , brightness temperature limitations require that a future large area cm telescopes have much of its collecting area on baselines shorter than 10 km . |
the hamiltonian of the fermionic van hemmem ( fvh ) model with a magnetic transverse field @xmath0 is given by @xmath9 with the operators @xmath10 , \ \
\hat{s}_{i}^{x}=\frac{1}{2}[c_{i\uparrow}^{\dagger}c_{i\downarrow}+ c_{i\downarrow}^{\dagger}c_{i\uparrow}]\ ] ] and the random coupling @xmath4 given by @xmath11 , \label{jij}\ ] ] where @xmath12 and @xmath13 are random variables which follow a bimodal distribution @xmath14 .
\label { prob}\ ] ] in the present work the partition function is obtained within the lagrangian path integral formalism . in that case , the spin operators are represented as bilinear combinations of grassmann fields ( @xmath15 ) . particularly , the @xmath8-fvh model admits only magnetic states .
this restriction for the corresponding partition function is obtained by using the kronecker @xmath16 function ( @xmath17 @xmath18 @xmath19 ) @xcite .
this procedure allows to write the partition function for @xmath8 and @xmath7 situations in a compact form as @xmath20 and @xmath21 where @xmath22 ( @xmath23 or @xmath24 ) is the number of states per site , @xmath25\varphi_{j\sigma}(\tau ) \right .
\\ & - \left .
h(\varphi^{*}_{j\sigma}(\tau),\varphi_{j\sigma}(\tau ) ) \right\rbrace \label{eqz2 } , \end{split}\ ] ] @xmath1 is chemical potential , @xmath26 and @xmath27 or @xmath28 for the @xmath7-fvh and @xmath8-fvh models , respectively . the action @xmath29 given in eq .
( [ eqz1 ] ) can be build using the hamiltonian ( [ ham ] ) .
thus , we have : @xmath30 where @xmath31\underline{\varphi}_{j}(\tau ) , \label{eqao}\ ] ] @xmath32 @xmath33 and @xmath34 the matrices in eqs .
( [ eqao])-([eqs ] ) are defined as : @xmath35 ; ~ \underline{\sigma}^x=\left ( \begin{tabular}{cc } $ 0 $ & $ 1$\\ $ 1 $ & $ 0 $ \end{tabular } \right ) ; \ \ \underline{\sigma}^z=\left ( \begin{tabular}{cc } $ 1 $ & $ 0$\\ $ 0 $ & $ -1 $ \end{tabular } \right ) . \ \ \label{pauli}\ ] ] the coupling @xmath4 given in eq .
( [ jij ] ) allows to rewrite the random part of the action as : @xmath36^{2 } - \left [ { \displaystyle\sum_{j=1}^{n}}\eta_{j}s_{j}(\tau)\right ] ^2\right .
\\ -\left . \left
[ { \displaystyle\sum_{j=1}^{n}}\xi_{j}s_{j}(\tau)\right ] ^2 -2{\displaystyle\sum_{j=1}^{n}}\eta_{j}\xi_{j}(s_{j}(\tau))^2\right\ } \label{eq9}. \end{split}\ ] ] similarly , the ferromagnetic part of the action is given as @xmath37 ^2- { \displaystyle\sum_{j}}\left [ s_{j}(\tau)\right]^2\right\}. \label{eq101}\ ] ] the hubbard - stratonovich transformation is used to linearize the first three terms and the first one from the right side of eqs .
( [ eq9 ] ) and ( [ eq101 ] ) , respectively .
the last terms in the right side of eqs .
( [ eq9])-([eq101 ] ) vanish in the thermodynamic limit .
therefore , @xmath38 defined in eqs .
( [ eqz])-([eqs ] ) is obtained as : @xmath39\ } \label{eq11 } \end{split}\ ] ] with @xmath40 and @xmath41 . \label{eq20}\ ] ] in eq .
( [ eq11 ] ) , it has been used the self - averaging property @xmath42 .
the average @xmath43 can be achieved by using the probability distribution @xmath44 where @xmath45 ( or @xmath46 ) is defined in eq .
( [ prob ] ) . in this work ,
the sa is assumed , in which @xmath47 and @xmath48 for @xmath491 , 2 and 3 .
therefore , the internal field in eq.([eq20 ] ) is @xmath50 the integrals over \{@xmath51 } and @xmath52 in eq .
( [ eq11 ] ) can be solved by the saddle point method which gives @xmath53 @xmath54 @xmath55 and @xmath56 the fourier transform can be used in eq .
( [ eq19 ] ) and the functional integrals over the grassmann fields ( @xmath15 ) performed as well as the subsequent sum over matsubara s frequencies following close ref .
@xcite to give : @xmath57 \label{lambda}\ ] ] with @xmath58^{2 } + \gamma^{2}$ ] . for the @xmath8-fvh model
the integral over @xmath59 in eq .
( [ eq11 ] ) is calculated which leads to the same thermodynamics found in ref .
@xcite for the quantum vh model .
while , in the case of @xmath7-fvh model , @xmath60 can also be obtained from eq.([eq11 ] ) which gives for the grand canonical potential @xmath61 , the following expression : @xmath62\rangle\rangle_{\eta\xi } \label{potential } \end{split}\ ] ] where it is assumed @xmath63 .
in order to investigate the interplay between charge and spin fluctuations in the @xmath7-fvh model in the presence of a transverse magnetic field @xmath0 , a set of phase diagrams can be built from the numerical solution of the saddle point sg order parameter @xmath64 and the magnetization @xmath65 which is obtained from eq .
( [ potential ] ) .
it should be remarked that temperature , @xmath66 ( the strength of the ferromagnetic component of @xmath4 ) , @xmath1 and @xmath0 are given in units of @xmath67 ( the strength of the random component of @xmath4 ) . for numerical results ,
@xmath68 is used . in figs .
[ fig1]-[fig23 ] , phase diagrams temperature versus @xmath66 for several values of @xmath1 and @xmath0 are displayed .
[ fig1](a ) shows the situation corresponding to the half - filling occupation ( @xmath69 ) and @xmath70 . for that case
, the phase diagram of the classical vh model is basically recovered . for small @xmath66 , when the temperature is decreased , there is first a transition from the pm phase to the sg one which has @xmath71 and @xmath72 .
then , there is a second transition from the sg phase to the mixed ( fm+sg ) one in which @xmath73 and @xmath72 . for larger
@xmath66 , sg and fm+sg phases are replaced by the ferromagnetic ( fm ) one with @xmath73 and @xmath74 . in the inset of fig .
[ fig1](a ) , the @xmath7-fvh model phase diagram with @xmath69 is compared to the @xmath8-fvh one .
these two models present qualitatively the same results . however , the phase boundaries of the @xmath7-fvh model at the half - filling appear at lower temperatures than the @xmath8-fvh model .
that is a consequence of the non - magnetic states allowed in the @xmath7-fvh model . figs .
[ fig1](b)-[fig1](d ) show how the increase of @xmath1 can affect phase boundaries even when @xmath70 . when the average occupation @xmath75 moves from the half - filling ( @xmath69 ) to doubly occupied sites ( @xmath76 ) , the transition temperature @xmath77 between paramagnetism and sg phase as well as that one ( here called @xmath78 ) between sg and fm+sg phases are depressed
. ultimately , this depressing of @xmath77 and @xmath78 as @xmath1 increases leads to the complete suppression of sg and fm+sg . similarly to the fisg model ( see discussion in ref .
@xcite ) , the increase of @xmath1 also changes the nature of the phase transitions which appear in fig .
[ fig1 ] . that can be seen , for instance , for @xmath79 in fig .
[ fig1](b ) when @xmath80 .
furthermore , for @xmath76 ( see fig . [ fig1](d ) ) , the nature of the curie temperature @xmath81 is also changed .
it appears a tricritical point which can be located ( see appendix ) at @xmath82 with @xmath83 .
one last consequence due to the increase of @xmath1 is that larger values of @xmath66 are necessary to obtain a fm solution . in fig [ fig2 ] , @xmath84 while @xmath1 assumes the same values already used in fig .
therefore , the results can be compared directly to fig .
[ fig1 ] which allows to understand the combined effects of @xmath0 and @xmath1 .
the phase diagram for @xmath69 is shown in fig . [ fig2](a ) . in this case , the effect of @xmath0 is to depress both @xmath85 and @xmath79 .
however , the subsequent increase of @xmath1 showed in figs .
[ fig2](b)-[fig2](d ) leads to a distinct situation as compared with fig [ fig1 ] .
for instance , the fm+sg phase is completely suppressed .
in contrast , the sg phase is still preserved . other important difference in comparison to the case displayed in fig .
[ fig1 ] is concerned with the nature of phase boundaries .
for instance , the line transition @xmath78 remains always a second order one . that is not the case for @xmath77 which becomes a first order one as displayed in fig .
[ fig2](d ) where @xmath76 .
in fact , for this value of chemical potential there is a completely new scenario .
the existence of the tricritical point in the curie temperature is preserved being located at @xmath86 .
nevertheless , there is now an additional triple point @xmath87 with the coexistence of sg , fm and pm phases .
[ figpot ] exhibits the grand canonical potential @xmath61 as a function of temperature for the sg , pm and fe phases near the triple point .
the first order transitions are located by comparing @xmath61 of these phases , where the stable solution is always that one with minimum @xmath61 .
particularly , the triple point is obtained when @xmath61 of the sg , pm and fe phases assume the same values ( see fig . [ figpot](c ) ) in fig .
[ fig23 ] phase diagrams with @xmath88 are displayed . in that case , the fm+sg is completely suppressed for @xmath69 ( see fig . [ fig23](a ) ) .
the remaining phase diagrams displayed in figs .
[ fig23](b ) -[fig23](d ) present only sg , fm and pm solutions for the order parameters .
when @xmath80 and @xmath89 , the second order line transition @xmath85 is even more diminished when compared to the respective previous situations given in figs .
[ fig1]-[fig2 ] . for @xmath76 , the complex scenario found in fig .
[ fig2](d ) is repeated in fig .
[ fig23](d ) with the presence of tricritical and triple points .
however , interestingly , @xmath77 , which is now a first order line transition , has a small increase as compared with fig [ fig2](d ) .
the effects of the @xmath0 increasing on the @xmath8-fvh model can be observed in the inset of figs . [ fig2](a ) and [ fig23](a ) , which exhibit phase diagrams of the @xmath8-fvh model compared to the @xmath7 one with @xmath69 for @xmath84 and 1.0 , respectively . as in the @xmath7-fvh model ,
the magnetic phase is also suppressed by @xmath0 in the @xmath8-fvh model . in addition , the phase boundaries converge to the same values when temperature decreases .
the present work has studied the fermionic version of the classical van hemmen model @xcite for sg in the presence of a transverse magnetic field @xmath0 .
the goal is to verify how the interplay between charge and spin fluctuations can affect the magnetic ordering in this particular model .
it should be mentioned that the knowledge of such interplay can also be quite useful in further applications for studying the disordered magnetism when other physical processes are present ( see , for instance , ref .
@xcite ) .
the mentioned fermionic version has been formulated in two versions . in the first one , the so called @xmath8-fvh , the two nonmagnetic eigenstates of the operator @xmath90 ( defined in eq .
( [ eqs ] ) have been eliminated .
therefore , the corresponding thermodynamics obtained is exactly the same one obtained by a previous quantum formulation of the van hemmen model given in ref .
@xcite . on the contrary , when the four original eigenstates are preserved ,
the solution for both saddle point order parameters , the sg ( @xmath64 ) and the magnetization ( @xmath65 ) one , indicates that not only the phase diagram but also the nature of the phase transitions are deeply affected when the chemical potential @xmath1 and/or @xmath0 are varied .
for instance , the mixed phase fm+sg can be suppressed as well as there is onset of tricritical and triple points by increasing @xmath1 and/or @xmath0 .
in fact , the role of @xmath0 as responsible for the suppression of the fm+sg phase has already been found in ref .
remarkably , in the present problem , the same result can be obtained by varying only @xmath1 , or in other words , only by changing the average occupation of sites @xmath75 .
therefore , one can say that the thermodynamics derived from van hemmen model is either sensitive to a spin flipping mechanism provided by @xmath0 as well a dilution one provided by @xmath1 .
surely , when both mechanisms are combined , the complexity of the phase diagram is even more intense as can be illustrated by a first order phase transition between the pm and sg ( or fe ) phases with the presence of tricritical and triple points . to conclude ,
we have shown in this work that mechanisms such as dilution and quantum spin flipping given by @xmath1 and @xmath0 , respectively , are sources of quite non - trivial effects in the fermionic van hemmen model .
more precisely , those effects indicate that the existence of magnetic phases and the nature of its boundaries in that model are extremely sensitive to the redistribution of charge provided by @xmath1 and @xmath0 .
it should be emphasized that thermodynamics in the present work could be derived without the use of sophisticated mathematical techniques such as the replica method .
it is well known that in the case of sg first order phase transitions , the use of this method implies in considerable difficulties as , for example , location of first order phases boundary @xcite .
this work was partially supported by the brazilian agencies cnpq and capes .
the landau expansion of the grand canonical potential ( eq .
( [ potential ] ) ) in powers of @xmath65 and @xmath64 is given by @xmath91 where @xmath92 @xmath93 and @xmath94 the second order phase transition from the paramagnetic phase to the sg one ( to the fe phase ) occurs when @xmath95 ( @xmath96 ) and @xmath97 .
the tricritical point is located when @xmath96 with @xmath98 .
e. dagotto , science 309 ( 2005 ) 257 - 262 .
m. brando , j. hemberger , t. westerkamp , m. deppe , r. kuchler , m. brando , c. geibel , p. gegenwart , a. p. pikul , f. steglich , phys .
102 ( 2009 ) 206404 .
n. marcano , j. c. gomez sal , j. i. espeso , j. m. de teresa , p. a. algarabel , c. paulsen , j. r. iglesias , phys .
98 ( 2007 ) 166406 .
b. coqblin , m. d. nunez - regueiro , a. theumann , j. r. iglesias , s. g. magalhaes , philosophical magazine 86 ( 2006 ) 2576 - 2580 .
a. theumann , b. coqblin , s. g. magalhaes , a. a. schmidt , phys .
b 63 ( 2001 ) 054409 . s. g. magalhaes , a. a. schmidt , a. theumann , b. coqblin , eur . phys
. j. b 30 ( 2002 ) 419 - 426 .
a. theumann , b. coqblin , phys .
b 69 ( 2004 ) 214418
. s. g. magalhaes , f. m. zimmer , p. r. krebs , b. coqblin , phys .
b 74 ( 2006 ) 014427 .
s. g. magalhaes , f. m. zimmer , b. coqblin physica b 403/5 ( 2008 ) 1395 - 1397 .
s. kirkpatrick , d. sherrington , phys .
b 17 ( 1978 ) 4384 - 4403 .
d. j. amit , modelling brain function .
the world of attractor neural networks , cambridge university press , cambridge , england , 1989 .
j. l. van hemmen , phys .
49 ( 1982 ) 409 - 412 .
s. g. magalhaes , f. m. zimmer , b. coqblin , phys .
b 81 ( 2010 ) 094424 . s. g. magalhaes , a. theumann , eur .
j. b vol 9 ( 1999 ) 5 - 14 .
a. theumann , a. a. schmidt , s. g. magalhaes , physica a 311 ( 2002 ) 498 - 506 .
w. wiethege , d. sherrington , j. phys .
c 19 ( 1986 ) 6983 - 6991 . k. binder , a. p.
young , rev .
mod . phys .
58 ( 1986 ) 801 - 976 .
a. j. bray , m. a. moore , j. phys .
c 13 ( 1980 ) l655-l660 .
a. theumann , m. v. gusmao , phys .
lett . a 105 ( 1984 ) 311 - 314 .
r. oppermann , a. muller - groeling , nucl .
b 401 ( 1993 ) 507 - 547 .
b. rosenow , r. oppermann , phys .
( 1996 ) 1608 - 1611 .
s. g. magalhaes , c. v. morais , f. m. zimmer , phys .
b 77 ( 2008 ) 134422
. s. g. magalhaes , c. v. morais , f. m. zimmer , phys . rev .
b 81 ( 2010 ) 014207
. s. k. ghatak , d. sherrington , j. phys .
c 10 ( 1977 ) 3149 - 3156 .
f. a. da costa , c. s. o. yokoi , s. r. a. salinas , j. phys .
a 27 ( 1994 ) 3365 - 3372 .
t. c. choy , d. sherrington , j. phys .
c : solid state phys .
17 ( 1984 ) 739 - 745 . j. l. van hemmen , j. phys .
c : solid state phys . 19 ( 1986 )
j. roberto viana , yamilles nogueira , j. ricardo de souza , phys .
b 66 ( 2002 ) 113307
. s. g. magalhaes , f. m. zimmer , c. v. morais , physica a 388 ( 2009 ) 2140 - 2148 .
p. mottishaw , d. sherrington , j. phys .
c 18 ( 1985 ) 5201 - 5213 .
p. mottishaw , europhys .
lett . 1 ( 1986 ) 409 - 414 . | in the present work it is studied the fermionic van hemmen model for the spin glass ( sg ) with a transverse magnetic field @xmath0 . in this model , the spin operators are written as a bilinear combination of fermionic operators , which allows the analysis of the interplay between charge and spin fluctuations in the presence of a quantum spin flipping mechanism given by @xmath0 .
the problem is expressed in the fermionic path integral formalism . as results , magnetic phase diagrams of temperature versus the ferromagnetic interaction
are obtained for several values of chemical potential @xmath1 and @xmath0 .
the @xmath0 field suppresses the magnetic orders .
the increase of @xmath1 alters the average occupation per site that affects the magnetic phases .
for instance , the sg and the mixed sg+ferromagnetic phases are also suppressed by @xmath1 . in addition
, @xmath1 can change the nature of the phase boundaries introducing a first order transition .
quantum spin glass , fermionic model , charge fluctuation it is well established that disorder can be a source of non - trivial effects in condensed matter physics . from anderson localization to spin glass ,
there are plenty of examples supporting such role .
as one more recent example there are indications that the presence of disorder is responsible by several complex effects in strongly correlated systems as , for instance , manganites or cerium compounds ( see ref . @xcite and references therein ) .
the important point is that disorder can affect the charge sector as well as the spin one in such way that it becomes a constant cause of new interesting problems . from the theoretical point of view , the task of facing this kind of problems is surely a source of new approaches .
recent experimental findings in two specific cerium compounds can illustrate the last comment in the previous paragraph .
@xmath2 @xcite and @xmath3 @xcite are physical systems in which disorder combined with rkky and kondo interaction is responsible by the onset of a complex scenario .
their phase diagrams display a glassy state , ferromagnetism and a region dominated by the kondo effect .
for @xmath3 , several works have used the so called kondo - ising lattice ( kil ) model @xcite to reproduce the experimental phase diagram of that compound .
the ising part of the model represents an intersite interaction between localized f - spin operators .
it is precisely in the spin - spin ising coupling @xmath4 that various types of randomness have been tested trying to represent the effects of the disorder in the @xmath5 @xcite .
for instance , the random coupling @xmath4 has been chosen as the sherrington - kirkpatrick ( sk ) model @xcite or as the hopfield model @xcite . however ,
none of the choices previously mentioned were entirely satisfactory to describe how the glassy state in the @xmath5 is replaced by an inhomogeneous ferromagnetism when the temperature is decreased @xcite .
very recently , a new approach has been proposed for the problem discussed above , this approach uses the random coupling @xmath4 as that one introduced in the van hemmen ( vh ) model @xcite .
this model was originally conceived to study the spin glass ( sg ) problem with classical ising spin variables .
the results indicated that the vh type of randomness in the kil model is more adequate than those ones used in sk or hopfield models to reproduce some aspects of the experimental phase diagram of the @xmath5 @xcite .
for instance , the previously mentioned ferromagnetic phase appears below the spin glass one .
the improvement obtained using the coupling @xmath4 given as the vh model to study the @xmath5 suggests the usefulness of this kind of randomness could be more general .
in other words , it could also be used when it is present in other quantum process than the kondo effect . in this case , models with one type of fermions could be built by adding to the vh ising term , for instance , a hopping , a pairing coupling @xcite or a transverse magnetic field .
nevertheless , surprisingly little consideration has been given to a fermionic version of the vh model .
in fact , even less consideration is given to the fermionic vh model when a term capable to introduce quantum dynamics in the problem is added to this model .
in fact , it lacks an investigation of the interplay of charge and spin fluctuations within a fermionic version of the vh model even for a simple quantum process provided by a transverse magnetic field . the purpose of the present work is to study the vh model in the presence of a transverse field @xmath0 in a fermionic formulation which means that the ising spin operators are written as bilinear combinations of creation and destruction fermionic operators . for this particular representation of the spin operators ,
the natural tool to obtain the thermodynamics is the functional integral formalism with grassmann fields .
a peculiarity of this representation is that @xmath6 acts in a space with four eigenstates . among them , there are two which are nonmagnetic @xcite . therefore , we calculate the grand partition function for two situations . in the first one , here called @xmath7 fermionic van hemmen ( @xmath7-fvh ) model , the four states are preserved .
for this particular model , the chemical potential @xmath1 appears as an important parameter in the problem . in the second one , the @xmath8 fermionic van hemmen ( @xmath8-fvh ) model , introduces a restriction , what allows considering only the two magnetic states @xcite .
one of the most important advantages of the vh model in both versions 4s and 2s is that the disorder can be treated without the use of the replica method .
therefore , the subsequent difficulties associated with this technique @xcite can be avoided in the present work ( see discussion below ) .
however , the problem is still treated within the static approximation ( sa ) in which the time fluctuations are neglected @xcite . indeed , the fermionic representation of ising spin operators for the study of the sg problem is well established . for the sk type of randomness , the so called fermionic ising spin glass ( fisg ) model @xcite shows an interesting characteristic .
charge and spin correlation functions are connected .
the average occupation of fermions per site and the replica diagonal element of the sg order parameter matrix are directly related .
this relationship implies that , for instance , the nature of the pm / sg phase boundary can be modified by variations of chemical potential with the onset of a tricritical point @xcite .
the presence of an additional transverse magnetic field @xmath0 in the fisg model also introduces important changes in the pm / sg phase boundary .
the first order part of that phase boundary is diminished when @xmath0 increases @xcite .
consequently , the position of the tricritical point is also affected .
therefore , one could expect that in the 4s - fvh model , variations of @xmath1 could also bring important consequences on the location and nature of phase boundaries .
it should be remarked that the location of a first order pm / sg phase boundary is not a trivial task for models in which the replica method is used as the fisg one @xcite .
this difficulty has already been a subject of controversy for the classical ghatak - sherrington model ( see , for instance , refs .
@xcite ) .
it is clear that this kind of controversy would not be needed in the fvh model . on the other hand , vh model ( classical or fermionic )
has a shortcoming .
it lacks to present a multiplicity of metastable - states @xcite .
nevertheless , the vh model produces some results which are close to those ones found experimentally in sg physical systems ( see ref .
@xcite ) .
thus , the proper understanding of the vh model in a fermionic formulation can be important not only as a sg problem itself but also as something useful for applications as pointed out above and very recently discussed in ref .
@xcite .
it should also be mentioned that the vh model with an additional transverse magnetic field @xmath0 has already been studied . however , in a formulation where the classical ising spin variables were replaced by pauli matrices @xcite . in this formulation
one can observe important consequences for the phase diagram produced by the presence of @xmath0 .
the obtained phase diagram shows that the increase of @xmath0 destroys the sg , the ferromagnetic ( fm ) and the mixed ( fm+sg ) phases found in the classical phase diagram .
nevertheless , this kind of formulation is not able to capture the consequences for the phase diagram when both charge and spin fluctuations are affected by @xmath0 .
this paper has the following structure . in the second section ,
the model is introduced and the grand canonical potential is derived . in the third section ,
several phase diagrams are constructed with particular attention for the effects of @xmath0 and @xmath1 by solving the equations for the order parameters .
the last section is dedicated to the conclusions . |
one of important problems in the supersymmetric standard model ( susy - sm ) is how a susy higgs mass term , the @xmath0-term , is derived naturally within the framework of supergravity ( sugra ) . because the @xmath0-term is not a soft susy breaking term and a natural order of such a mass @xmath0 is the gravitational scale @xmath5 in sugra . ] .
this problem is called the @xmath0-problem @xcite
. some types of natural mechanisms of @xmath0-term generation have been proposed , e.g. the case with khler potential including a higgs mixing term @xcite and a superpotential including a suitably suppressed mass term @xcite .
in addition to the supersymmetric mass term , soft susy breaking mass terms also contribute to the higgs mass terms .
explication of susy breaking mechanism is an important and unsolved problem .
superstring theory ( sst ) is only the known candidate for unified theory including gravity and is expected to give a natural solution due to some non - perturbative effects .
recently there have been various remarkable developments in studying non - perturbative aspects of sst @xcite , but susy breaking mechanism has not been fully understood yet . for the present
, we can derive formulae for soft susy breaking terms assuming the existence of nonperturbative superpotential which induces susy breaking in string models @xcite .
further these soft terms can be parametrized simply using several parameters in most cases .
for example , under the assumption that only the dilaton field @xmath6 and/or the overall moduli field @xmath7 trigger susy breaking by their @xmath1-term condensations , soft scalar masses , gaugino masses and @xmath8-parameters are written model - independently by the gravitino mass @xmath9 and the goldstino angle @xmath10 in the case with the vanishing vacuum energy @xcite .
we can extend such a parametrization into cases with several moduli fields @xcite .
on the other hand , the @xmath11-parameter depends on the @xmath0-term generation mechanisms @xcite .
therefore @xmath0 and @xmath11 parameters are usually treated as arbitrary parameters .
the more arbitrary parameters exist , the harder we analyze and the less we have a predictivity .
it is important to select @xmath0-term generation mechanism in some way . in this paper , we study several @xmath0-term generation mechanisms based on the radiative breaking scenario of electroweak symmetry breaking within the framework of string theory .
examining the parameter space leading to successful electroweak symmetry breaking , we probe a realistic @xmath0-term generation mechanism . we take into account @xmath2-term contributions to soft scalar masses @xcite in our analysis .- term contribution @xcite and susy grand unified theory ( susy - gut ) with @xmath2-term contributions @xcite . ]
the reason is as follows .
string models , in general , have extra gauge symmetries including an anomalous @xmath12 symmetry and extra massless matter fields other than the susy - sm ones .
we can reduce gauge group and massless spectrum through symmetry breaking along flat directions @xcite . as discussed in recent papers @xcite , sizable @xmath2-term contributions can appear through such flat direction breaking .
this paper is organized as follows . in the next section ,
we give our basic assumptions , formulae of @xmath0-term and @xmath11-term , and our strategy . in section 3 ,
we study several types of @xmath0-term generation mechanisms based on the radiative breaking scenario and show which type is hopeful .
section 4 is devoted to conclusions and discussions .
we take the susy - sm with soft susy breaking parameters expected to be derived as a low - energy theory from orbifold models through flat direction breaking as a starting point of our analysis .
let us first list our basic assumptions . 1 .
the susy is broken by the @xmath1-term condensations of dilaton field @xmath6 and/or moduli fields @xmath13 .
2 . the vacuum energy @xmath14 vanishes , i.e. , @xmath15 , at the tree level .
3 . for the third family of the quark doublet @xmath16 , the up - type singlet @xmath17 and the higgs doublet @xmath18
, their modular weights are @xmath19 .
4 . the top yukawa coupling @xmath20 does not depend on moduli fields , i.e. , @xmath21 .
the modular weight of the higgs doublet @xmath22 is @xmath23 ( or @xmath24 ) .
the product @xmath25 of higgs doublets is gauge invariant under all gauge symmetries which are broken along a flat direction .
7 . effects of moduli - dependent threshold corrections for gauge couplings and gaugino masses are negligibly small .
the kac - moody levels satisfy @xmath26 .
the gaugino masses @xmath27 , @xmath11 and @xmath0 are all real . 10 .
the dependence of dilaton and moduli fields is small in the @xmath0-parameter , i.e. , @xmath28 .
the third and fourth assumptions can be justified from the fact that the coupling @xmath20 is strong and allowed as a renormalized coupling in the untwisted sector . here
we discuss the overall - moduli case ( @xmath29 ) for simplicity . under the above assumptions ,
we can obtain the following formulae of soft scalar masses @xmath30 , the @xmath8-parameter @xmath31 among @xmath32 , @xmath33 and @xmath18 , and the gaugino mass @xmath34 , @xmath35 where we use the following parameterization @xmath36 here @xmath37 is a @xmath2-term contribution and we get the relation @xmath38 from the assumption 6 and the fact that its contribution is proportional to broken diagonal charges of scalar field .
the natural order of @xmath39 is expected to be @xmath40 from some explicit models @xcite . in the limit of the moduli dominant susy breaking , we can not neglect moduli - dependent threshold corrections , but we do not consider such a case for simplicity in this paper .
the top quark mass is given as @xmath41 .
\nonumber\end{aligned}\ ] ] ] where @xmath42 and @xmath43 .
( @xmath44 s are neutral components of @xmath45 s . )
we use the value @xmath46gev as the top quark mass from the current experiments . through our analysis , we take the gauge coupling unification scale @xmath47gev as an energy scale where boundary conditions are imposed . hence it is supposed that the above formulae ( [ smass0])([gmass0 ] ) hold at @xmath48 . by the use of renormalization group equations ( rges ) of the susy - sm @xcite , soft scalar masses of @xmath22 and @xmath18 at the weak scale ( we use the @xmath49-boson mass @xmath50 )
are given as , @xmath51 where @xmath52 and @xmath53 . here the second term in rhs of ( [ barh ] ) represents the effect of the top yukawa coupling and @xmath54 is a function of @xmath55 ( or the top yukawa coupling ) .
the values of @xmath56 and @xmath54 are given in table 1 .
several types of solutions for the @xmath0-problem have been proposed @xcite . here
we explain some of them briefly and we give formulae of the @xmath11-parameter for each @xmath0-term generation mechanism .
( @xmath0 - 1 ) the @xmath0-term @xmath57 of @xmath58 appears after susy breaking in the case where a khler potential includes a term such as @xmath59 @xcite . in this case
we have @xmath60 hereafter we take @xmath61 and then @xmath57 and @xmath62 are given as @xmath63 ( @xmath0 - 2 ) the @xmath0-term @xmath64 of @xmath58 appears after susy breaking in the case where a superpotential @xmath65 includes a term such as @xmath66 @xcite . here
@xmath67 is a superpotential which induces susy breaking . in this case
we have @xmath68 ( @xmath0 - 3 ) the @xmath0-term @xmath69 can be generated through some non - perturbative effects such as gaugino condensation @xcite and it generally depends on the vevs of @xmath6 and @xmath7 . in this case
we have @xmath70 ( @xmath0 - 4 ) in the model with a singlet field @xmath71 which has a coupling @xmath72 , the @xmath0-term appears when the field @xmath71 develops its vev at some lower energy scale near @xmath58 @xcite . as a result , the @xmath11-parameter can be generated from the @xmath8-term @xmath73 .
there can be an admixture of several @xmath0-term generation mechanisms and , in this case , @xmath0 and @xmath11 parameters are given as @xmath74 where the indices @xmath75 and @xmath76 run over all @xmath0-term generation mechanisms .
finally we discuss a multi - moduli case briefly .
@xmath77 and @xmath78 orbifold models @xcite have @xmath79-type of moduli fields corresponding to complex structures of orbifolds @xcite and a mixing term in the khler potential as @xmath80 in this case , the higgs fields @xmath22 and @xmath18 belong to the untwisted sector .
we assume @xmath1-terms of @xmath6 , @xmath81 @xmath82 and @xmath83 contribute susy breaking and these are parametrized by @xmath9 , @xmath10 , @xmath84 and @xmath85 following refs.@xcite .
then the soft scalar masses , the @xmath0-term and @xmath11-term are written at the tree level as @xcite @xmath86 it is easy to show that @xmath87 and @xmath88 reduce to @xmath62 and @xmath57 , respectively if we set @xmath89 and @xmath90 .
we give formulae of @xmath91 and @xmath92 in table 2 . here
we choose @xmath93 , @xmath94 and @xmath95 and use this choice in our analysis . in general , the @xmath92 contains a small number of free parameters compared with @xmath91 and so the analysis of @xmath92 can be more predictable . in this subsection , we give an outline of our strategy to probe @xmath0-term generation mechanism based on the radiative electroweak symmetry breaking scenario . the neutral fields @xmath96 and
@xmath97 have the following potential @xcite : @xmath98 where all parameters correspond to the values at @xmath50 .
the condition for the symmetry breaking is given as @xmath99 the bounded from below ( bfb ) condition along the @xmath2-flat direction requires @xmath100 further the conditions that minimizing the potential are given as @xmath101 where we use the relation @xmath102 .
by the use of stationary conditions ( [ mini1 ] ) and ( [ mini2 ] ) , @xmath0 and @xmath11 are expressed by other parameters ( @xmath9 , @xmath103 , @xmath55,@xmath104 ) , such that @xmath105 by the use of rges , we can obtain the @xmath0 and @xmath11 parameters at @xmath48 ( they are denoted by @xmath106 and @xmath107 , respectively . ) as follows @xcite , @xmath108 and @xmath109 where @xmath110 here @xmath111 and @xmath112 are the gauge coupling and top yukawa coupling at @xmath48 , respectively . by comparing @xmath113 and
@xmath114 derived from the stationary conditions of radiative breaking with @xmath115 and @xmath116 ( @xmath117 ) given in the last subsection , we can find allowable parameter regions for ( @xmath9 , @xmath103 , @xmath84 , @xmath85 , @xmath55 , @xmath118 , @xmath119 , @xmath69 ) leading to successful electroweak symmetry breaking and know which type of @xmath0-term generation mechanism is hopeful .
our strategy , @xmath120bottom - up approach , is more generic and applicable than the usual one , @xmath120top - down approach , where a realization of the radiative breaking scenario is examined by defining the parameters @xmath121 , @xmath0 and @xmath11 at @xmath48 and checking if the quantities renormalized at @xmath50 satisfy stationary conditions ( [ mini1])([mini2 ] ) .
in this section , we examine which type of @xmath0-term generation is hopeful by taking the cases ( @xmath0 - 1)(@xmath0 - 3 ) as examples and comparing @xmath113 and @xmath114 with @xmath115 and @xmath116 determined by @xmath0-term generation mechanism .
we also give remarks of several extensions . for our analysis , it is convenient to define the following functions , @xmath122 the @xmath123 and @xmath124 should be satisfied the conditions @xmath125 and @xmath126 by definition if we assume the radiative breaking of electroweak symmetry and the @xmath75-th type of @xmath0-term generation mechanism .
first we give allowed regions for @xmath113 and @xmath114 in the limit of dilaton dominant susy breaking in fig.1 and 2 , respectively .
the ranges of @xmath127 and @xmath128 are @xmath129 and @xmath130 .
the ranges of @xmath131 and @xmath132 are @xmath133 and @xmath134 . here
we take @xmath135 , it is necessary to incorporate the contribution of yukawa couplings other than top quark neglected here . ] , @xmath136 and @xmath137gev .
the value @xmath138 ( @xmath139 ) corresponds to @xmath140 ( @xmath141 ) , @xmath142 ( @xmath143 ) and @xmath144 ( @xmath145 ) .
the inequality @xmath137gev is derived from the phenomenological constraint @xmath146gev when @xmath147 .
the fig.3 shows an allowable region of ( @xmath113 , @xmath118 ) in the limit of @xmath148 . in this way , we come to a conclusion that the @xmath0-term generation mechanism can be realistic if the values of @xmath149 and @xmath150 hit the above ranges .
we study the first case ( @xmath0 - 1 ) .
it is shown that there is no region satisfying @xmath151 with @xmath136 from fig.1 and @xmath152 . when @xmath153 , there is a solution of @xmath151 . however , in this case , it is not realistic since the top yukawa coupling blows up below @xmath48 .
this result is consistent with those in ref.@xcite .
further we find no region satisfying @xmath154 with @xmath136 from in fig.2 and @xmath155 .
as discussed in ref.@xcite , @xmath2-term contribution can survive even in the limit of dilaton dominant susy breaking if string model contains an anomalous @xmath12 symmetry which is cancelled by the green - schwarz mechanism @xcite . on the other hand ,
@xmath2-term contributions related to anomaly - free symmetries vanish at the tree level in the limit of dilaton dominant susy breaking .
if we assume the existence of @xmath2-term contribution , there appears a region consistent with the condition @xmath151 .
however it is a very narrow region with a relatively large positive value of @xmath118 .
for example , the region with @xmath156 and @xmath157 is allowed as given in fig.3 . on the other hand ,
negative @xmath2-term contribution is needed to lower the value of @xmath131 .
hence it is impossible to realize these @xmath0-term generation even with @xmath2-term contribution . in the second case ( @xmath0 - 2 )
, we get the same result for @xmath11-parameter as the first one . on @xmath0-parameter
, we can estimate the value of @xmath119 using the condition @xmath158 .
hence the radiative breaking scenario can be realized in the models with a relatively large positive @xmath2-term contribution , i.e. , @xmath159 . in the third case ( @xmath0 - 3 )
, we have a solution even in the absence of @xmath2-term contribution .
the allowable region of ( @xmath160 ) is given in fig.4 .
the favorable value is @xmath161 with @xmath136 .
the introduction of @xmath2-term contribution yields @xmath162 .
the @xmath0-parameter is treated as a free parameter as well as the second one since the origin is unknown .
we consider the effect of overall moduli @xmath1-term condensation .
we have the same qualitative result for @xmath163 as the dilaton dominant case .
that is , there is no allowed region in the first case , but very narrow region exists with large positive @xmath118 in the second case and there exists an allowed region with natural values of ( @xmath9 , @xmath55 , @xmath118 ) in the third case .
the above fact can be understood by the use of the eqs .
( [ mu])([tz ] ) directly
. that is , both @xmath114 and @xmath113 are proportional to @xmath164 in the absence of @xmath2-term contribution and the limit @xmath165 , and so they decrease as @xmath166 increases .
hence it is impossible to satisfy @xmath167 in the first and second case with @xmath136 .
the introduction of @xmath118 does not improve the situation drastically .
we discuss a multi - moduli case ( [ multi - h])([multi - b ] ) .
for the case with @xmath89 and @xmath90 , we have no allowable regions for @xmath167 and @xmath168 because this case corresponds to the first case ( @xmath0 - 1 ) .
we can check that there exists an allowable region with @xmath169 , @xmath170 and @xmath171 .
we can carry out the case with @xmath172 .
for @xmath169 and @xmath136 , we find the following fact .
compared with the case with @xmath95 , the value of @xmath127 and @xmath128 decreases and increases , respectively and the absolute value of @xmath114 decreases .
the difference between the values in the case with @xmath95 and @xmath172 increases as @xmath166 increases .
hence the similar conclusion holds for the reality of the radiative breaking scenario as the dilaton dominant case with @xmath95 . in the case of an admixture of several @xmath0-term generation mechanisms , we need a dominant contribution of the third mechanism ( @xmath0 - 3 ) to get an allowable region with natural values of ( @xmath9 , @xmath103 , @xmath173 , @xmath118 ) .
we discuss the case ( @xmath0 - 4 ) .
as we have an extra light singlet field @xmath71 , the rg flows of @xmath174 and @xmath175 should be modified owing to the effect of the yukawa coupling @xmath176 .
this case can be applied to a similar strategy discussed in the last section .
the difference is that @xmath107 receives rge effects as an @xmath8-parameter above @xmath50 and so we must compare the renormalized quantity with not @xmath11-parameter but @xmath177 at @xmath48 .
it is not discussed here further because the renormalized quantity contains an unknown parameter @xmath176 .
we have given a generic method to select a realistic @xmath0-term generation mechanism based on the radiative electroweak symmetry breaking scenario and studied which type is hopeful within the framework of string theory .
the @xmath0-term generated by some non - perturbative effects , i.e. , ( @xmath0 - 3 ) , can be hopeful to realize the radiative symmetry breaking scenario even in dilaton dominant supersymmetry breaking .
we have discussed effects of the moduli @xmath1-term condensation and @xmath2-term contribution to soft scalar masses . in the case of overall moduli
, we have the same qualitative result as in the limit of dilaton dominant susy breaking , that is , the first mechanism is impossible to realize the radiative scenario , the second one is required to a large @xmath2-term contribution of @xmath178 and the third one is hopeful .
our method to select a realistic @xmath0-term generation mechanism is so generic and powerful that we can apply it to the case with an improvement of approximation and more complex situations .
for example , the improvement by the incorporation of 1-loop effective potential @xcite , the case with a large @xmath55 , the case with large moduli - dominant threshold corrections for gaugino masses , other assignments of modular weight for matter fields and the modular dominant susy breaking case . and
@xmath11 are studied in the multi - moduli case with a large non - universality between higgs masses . ] in the above situations , extra contributions @xmath179 and @xmath180 are added to eqs.([h ] ) and ( [ barh ] ) , respectively .
these studies have to be considered systematically to select a realistic string model .
the authors are grateful to s. khalil for useful discussions . 99 j.e . kim and h.p .
nilles , phys . lett .
* b138 * ( 1984 ) 150 .
giudice and a. masiero , phys . lett .
* b206 * ( 1988 ) 480 .
nilles , m. srednicki and d. wyler , phys .
b120 * ( 1983 ) 346 ; + j.m .
frere , d.r.t .
jones and s. raby , nucl . phys .
* b222 * ( 1983 ) 11 ; + j.p .
derendinger and c. savoy , nucl .
b237 * ( 1984 ) 307 .
kim and h.p .
nilles , phys .
b263 * ( 1991 ) 79 ; + e.j .
chun , j.e .
kim and h.p .
nilles , nucl .
phys . * b370 * ( 1992 ) 105 .
casas and c. muoz , phys .
* b306 * ( 1993 ) 288 .
see e.g. e. witten , nucl .
b433 * ( 1995 ) 85 ; nucl . phys . *
b460 * ( 1996 ) 541 ; + n. seiberg and e. witten , ru-96 - 12 , hep - th/9603003 .
ibez and d. lst , nucl . phys . *
b382 * ( 1992 ) 305 .
kaplunovsky and j. louis , phys . lett .
* b306 * ( 1993 ) 269 .
a. brignole , l.e .
ibez and c. muoz , nucl . phys . *
b422 * ( 1994 ) 125 .
t. kobayashi , d. suematsu , k. yamada and y. yamagishi , phys .
b348 * ( 1995 ) 402 .
a. brignole , l.e .
ibez , c. muoz and c. scheich , ftuam 95/26 , hep - ph/9508258 .
j. louis and y. nir , nucl .
* b447 * ( 1995 ) 18 .
m. drees , phys .
b181 * ( 1986 ) 279 ; + j.s .
hagelin and s. kelley , nucl . phys .
* b342 * ( 1990 ) 95 ; + a.e .
faraggi , j.s .
hagelin , s. kelley and d.v .
nanopoulos , phys . rev . *
d45 * ( 1992 ) 3272 ; + y. kawamura , h. murayama and m. yamaguchi , phys . rev . * d51 * ( 1995 ) 1337 ; + y. kawamura , phys .
* d53 * ( 1996 ) 3779 ; prog .
* 123 * ( 1996 ) 421 .
r. barbieri , j. louis and m. moretti , phys .
* b312 * ( 1993 ) 451 , + a. brignole , l.e .
ibez and c. muoz , cern - th/96 - 183 , hep - ph/9607405 .
c.kolda and s.p .
martin , phys . rev . * d53 * ( 1996 ) 3871 .
a. font , l.e .
ibez , h.p .
nilles and f. quevedo , nucl .
phys . * b307 * ( 1988 ) 109 ; + j.a .
casas , e.k .
katehou and c. muoz , nucl . phys .
* b317 * ( 1989 ) 171 ; + y. kawamura and t. kobayashi , nucl . phys .
* b481 * ( 1996 ) 539 .
y. kawamura and t. kobayashi , phys .
* b375 * ( 1996 ) 141 ; preprint , dpsu-96 - 11 , ins - rep-1153 , hep - ph/9608233 ; + y. kawamura , t. kobayashi and t. komatsu , preprint , dpsu-96 - 12 , ins - rep-1161 , hep - ph/9609462 , to appear in phys . lett . * b*. j.p .
derendinger , l.e .
ibez and h.p .
nilles , phys .
* b155 * ( 1985 ) 65 ; + m. dine , r. rohm , n. seiberg and e. witten , phys . lett .
* b156 * ( 1985 ) 55 .
k. inoue , a. kakuto , h. komatsu and s. takeshita , prog .
* 68 * ( 1982 ) 927 ; * 71 * ( 1984 ) 413 ; + l.e .
ibez , phys . lett .
* b118 * ( 1982 ) 73 ; nucl . phys .
* b218 * ( 1983 ) 514 ; + l. alvarez - gaum , j. polchinski and m. wise , nucl . phys . *
b221 * ( 1983 ) 495 ; + j. ellis , j.s .
hagelin , d.v .
nanopoulos and k. tamvakis , phys . lett . *
b125 * ( 1983 ) 275 ; + l.e .
ibez and c. lopez , nucl . phys .
* b233 * ( 1984 ) 511 ; + l.e .
ibez , c. lopez and c. muoz , nucl . phys .
* b256 * ( 1985 ) 218 .
k. hikasa , _ minimal supersymmetry for collider physicists _ , unpublished
. for review on @xmath0-problem , see , e.g. c. muoz , ftuam 95/20 .
l. dixon , j. harvey , c. vafa and e. witten , nucl
b261 * ( 1985 ) 651 ; * b274 * ( 1986 ) 285 ; + l.e .
ibez , j.e .
kim , h.p .
nilles and f. quevedo , phys .
b191 * ( 1987 ) 282 ; + a. font , l.e .
ibez and f. quevedo , phys .
* b217 * ( 1989 ) 272 ; + y. katsuki , y. kawamura , t. kobayashi , n. ohtsubo , y. ono and k. tanioka , nucl .
* b341 * ( 1990 ) 611 ; + t. kobayashi and n. ohtsubo , phys . lett . *
b262 * ( 1991 ) 425 ; int .
j. mod .
a9 * ( 1994 ) 87 .
g. lopes - cardoso , d. lst and t. mohaupt , nucl .
* b432 * ( 1994 ) 68 ; + i. antoniadis , e. gava , k.s .
narain and t.r .
taylor , nucl . phys .
* b432 * ( 1994 ) 187 . for review on the supersymmetric higgs sector ,
see , e.g. l.e .
ibez and g.g .
ross , _ perspectives in higgs physics _ , ed .
g. kane .
green and j.h .
schwarz , phys .
* b149 * ( 1984 ) 117 ; + l.e .
ibez , phys .
b303 * ( 1993 ) 55 .
g. gamberini , g. ridolfi and f. zwirner , nucl . phys .
* b331 * ( 1990 ) 187 .
y. kawamura , s. khalil and t. kobayashi , preprint , dpsu-97 - 4 , ins - rep-1189 , hep - ph/9703239 .
1 . the values of @xmath181 and @xmath54 . here
we use @xmath182gev , i.e. , @xmath183gev and @xmath184 .
2 . the formulae of @xmath116 and @xmath115 . the second column shows the dilaton dominant susy breaking case and the third one is the dilaton and overall moduli mixed case .
here we take modular weights @xmath185 .
table 1 [ cols="^,^,^,^",options="header " , ]
1 . the values of @xmath113 versus @xmath9 with @xmath186
. 2 . the values of @xmath114 versus @xmath187 with @xmath186 .
3 . the values of @xmath113 versus @xmath118 with @xmath188 in the limit of @xmath148 .
4 . the values of @xmath55 versus @xmath9 with @xmath189 . | we give a generic method to select a realistic @xmath0-term generation mechanism based on the radiative electroweak symmetry breaking scenario and study which type is hopeful within the framework of string theory .
we discuss effects of the moduli @xmath1-term condensation and @xmath2-term contribution to soft scalar masses .
= 10000 = 10000 dpsu-97 - 5 + kek preprint 97 - 11 + may , 1997 + * probing @xmath0-term generation mechanism + in string models * * yoshiharu kawamura @xmath3 * , * tatsuo kobayashi @xmath4 * + and + * manabu watanabe @xmath3 * @xmath3 department of physics , shinshu university + matsumoto 390 , japan + and + @xmath4 institute of particle and nuclear studies + high energy accelerator research organization + midori - cho , tanashi , tokyo 188 , japan + |
integrity constraints provide means for ensuring that database evolution does not result in a loss of consistency or in a discrepancy with the intended model of the application domain @xcite .
a relational database that do not satisfy some of these constraints is said to be inconsistent . in practice
it is not unusual that one has to deal with inconsistent data @xcite , and when a conjunctive query ( cq ) is posed to an inconsistent database , a natural problem arises that can be formulated as : _ how to deal with inconsistencies to answer the input query in a consistent way ? _
this is a classical problem in database research and different approaches have been proposed in the literature .
one possibility is to clean the database @xcite and work on one of the possible coherent states ; another possibility is to be tolerant of inconsistencies by leaving intact the database and computing answers that are `` consistent with the integrity constraints '' @xcite . in this paper
, we adopt the second approach which has been proposed by under the name of _ consistent query answering _
( cqa ) and focus on the relevant class of _ primary key _ constraints . formally , in our setting : @xmath0 a database @xmath1 is _ inconsistent _ if there are at least two tuples of the same relation that agree on their primary key ; @xmath2 a _ repair _ of @xmath1 is any maximal consistent subset of @xmath1 ; and @xmath3 a tuple @xmath4 of constants is in the _ consistent answer _ to a cq @xmath5 over @xmath1 if and only if , for each repair @xmath6 of @xmath1 , tuple @xmath4 is in the ( classical ) answer to @xmath5 over @xmath6 .
intuitively , the original database is ( virtually ) repaired by applying a minimal number of corrections ( deletion of tuples with the same primary key ) , while the consistent answer collects the tuples that can be retrieved in every repaired instance .
cqa under primary keys is conp - complete in data complexity @xcite , when both the relational schema and the query are considered fixed . due to its complex nature ,
traditional rdbms are inadequate to solve the problem alone via sql without focusing on restricted classes of cqs @xcite . actually ,
in the unrestricted case , cqa has been traditionally dealt with logic programming @xcite .
however , it has been argued @xcite that the practical applicability of logic - based approaches is restricted to data sets of moderate size .
only recently , an approach based on binary integer programming @xcite has revealed good performances on large databases ( featuring up to one million tuples per relation ) with primary key violations . in this paper
, we demonstrate that logic programming can still be effectively used for computing consistent answers over large relational databases .
we design a novel decomposition strategy that reduces ( in polynomial time ) the computation of the consistent answer to a cq over a database subject to primary key constraints into a collection of smaller problems of the same sort . at the core of the strategy
is a cascade pruning mechanism that dramatically reduces the number of key violations that have to be handled to answer the query .
moreover , we implement the new strategy using answer set programming ( asp ) @xcite , and we prove empirically the effectiveness of our asp - based approach on existing benchmarks from the database world .
in particular , we compare our approach with some classical @xcite and optimized @xcite encodings of cqa in asp that were presented in the literature .
the experiment empirically demonstrate that our logic - based approach implements cqa efficiently on large data sets , and can even perform better than state - of - the - art methods .
we are given two disjoint countably infinite sets of _ terms _ denoted by @xmath7 and @xmath8 and called _ constants _ and _ variables _ , respectively .
we denote by @xmath9 sequences ( or sets , with a slight abuse of notation ) of variables @xmath10 , and by @xmath11 sequences of terms @xmath12 .
we also denote by @xmath13 $ ] the set @xmath14 , for any @xmath15 . given a sequence @xmath16 of terms and a set @xmath17 $ ]
, @xmath18 is the subsequence @xmath19 .
for example , if @xmath20 and @xmath21 , then @xmath22 .
a ( _ relational _ ) _ schema _ is a triple @xmath23 where @xmath24 is a finite set of _ relation symbols _ ( or _ predicates _ ) , @xmath25 is a function associating an _
arity _ to each predicate , and @xmath26 is a function that associates , to each @xmath27 , a nonempty set of positions from @xmath28 $ ] , which represents the _ primary key _ of @xmath29 .
moreover , for each relation symbol @xmath27 and for each position @xmath30 $ ] , @xmath31 $ ] denotes the @xmath32-th _ attribute _ of @xmath29 . throughout ,
let @xmath33 denote a relational schema .
an _ atom _ ( over @xmath34 ) is an expression of the form @xmath35 , where @xmath27 , and @xmath36 .
an atom is called a _ fact _ if all of its terms are constants of @xmath7 .
conjunctions of atoms are often identified with the sets of their atoms . for a set @xmath37 of atoms ,
the variables occurring in @xmath37 are denoted by @xmath38 .
a _ database _ @xmath1 ( over @xmath34 ) is a finite set of facts over @xmath39 .
given an atom @xmath40 , we denote by @xmath41 the sequence @xmath42 .
we say that @xmath1 is _ inconsistent _ ( w.r.t .
@xmath34 ) if it contains two different atoms of the form @xmath43 and @xmath44 such that @xmath45 .
otherwise , it is _ consistent_. a _ repair _ @xmath6 of @xmath1 ( w.r.t .
@xmath34 ) is any maximal consistent subset of @xmath1 .
the set of all the repairs of @xmath1 is denoted by @xmath46 .
a _ substitution _ is a mapping @xmath47 which is the identity on @xmath7 .
given a set @xmath37 of atoms , .
the restriction of @xmath48 to a set @xmath49 , is denoted by @xmath50 . a _ conjunctive query _ ( cq ) @xmath5 ( over @xmath34 ) is an expression of the form @xmath51 , where @xmath52 are variables of @xmath8 , and @xmath53 is a conjunction of atoms ( possibly with constants ) over @xmath34 . to highlight the free variables of @xmath5 ,
we often write @xmath54 instead of @xmath5 .
if @xmath9 is empty , then @xmath5 is called a _ boolean conjunctive query _ ( bcq ) . assuming that @xmath9 is the sequence @xmath10 , the _ answer _ to @xmath5 over a database @xmath1 , denoted @xmath55 , is the set of all @xmath56-tuples @xmath57 for which there exists a substitution @xmath48 such that @xmath58 and @xmath59 , for each @xmath60 $ ] .
a bcq is _ true _ in @xmath1 , denoted @xmath61 , if @xmath62 .
consistent answer _ to a cq @xmath54 over a database @xmath1 ( w.r.t .
@xmath34 ) , denoted @xmath63 , is the set of tuples @xmath64 .
clearly , @xmath65 holds .
a bcq @xmath5 is _ consistently true _ in a database @xmath1 ( w.r.t .
@xmath34 ) , denoted @xmath66 , if @xmath67 .
to deal with large inconsistent data , we design a strategy that reduces in polynomial time the problem of computing the consistent answer to a cq over a database subject to primary key constraints to a collection of smaller problems of the same sort . to this end , we exploit the fact that the former problem is logspace turing reducible to the one of deciding whether a bcq is consistently true ( recall that the consistent answer to a cq is a subset of its answer ) . hence , given a database @xmath1 over a schema @xmath34 , and a bcq @xmath5 , we would like to identify a set @xmath68 of pairwise disjoint subsets of @xmath1 , called fragments , such that : _
@xmath69 iff there is @xmath70 $ ] such that @xmath71_. at the core of our strategy we have : @xmath0 a cascade pruning mechanism to reduce the number of `` crucial '' inconsistencies , and @xmath2 a technique to identify a suitable set of fragments from any ( possibly unpruned ) database . for the sake of presentation
, we start with principle @xmath2 . in the last two subsections , we provide complementary techniques to further reduce the number of inconsistencies to be handled for answering the original cq .
the proofs of this section are given in [ sec : app - proofs ] .
given a database @xmath1 , a _
key component _
@xmath72 of @xmath1 is any maximal subset of @xmath1 such that if @xmath73 and @xmath74 are in @xmath72 , then both @xmath75 and @xmath45 hold .
namely , @xmath72 collects only atoms that agree on their primary key .
hence , the set of all key components of @xmath1 , denoted by @xmath76 , forms a partition of @xmath1 .
if a key component is a singleton , then it is called _ safe _ ; otherwise it is _
conflicting_. let @xmath77 .
it can be verified that @xmath78 .
let us now fix throughout this section a bcq @xmath5 over @xmath34 .
for a repair @xmath79 , if @xmath5 is true in @xmath6 , then there is a substitution @xmath48 such that @xmath80 .
but since @xmath81 , it also holds that @xmath82 .
hence , is an overestimation of the substitutions that map @xmath5 to the repairs of @xmath1 . inspired by the notions of conflict - hypergraph @xcite and conflict - join graph @xcite
, we now introduce the notion of conflict - join hypergraph . given a database @xmath1 , the _ conflict - join hypergraph _ of @xmath1 ( w.r.t .
@xmath5 and @xmath34 ) is denoted by @xmath83 , where @xmath1 are the vertices , and @xmath84 are the hyperedges partitioned in @xmath85 and @xmath86 .
@xmath87 of vertices of @xmath88 is any minimal nonempty subset of @xmath1 such that , for each @xmath89 , either @xmath90 or @xmath91 holds .
intuitively , every edge of @xmath88 collects the atoms in a key component of @xmath1 or the atoms in @xmath92 , for some @xmath93 .
moreover , each bunch collects the vertices of some connected component of @xmath88 . before we proceed further ,
let us fix these preliminary notions with the aid of the following example .
[ ex : hypergraph ] consider the schema @xmath33 , where @xmath94 , @xmath95 , and @xmath96 .
consider also the database @xmath97 @xmath98 @xmath99 @xmath100 , @xmath101 , and the bcq @xmath102 .
the conflicting components of @xmath1 are @xmath103 @xmath104 and @xmath105 , @xmath101 , while its safe component is @xmath106 .
the repairs of @xmath1 are @xmath107 @xmath99 @xmath108 , @xmath109 @xmath99 @xmath101 ,
@xmath110 @xmath99 @xmath108 , and @xmath111 @xmath99 @xmath101 .
moreover , @xmath112 contains the substitutions : @xmath113 , @xmath114 , @xmath115 , and @xmath116 .
the conflict - join hypergraph @xmath83 is depicted in figure [ fig : repairs ] .
solid ( resp .
, dashed ) edges form the set @xmath117 ( resp . , @xmath118 ) .
since @xmath119 maps @xmath5 to @xmath120 and @xmath121 , and @xmath122 maps @xmath5 to @xmath123 and @xmath124 , we conclude that @xmath69 .
finally , @xmath1 is the only bunch of @xmath88 .
@xmath125 in example [ ex : hypergraph ] we observe that @xmath126 can be safely ignored in the evaluation of @xmath5 .
in fact , even if both @xmath127 and @xmath128 contain an atom of @xmath126 , @xmath119 and @xmath122 are sufficient to prove that @xmath5 is consistently true .
this might suggest to focus only on the set @xmath129 , and on its repairs @xmath130 @xmath131 and @xmath132 @xmath131 .
also , since @xmath133 , @xmath134 represents the `` small '' fragment of @xmath1 that we need to evaluate @xmath5
. the practical advantage of considering @xmath134 instead of @xmath1 should be already clear : @xmath0 the repairs of @xmath134 are smaller than the repairs of @xmath1 ; and @xmath2 @xmath134 has less repairs than @xmath1 .
we are now ready to introduce the the formal notion of fragment .
[ def : fragment ] consider a database @xmath1 .
for any set @xmath135 of key components of @xmath1 , we say that the set @xmath136 is a ( _ well - defined _ ) _ fragment _ of @xmath1 .
@xmath125 according to definition [ def : fragment ] , the set @xmath129 in example [ ex : hypergraph ] is a fragment of @xmath1 .
the following proposition , states a useful property that holds for any fragment .
[ prop : fragment ] consider a database @xmath1 , and two fragments @xmath137 of @xmath1 .
if @xmath138 , then @xmath139 . by definition [ def :
fragment ] , @xmath1 is indeed a fragment of itself .
hence , if @xmath5 is consistently true , then there is always the fragment @xmath140 such that @xmath133 .
but now the question is : _ how can we identify a convenient set of fragments of @xmath1 _ ?
the naive way would be to use as fragments the bunches of @xmath88 .
soundness is guaranteed by proposition [ prop : fragment ] . regarding completeness
, we rely on the following result .
[ thm : bunches ] consider a database @xmath1 .
if @xmath69 , then there is a bunch @xmath87 of @xmath88 s.t . @xmath141 . by combining proposition [ prop : fragment ] with theorem [ thm : bunches ]
we are able to reduce , in polynomial time , the original problem into a collection of smaller ones of the same sort .
the technique proposed in the previous section alone is not sufficient to deal with large data sets .
in fact , since it considers all the bunches of the conflict - join hypergraph , it unavoidably involves the entire database . to strengthen its effectiveness ,
we need an algorithm that realizes , for instance , that @xmath126 is `` redundant '' in example [ ex : hypergraph ] . but
before that , let us define formally what we mean by the term redundant .
[ def : redundant ] a key component @xmath72 of a database @xmath1 is called _ redundant _ ( w.r.t .
@xmath5 ) if the following condition is satisfied : for each fragment @xmath134 of @xmath1 , @xmath133 implies @xmath142 .
@xmath125 the above definition states that a key component is redundant independently from the fact that some other key component is redundant or not .
therefore : [ prop : redundancyclosure ] consider a database @xmath1 and a set @xmath143 of redundant components of @xmath1 .
it holds that @xmath69 iff @xmath144 . in light of proposition
[ prop : redundancyclosure ] , if we can identify all the redundant components of @xmath1 , then after removing from @xmath1 all these components , what remains is either : @xmath0 a nonempty set of ( minimal ) bunches , each of which entails consistently @xmath5 whenever @xmath69 ; or @xmath2 the empty set , whenever @xmath145 . more formally : [ prop : truequeryhasnonredundantcomponent ] given a database @xmath1 , each key component of @xmath1 is redundant iff @xmath145 . however , assuming that @xmath146 , any algorithm for the identification of all the redundant components of @xmath1 can not be polynomial because , otherwise , we would have a polynomial procedure for solving the original problem .
our goal is therefore to identify sufficient conditions to design a pruning mechanism that detects in polynomial time as many redundant conflicting components as possible . to give an intuition of our pruning mechanism
, we look again at example [ ex : hypergraph ] .
actually , @xmath126 is redundant because it contains an atom , namely @xmath147 , that is not involved in any substitution ( see figure [ fig : repairs ] ) .
assume now that this is the criterion that we use to identify redundant components . since , by definition [ def : redundant ] , we know that @xmath69 iff @xmath148 , this means that we can now forget about @xmath1 and consider only @xmath149 .
but once we focus on @xmath150 , we realize that it contains only @xmath119 and @xmath122 .
then , a smaller number of substitutions in @xmath150 w.r.t . those in @xmath112 motivates us to reapply our criterion .
indeed , there could also be some atom in @xmath151 not involved in any of the substitutions of @xmath150 .
this is not the case in our example since the atoms in @xmath151 are covered by @xmath152 or @xmath153 .
however , in general , in one or more steps , we can identify more and more redundant components .
we can now state the main result of this section .
[ thm : pruning ] consider a database @xmath1 , and a key component @xmath72 of @xmath1 .
let @xmath83 be the conflict - join hypergraph of @xmath1 .
if @xmath154 , then @xmath72 is redundant . in what follows , a redundant component that can be identified via theorem [ thm : pruning ]
is called _
strongly redundant_. as discussed just before theorem [ thm : pruning ] , an indirect effect of removing a redundant component @xmath72 from @xmath1 is that all the substitutions in the set can be in a sense ignored .
in fact , @xmath155 . whenever a substitution can be safely ignored , we say that it is unfounded .
let us formalize this new notion in the following definition .
[ def : unfoundedsub ] consider a database @xmath1 .
a substitution @xmath48 of @xmath112 is _ unfounded _ if : for each fragment @xmath134 of @xmath1 , @xmath133 implies that , for each repair @xmath156 , there exists a substitution @xmath157 different from @xmath48 such that @xmath158 .
@xmath125 we now show how to detect as many unfounded substitutions as possible .
[ thm : suffunfoundedsub ] consider a database @xmath1 , and a substitution @xmath93 .
if there exists a redundant component @xmath72 of @xmath1 such that @xmath159 , then @xmath48 is unfounded . clearly , theorem [ thm : suffunfoundedsub ] alone is not helpful since it relies on the identification of redundant components .
however , if combined with theorem [ thm : pruning ] , it forms the desired cascade pruning mechanism . to this end , we call _ strongly unfounded _ an unfounded substitution that can be identified by applying theorem [ thm : suffunfoundedsub ] by only considering strongly redundant components .
hereafter , let us denote by @xmath160 the subset of @xmath112 containing only strongly unfounded substitutions .
hence , both substitutions @xmath161 and @xmath162 in example [ ex : hypergraph ] are strongly unfounded , since @xmath126 is strongly redundant . moreover , we reformulate the statement of theorem [ thm : pruning ] by exploiting the notion of strongly unfounded substitution , and the fact that the set @xmath163 is nonempty if and only if there exists an atom @xmath164 such that the set @xmath165 or equivalently the set @xmath166 is empty . for example , according to figure [ fig : repairs ] , the set @xmath167 is nonempty since it contains the atom @xmath147 .
but this atoms makes the set @xmath168 empty since no substitution of @xmath112 ( or no hyperedge of @xmath118 ) involves @xmath147 .
[ prop : stronglyredundant ] a key component @xmath72 of @xmath1 is strongly redundant if there is an atom @xmath164 such that one of the two following conditions is satisfied : ( 1 ) @xmath169 , or ( 2 ) @xmath170 . by combining theorem [ thm : suffunfoundedsub ] and proposition [ prop : stronglyredundant ] , we have a declarative ( yet inductive ) specification of all the strongly redundant components of @xmath1 .
importantly , the process of identifying strongly redundant components and strongly unfounded substitutions by exhaustively applying theorem [ thm : suffunfoundedsub ] and proposition [ prop : stronglyredundant ] is monotone and reaches a fixed - point ( after no more than @xmath171 steps ) when no more key component can be marked as strongly redundant .
previously , we have described a technique to reduce inconsistencies by progressively eliminating key components that are involved in query substitutions but are redundant . in the following ,
we show how to reduce inconsistencies by reducing the cardinality of conflicting components , which in some cases can be even treated as safe components .
the act of _ removing _ an attribute @xmath31 $ ] from a triple @xmath172 consists of reducing the arity of @xmath29 by one , cutting down the @xmath32-th term of each @xmath29-atom of @xmath1 and @xmath5 , and adapting the positions of the primary key of @xmath29 accordingly .
moreover , let @xmath173~|~r \in { \mathcal{r}}~\textrm { and } ~i \in [ \alpha(r)]\}$ ] , let @xmath174 , and let @xmath175 .
the _ projection _ of @xmath172 on @xmath37 , denoted by @xmath176 , is the triple that is obtained from @xmath172 by removing all the attributes of @xmath87 . consider a cq @xmath5 and a predicate @xmath27 .
the attribute @xmath31 $ ] is _ relevant _ ( w.r.t .
@xmath5 ) if @xmath5 contains an atom of the form @xmath177 such that at least one of the following conditions is satisfied : @xmath0 @xmath178 ; or @xmath2 @xmath179 is a constant ; or @xmath3 @xmath179 is a variable that occurs more than once in @xmath5 ; or @xmath180 @xmath179 is a free variable of @xmath5 .
an attribute which is not relevant is _ idle _
@xmath5 ) .
an example is reported in [ sec : exampleidle ] .
the following theorem states that the consistent answer to a cq does not change after removing the idle attributes .
[ thm : idle ] consider a cq @xmath5 , the set @xmath181\in \mathit{attrs}({\sigma})~|~r[i]~\textrm { is relevant w.r.t . }
q\}$ ] , and a database @xmath1 .
it holds that @xmath182 .
let @xmath39 be a relational schema , @xmath1 be a database , and @xmath183 be a cq , where we assume that @xmath184 contains only relevant attributes w.r.t .
@xmath5 ( idle attributes , if any , have been already removed ) . since @xmath65 , for each candidate answer @xmath185 , one should evaluate whether the bcq @xmath186 is ( or is not ) consistently true in @xmath1 . before constructing the conflict - join hypergraph of @xmath1 ( w.r.t .
@xmath187 and @xmath34 ) , however , one could check whether there is a substitution @xmath48 that maps @xmath187 to @xmath1 with the following property : for each @xmath188 , the singleton @xmath189 is a safe component of @xmath1 . and
, if so , it is possible to conclude immediately that @xmath190 .
intuitively , whenever the above property is satisfied , we say that @xmath191 is a
_ safe answer _ to @xmath5 because , for each @xmath192 , it is guaranteed that @xmath193 .
the next result follows .
consider a cq @xmath183 , and a tuple @xmath191 of @xmath55 . if there is a substitution @xmath48 s.t .
each atom of @xmath194 forms a safe component of @xmath1 , then @xmath190 .
in this section , we propose an asp - based encoding to cqa that implements the techniques described in section [ sec : pruning ] , and that is able to deal directly with cqs , instead of evaluating separately the associated bcqs .
hereafter , we assume the reader is familiar with answer set programming @xcite and with the standard syntax of asp competitions @xcite . a nice introduction to asp
can be found in @xcite , and in the asp core 2.0 specification in @xcite . given a relational schema @xmath195 , a database @xmath1 , and a cq @xmath183
, we construct a program @xmath196 s.t .
a tuple @xmath197 belongs to @xmath63 iff each answer set of @xmath198 contains an atom of the form @xmath199 , for some constant @xmath200 .
importantly , a large part of @xmath196 does not depend on @xmath5 or @xmath39 . to lighten the presentation
, we provide a simplified version of the encoding that has been used in our experiments .
in fact , for efficiency reasons , idle attributes should be `` ignored on - the - fly '' without materializing the projection of @xmath201 on the relevant attributes ; but this makes the encoding a little more heavy .
hence , we first provide a naive way to consider only the relevant attributes , and them we will assume that @xmath184 contains no idle attribute . let @xmath6 collect all the attributes of @xmath39 that are relevant w.r.t .
@xmath5 . for each @xmath27 that occurs in @xmath5 ,
let @xmath202 be a sequence of @xmath203 different variables and @xmath204\in r\}$ ] , the terms of the @xmath29-atoms of @xmath1 that are associated to idle attributes can be removed via the rule @xmath205 .
hereafter , let us assume that @xmath184 contains no idle attribute , and @xmath206 .
program @xmath196 is depicted in figure [ fig : encoding ] . ' '' '' = = = + % + @xmath207 .
+ @xmath208 . `
@xmath209 + @xmath210 .
+ @xmath211 . +
[ 2 mm ] % + @xmath212 .
+ @xmath213 . `
@xmath209 + @xmath214 . `
@xmath209 + @xmath215 . `
@xmath209 + [ 2 mm ] % + @xmath216 + @xmath217 .
+ @xmath218 .
+ @xmath219 , + @xmath220 . + @xmath221 . +
[ 2 mm ] % + @xmath222 + @xmath223 .
+ @xmath224
. + @xmath225
. + @xmath226
. + @xmath227
. + @xmath228
. + @xmath229 + @xmath230
. + @xmath231 . +
[ 2 mm ] % + @xmath232 .
+ @xmath233
. + @xmath234 .
+ [ 2 mm ] % + @xmath235
. + @xmath236 .
+ @xmath237 + @xmath238 .
+ ' '' '' _ computation of the safe answer .
_ via rule @xmath239 , we identify the set @xmath240 is a substitution and @xmath241 .
it is now possible ( rule @xmath242 ) to identify the atoms of @xmath1 that are involved in some substitution . here
, for each atom @xmath243 , we recall that @xmath244 is the subsequence of @xmath4 containing the terms in the positions of the primary key of @xmath29 , and we assume that @xmath245 are the terms of @xmath4 in the remaining positions . in particular , we use two function symbols , @xmath246 and @xmath247 , to group the terms in the key of @xmath29 and the remaining ones , respectively .
it is now easy ( rule @xmath248 ) to identify the conflicting components involved in some substitution .
let @xmath249 .
we now compute ( rule @xmath250 ) the safe answers
. _ hypergraph construction .
_ for each candidate answer @xmath185 that has not been already recognized as safe , we construct the hypergraph @xmath251 associated to the bcq @xmath252 , where @xmath253 , as usual .
hypergraph @xmath254 is identified by the functional term @xmath255 , the substitutions of @xmath118 ( collected via rule @xmath256 ) are identified by the set @xmath257 and @xmath258 of functional terms , while the key components of @xmath117 ( collected via rule @xmath259 ) are identified by the set @xmath260 and @xmath261 and @xmath262 of functional terms . to complete the construction of the various hypergraphs
, we need to specify ( rules @xmath263 and @xmath264 ) which are the atoms in each hyperedge .
_ pruning .
_ we are now ready to identify ( rules @xmath265 ) the strongly redundant components and the strongly unfounded substitutions ( as described in section [ sec : pruning ] ) to implement our cascade pruning mechanism .
hence , it is not difficult to collect ( rule @xmath266 ) the substitutions that are not unfounded , that we call _
residual_. _ fragments identification .
_ key components involving at least a residual substitution ( i.e. , not redundant ones ) , can be aggregated in fragments ( rules @xmath267 ) by using the notion of bunch introduced in section [ sec : fragments ] . in particular ,
any given fragment @xmath134 associated to a candidate answer @xmath185 , and collecting the key components @xmath268 is identified by the functional term @xmath269 where , for each @xmath270 , the functional term associated to @xmath271 lexicographically precedes the functional term associated to @xmath272 . _ repair construction .
_ rules @xmath273 can be evaluated in polynomial time and have only one answer set , while the remaining part of the program can not in general .
in particular , rules @xmath274 generate the search space .
actually , each answer set @xmath275 of @xmath196 is associated ( rule @xmath276 ) with only one fragment , say @xmath134 , that we call _ active _ in @xmath275 .
moreover , for each key component @xmath72 of @xmath134 , answer set @xmath275 is also associated ( rule @xmath277 ) with only one atom of @xmath72 , that we also call _ active _ in @xmath275 .
consequently , each substitution which involves atoms of @xmath134 but also at least one atom which is not active , must be ignored in @xmath275 ( rule @xmath278 ) . _
new query .
_ finally , we compute the atoms of the form @xmath199 via rules @xmath279 .
the experiment for assessing the effectiveness of our approach is described in the following .
we first describe the benchmark setup and , then , we analyze the results .
_ benchmark setup .
_ the assessment of our approach was done using a benchmark employed in the literature for testing cqa systems on large inconsistent databases @xcite .
it comprises 40 instances of a database schema with 10 tables , organized in four families of 10 instances each of which contains tables of size varying from 100k to 1 m tuples ; also it includes 21 queries of different structural features split into three groups depending on whether cqa complexity is conp - complete ( queries @xmath280 ) , ptime but not fo - rewritable @xcite ( queries @xmath281 ) , and fo - rewritable ( queries @xmath282 ) .
( see [ app : bench ] ) .
we compare our approach , named _ pruning _ , with two alternative asp - based approaches . in particular ,
we considered one of the first encoding of cqa in asp that was introduced in @xcite , and an optimized technique that was introduced more recently in @xcite ; these are named _ bb_and _ mrt _
, respectively .
_ bb_and _ mrt_can handle a larger class of integrity constrains than _ pruning _ , and only _
mrt_features specific optimization that apply also to primary key violations handling .
we constructed the three alternative encodings for all 21 queries of the benchmark , and we run them on the asp solver wasp 2.0 @xcite , configured with the iterative coherence testing algorithm @xcite , coupled with the grounder gringo ver .
4.4.0 @xcite.for completeness we have also run claspver .
3.1.1 @xcite obtaining similar results
. waspperformed better in terms of number of solved instances on _ mrt_and _ bb_.
the experiment was run on a debian server equipped with xeon e5 - 4610 cpus and 128 gb of ram . in each execution ,
resource usage was limited to 600 seconds and 16 gb of ram .
execution times include the entire computation , i.e. , both grounding and solving .
all the material for reproducing the experiment ( asp programs , and solver binaries ) can be downloaded from www.mat.unical.it/ricca/downloads/mrticlp2015.zip .
_ analysis of the results .
_ concerning the capability of providing an answer to a query within the time limit , we report that _
pruning_was able to answer the queries in all the 840 runs in the benchmark with an average time of 14.6s . _ mrt _ , and _ bb_solved only 778 , and 768 instances within 600 seconds , with an average of 80.5s and 52.3s , respectively .
the cactus plot in figure [ fig : comp : cac ] provides an aggregate view of the performance of the compared methods .
recall that a cactus plot reports for each method the number of answered queries ( solved instances ) in a given time .
we observe that the line corresponding to _ pruning_in figure [ fig : comp : cac ] is always below the ones of _ mrt_and _ bb_. in more detail , _
pruning_execution times grow almost linearly with the number of answered queries , whereas _ mrt_and _
bb_show an exponential behavior .
we also note that _
mrt_behaves better than _ bb _ , and this is due to the optimizations done in _
mrt_that reduce the search space . the performance of the approaches w.r.t .
the size of the database is studied in figure [ fig : comp : avgsol ] . the x - axis reports the number of tuples per relation in tenth of thousands , in the upper plot
is reported the number of queries answered in 600s , and in the lower plot is reported the corresponding the average running time .
we observe that all the approaches can answer all 84 queries ( 21 queries per 4 databases ) up to the size of 300k tuples , then the number of answered queries by both _ bb_and _
mrt_starts decreasing .
indeed , they can answer respectively 74 and 75 queries of size 600k tuples , and only 67 and 71 queries on the largest databases ( 1 m tuples ) . instead , _
pruning_is able to solve all the queries in the data set . the average time elapsed by running _
pruning_grows linearly from 2.4s up to 27.4s .
_ mrt_and _ bb_average times show a non - linear growth and peak at 128.9s and 85.2s , respectively .
( average is computed on queries answered in 600s , this explains why it apparently decreases when a method can not answer some instance within 600s . ) the scalability of _
pruning_is studied in detail for each query in figures [ fig : ratioscalability](d - f ) , each plotting the average execution times per group of queries of the same theoretical complexity .
it is worth noting that _ pruning_scales almost linearly in all queries , and independently from the complexity class of the query .
this is because _
pruning_is able to identify and deal efficiently with the conflicting fragments .
we now analyze the performance of _ pruning_from the perspective of a measure called _ overhead _ , which was employed in @xcite for measuring the performance of cqa systems . given a query q the overhead is given by @xmath283 , where @xmath284 is time needed for computing the consistent answer of q , and @xmath285 is the time needed for a plain execution of q where the violation of integrity constraints are ignored .
note that the overhead measure is independent of the hardware and the software employed , since it relates the computation of cqa to the execution of a plain query on the same system .
thus it allows for a direct comparison of _
pruning_with other methods having known overheads . following what was done in @xcite , we computed the average overhead measured varying the database size for each query , and we report the results by grouping queries per complexity class in figures [ fig : ratioscalability](a - c ) .
the overheads of _
pruning_is always below 2.1 , and the majority of queries has overheads of around 1.5 .
the behavior is basically ideal for query q5 and q4 ( overhead is about 1 ) .
the state of the art approach described in @xcite has overheads that range between 5 and 2.8 on the very same dataset ( more details on [ app : bench ] ) .
thus , our approach allows to obtain a very effective implementation of cqa in asp with an overhead that is often more than two times smaller than the one of state - of - the - art approaches .
we complemented this analysis by measuring also the overhead of _ pruning_w.r.t .
the computation of safe answers , which provide an underestimate of consistent answers that can be computed efficiently ( in polynomial time ) by means of stratified asp programs .
we report that the computation of the consistent answer with _
pruning_requires only at most 1.5 times more in average than computing the safe answer ( detailed plots in [ app : bench ] ) .
this further outlines that _
pruning_is able to maintain reasonable the impact of the hard - to - evaluate component of cqa .
finally , we have analyzed the impact of our technique in the various solving steps of the evaluation .
the first three histograms in figure [ fig : grounding ] report the average running time spent for answering queries in databases of growing size for _ pruning_(fig .
[ fig : hpr ] ) , _ bb_(fig .
[ fig : hbb ] ) , and _ mrt_(fig .
[ fig : hmrt ] ) . in each bar
different colors distinguish the average time spent for grounding and solving .
in particular , the average solving time over queries _ answered within the timeout _ is labeled solving - sol , and each bar extends up to the average cumulative execution time computed over all instances , where each timed out execution counts 600s .
recall that , roughly speaking , the grounder solves stratified normal programs , and the hard part of the computation is performed by the solver on the residual non - stratified program ; thus , we additionally report in figure [ fig : grperc ] the average number of facts ( knowledge inferred by grounding ) and of non - factual rules ( to be evaluated by the solver ) in percentage of the total for the three compared approaches .
the data in figure [ fig : grounding ] confirm that most of the computation is done with _
pruning_during the grounding , whereas this is not the case for _ mrt_and _ bb_. figure [ fig : grperc ] shows that for _ pruning_the grounder produces a few non - factual rules ( below 1% in average ) , whereas _ mrt_and _
bb_produce 5% and 63% of non - factual rules , respectively .
roughly , this corresponds to about 23k non - factual rules ( resp .
, 375k non - factual rules ) every 100k tuples per relation for _ mrt_(resp .
, _ bb _ ) , whereas our approach produces no more than 650 non - factual rules every 100k tuples per relation .
logic programming approaches to cqa were recently considered not competitive @xcite on large databases affected by primary key violations . in this paper , we proposed a new strategy based on a cascade pruning mechanism that dramatically reduces the number of primary key violations to be handled to answer the query .
the strategy is encoded naturally in asp , and an experiment on benchmarks already employed in the literature demonstrates that our asp - based approach is efficient on large datasets , and performs better than state - of - the - art methods in terms of overhead . as far as future work
is concerned , we plan to extend the _
pruning_method for handling inclusion dependencies , and other tractable classes of tuple - generating dependencies . ,
hull , r. , and vianu , v. 1995 . .
addison - wesley . , dodaro , c. , and ricca , f. 2014a .
anytime computation of cautious consequences in answer set programming . _
14 , _ 4 - 5 , 755770 . ,
dodaro , c. , and ricca , f. 2014b .
preliminary report on wasp 2.0 . _
abs/1404.6999_. , bertossi , l. e. , and chomicki , j. 1999 .
consistent query answers in inconsistent databases . in _ proceedings of pods
99_. 6879 . ,
bertossi , l. e. , and chomicki , j. 2003 . answer sets for consistent query answering in inconsistent databases .
_ 3 , _ 4 - 5 , 393424
. \2003 . .
cambridge university press .
logic programs for querying inconsistent databases . in _ proceedings of padl03_. lncs , vol
springer , 208222 .
synthesis lectures on data management .
morgan & claypool publishers . , hunter , a. , and schaub , t. , eds .
lncs , vol .
springer , berlin / heidelberg . ,
eiter , t. , and truszczynski , m. 2011 .
answer set programming at a glance .
_ 54 , _ 12 , 92103 . ,
faber , w. , gebser , m. , ianni , g. , kaminski , r. , krennwallner , t. , leone , n. , ricca , f. , and schaub , t. 2013 .
asp - core-2 input language format .
available at https://www.mat.unical.it/aspcomp2013/files/asp-core-2.03b.pdf . ,
ianni , g. , and ricca , f. 2014 . the third open answer set programming competition . _ 14 , _ 1 , 117135 .
minimal - change integrity maintenance using tuple deletions .
_ 197 , _ 1 - 2 , 90121 . ,
fink , m. , greco , g. , and lembo , d. 2003 .
efficient evaluation of logic programs for querying data integration systems . in _ proceedings of iclp03_. lncs , vol .
springer , 163177 . , ipeirotis , p. g. , and verykios , v. s. 2007 .
duplicate record detection : a survey .
_ 19 , _ 1 , 116 . ,
fazli , e. , and miller , r. j. 2005 .
conquer : efficient management of inconsistent databases . in _ proceedings of sigmod05_. acm , 155166 .
first - order query rewriting for inconsistent databases .
_ 73 , _ 4 , 610635 . , kaminski , r. , knig , a. , and schaub , t. 2011 .
advances in _ gringo _ series 3 . in _ logic programming and nonmonotonic reasoning - 11th international conference , lpnmr 2011 , vancouver , canada , may 16 - 19 , 2011 . proceedings _ , j. p. delgrande and w. faber , eds .
lecture notes in computer science , vol . 6645 .
springer , 345351 . ,
kaufmann , b. , and schaub , t. 2013 .
advanced conflict - driven disjunctive answer set solving . in _
ijcai 2013 , proceedings of the 23rd international joint conference on artificial intelligence , beijing , china , august 3 - 9 , 2013 _ , f. rossi , ed .
ijcai / aaai .
classical negation in logic programs and disjunctive databases .
_ 9 , _ 3/4 , 365386 . ,
greco , s. , and zumpano , e. 2001 . a logic programming approach to the integration , repairing and querying of inconsistent databases . in _ proceedings of iclp01_. lncs , vol .
springer , 348364 . , greco , s. , and zumpano , e. 2003 . a logical framework for querying and repairing inconsistent databases .
_ 15 , _ 6 , 13891408
. \2012 .
a dichotomy in the complexity of consistent query answering for queries with two atoms .
_ 112 , _ 3 , 7785 . ,
pema , e. , and tan , w .- c .
efficient querying of inconsistent databases with binary integer programming .
_ 6 , _ 6 , 397408 . ,
ricca , f. , and terracina , g. 2013 .
consistent query answering via asp from different perspectives : theory and practice .
_ 13 , _ 2 , 227252 .
\2009 . on the consistent rewriting of conjunctive queries under primary key constraints .
_ 34 , _ 7 , 578601 .
certain conjunctive query answering in first - order logic .
_ 37 , _ 2 , 9 .
[ sec : appendix ]
here we report the proofs of theorems and propositions reported in section [ sec : pruning ] .
let us assume that @xmath138 .
this means that @xmath5 is true in every repair of @xmath286 .
since , by definition , for each repair @xmath121 of @xmath287 , there exists a repair @xmath120 of @xmath286 such that @xmath288 , we conclude that @xmath5 must be true also in every repair of @xmath287 .
we we will prove the contrapositive . to this end , let @xmath289 be the bunches of @xmath88 . assume that , for each @xmath290 $ ] , @xmath291 .
this means that , for each @xmath70 $ ] , there exists a repair @xmath292 such that @xmath293 .
consider now the instance @xmath294 } r_i$ ] .
since @xmath289 always form a partition of @xmath1 , since for each @xmath93 , @xmath92 is entirely contained in exactly one bunch , and since each key component of @xmath1 is entirely contained in exactly one bunch , we conclude that @xmath6 is a repair of @xmath1 and @xmath295 .
hence @xmath296 .
@xmath297 if @xmath145 , then by proposition [ prop : fragment ] we have that , for each fragment @xmath134 of @xmath1 , @xmath298 .
moreover , by rephrasing definition [ def : redundant ] , we have that any key component @xmath72 of @xmath1 is redundant if the following condition is satisfied : for each fragment @xmath134 of @xmath1 , @xmath299 .
hence , by combining the two , we conclude that each key component of @xmath1 is redundant .
@xmath300 if each key component @xmath72 of @xmath1 is redundant , by proposition [ prop : redundancyclosure ] , we can conclude that @xmath145 , since the empty database can not entail @xmath5 .
let @xmath134 be a fragment of @xmath1 such that @xmath133 .
by considering @xmath134 as a database and by theorem [ thm : bunches ] , we have that there exists at least a bunch @xmath87 of the conflict - join hypergraph @xmath301 of @xmath134 such that @xmath141 . if @xmath302 , then @xmath303 , and therefore , by proposition [ prop : fragment ] , since @xmath87 is a fragment of @xmath304 , we have that @xmath142 .
if @xmath305 , then let us consider one of the atoms @xmath164 that is not involved in any substitution .
but since @xmath5 is true in every repair of @xmath87 containing @xmath306 , this means that @xmath5 is true also in every repair of @xmath307 . and
since @xmath307 is a fragment of @xmath308 , also in this case we can conclude that @xmath142 .
let @xmath72 be a redundant component of @xmath1 , and @xmath48 be a substitution of @xmath112 such that @xmath159 .
moreover , let @xmath134 be a fragment of @xmath1 such that @xmath133 . since @xmath72 is redundant , by definition [ def : redundant ]
, we have that @xmath142 .
but since @xmath92 necessarily contains an atom of @xmath72 , this means that for each repair @xmath309 , there exists a substitution @xmath157 different from @xmath48 such that @xmath158 .
but since the union of all these substitutions different from @xmath48 can be also used to entail @xmath5 in every repair of @xmath134 , by definition [ def : unfoundedsub ] , we can conclude that @xmath48 is unfounded .
consider , for example , the schema @xmath33 , where @xmath94 , @xmath310 , @xmath311 , and @xmath96 .
consider also the database @xmath312 @xmath313 @xmath314 , and the bcq @xmath315 .
the key components of @xmath1 are @xmath316 @xmath317 and @xmath318 , while the repairs of @xmath1 and @xmath34 are @xmath319 @xmath314 and @xmath320 @xmath314 . moreover , the set @xmath112 contains substitutions @xmath321 and @xmath322 . finally , since @xmath119 maps @xmath5 to @xmath120 , and @xmath122 maps @xmath5 to @xmath121
, we can conclude that @xmath69 .
however , one can observe that @xmath323 could be considered as a safe component with respect to @xmath5 .
in fact , variable @xmath324 of @xmath5 being in a position that does not belong to @xmath325 occurs only once in @xmath5 . and
this intuitively means that whenever there exists a substitution that maps @xmath5 in a repair containing @xmath326 , there must exist also a substitution that maps @xmath5 in a repair containing @xmath327 .
therefore , to avoid that @xmath323 produces two repairs , one can consider only the first two attributes of @xmath328 and modify @xmath5 accordingly .
hence , we can consider @xmath329 , where @xmath330 , @xmath331 and @xmath332 , the database @xmath333 @xmath314 , and the bcq @xmath334 .
clearly , @xmath151 is now consistent and entails @xmath335 .
the benchmark considered in the paper was firstly used in @xcite .
it comprises several instances of varying size of a synthetic database specifically conceived to simulate reasonably high selectivities of the joins and a large number of potential answers .
moreover it includes a set of queries of varying complexity and 40 instances of a randomly generated database . in the following we report the main characteristics of the data set and a link to an archive where the encoding and the binaries of the asp system employed in the experiment can be also obtained .
it contains the following queries organized in groups depending on the respective complexity of cqa ( existential quantifiers are omitted for simplicity ) : * * co - np , not first - order rewritable * * @xmath336 * @xmath337 * @xmath338 * @xmath339 * @xmath340 * @xmath341 * @xmath342 * * ptime , not first order rewritable * * @xmath343 * @xmath344 * @xmath345 * @xmath346 * @xmath347 * @xmath348 * @xmath349 * * first order rewritable * * @xmath350 * @xmath351 * @xmath352 * @xmath353 * @xmath354 * @xmath355 * @xmath356 we used exactly the same datasets employed in @xcite .
it comprises 40 samples of the same database , organized in four families of 10 instances each of which contains 10 tables of size varying from 100000 to 100000 tuples with increments 100000 . quoting @xcite ,
the generation of databases has been done according with the following criterion : for every two atoms @xmath357 , @xmath358 that share variables in any of the queries , approximately 25 of the facts in @xmath357 join with some fact in @xmath358 , and vice - versa .
the third attribute in all of the ternary relations , which is sometimes projected out and never used as a join attribute in table 1 , takes values from a uniform distribution in the range @xmath359 $ ] .
hence , in each relation , there are approximately @xmath360 distinct values in the third attribute , each value appearing approximately 10 times .
we refrain from reporting here all the asp encodings employed in the experiment since they are very lengthy .
instead we report as an example the asp program used for answering query q7 , and provide all the material in an archive that can be downloaded from www.mat.unical.it/ricca/downloads/mrticlp2015.zip .
the zip package also contains the binaries of the asp system employed in the experiment .
let us classify the variables of @xmath361 : * all the variables are : @xmath362 ; * the free variables are : @xmath363 ; * the variables involved in some join are : @xmath364 ; * the variables in primary - key positions are : @xmath365 ; * the variable in idle positions are : @xmath366 * the variable occurring in relevant positions are : @xmath367 [ [ computation - of - the - safe - answer . ] ] computation of the safe answer .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + ` sub(x , y , z , x1,w , d ) ` + ` : - ` + ` r5(x , y , z ) , r6(x1,y , w ) , r7(y , u , d ) . ` + ` involvedatom(k - r5(x ) , nk - r5(v2,v3 ) ) : - sub(x , y , z , x1,w , d ) , r5(x , v2,v3 ) . `
+ ` involvedatom(k - r6(x1 ) , nk - r6(v2,v3 ) ) : - sub(x , y , z , x1,w , d ) , r6(x1,v2,v3 ) . `
+ ` involvedatom(k - r7(y ) , nk - r7(v3 ) ) : - sub(x , y , z , x1,w , d ) , r7(y , v2,v3 ) . `
+ ` confcomp(k ) : - involvedatom(k , nk1 ) , involvedatom(k , nk2 ) ,
nk1 > nk2 .
` + ` safeans(z , w , d ) : - sub(x , y , z , x1,w , d ) , not confcomp(k - r5(x ) ) , ` + ` not confcomp(k - r6(x1 ) ) , not confcomp(k - r7(y ) ) . `
[ [ hypergraph - construction . ] ] hypergraph construction .
+ + + + + + + + + + + + + + + + + + + + + + + + + ` subeq(sid(x , y , z , x1,w ,
d ) , ans(z , w , d ) ) : - sub(x , y , z , x1,w , d ) , not safeans(z , w , d ) . ` + + ` compek(k - r5(x ) , ans ) : - subeq(sid(x , y , z , x1,w , d ) , ans ) . `
+ ` compek(k - r6(x1 ) , ans ) : - subeq(sid(x , y , z , x1,w , d ) , ans ) . `
+ ` compek(k - r7(y ) , ans ) : - subeq(sid(x , y , z , x1,w , d ) , ans ) . `
+ ` insubeq(atom - r5(x , y , z ) , sid(x , y , z , x1,w , d ) ) : - subeq(sid(x , y , z , x1,w , d ) , _ ) . `
+ ` insubeq(atom - r6(x1,y , w ) , sid(x , y , z , x1,w , d ) ) : - subeq(sid(x , y , z , x1,w , d ) , _ ) . ` + ` insubeq(atom - r7(y , d ) , sid(x , y , z , x1,w , d ) ) : - subeq(sid(x , y , z , x1,w , d ) , _ ) . ` + + ` incompek(atom - r5(x , v2,v3 ) , k - r5(x ) ) : - compek(k - r5(x ) , ans ) , ` + ` involvedatom(k - r5(x ) , nk - r5(v2,v3 ) ) . `
+ ` incompek(atom - r6(x1,v2,v3 ) , k - r6(x1 ) ) : - compek(k - r6(x1 ) , ans ) , ` + ` involvedatom(k - r6(x1 ) , nk - r6(v2,v3 ) ) . ` + ` incompek(atom - r7(y , v3 ) , k - r7(y ) ) : - compek(k - r7(y ) , ans ) , ` + ` involvedatom(k - r7(y ) , nk - r7(v3 ) ) . `
[ [ pruning . ] ] pruning .
+ + + + + + + + + ` redcomp(k , ans ) : - compek(k , ans ) , incompek(a , k ) , ` + ` # count{s : insubeq(a , s ) , subeq(s , ans ) } = 0 . `
+ + ` unfsub(s , ans ) : - subeq(s , ans ) , insubeq(a , s ) , incompek(a , k ) , redcomp(k , ans ) . ` + + ` redcomp(k , ans ) : - compek(k , ans ) , incompek(a ,
k ) , ` + ` x = # count{s : insubeq(a , s ) , subeq(s , ans ) } ` + ` # count{s : insubeq(a , s ) , unfsub(s , ans ) } > = x. ` + + ` residualsub(s , ans ) : - subeq(s , ans ) , not unfsub(s , ans ) . ` [ [ fragments - identification . ] ] fragments identification .
+ + + + + + + + + + + + + + + + + + + + + + + + + + ` sharesub(k1,k2,ans ) : - residualsub(s , ans ) , insubeq(a1,s ) , insubeq(a2,s ) , ` + ` a1 < > a2 , incompek(a1,k1 ) , incompek(a2,k2 ) , k1 < > k2 . ` + + ` ancestorof(k1,k2,ans ) : - sharesub(k1,k2,ans ) , k1 < k2 . `
+ ` ancestorof(k1,k3,ans ) : - ancestorof(k1,k2,ans ) , sharesub(k2,k3,ans ) , k1 < k3 . `
+ + ` child(k , ans ) : - ancestorof(_,k , ans ) . `
+ + ` keycompinfrag(k1 , fid(k1,ans ) ) : - ancestorof(k1,_,ans ) , not child(k1,ans ) . `
+ ` keycompinfrag(k2 , fid(k1,ans ) ) : - ancestorof(k1,k2,ans ) , not child(k1,ans ) . `
+ + ` subinfrag(s , fid(kf , ans ) ) : - residualsub(s , ans ) , insubeq(a , s ) , ` + ` incompek(a , k ) , keycompinfrag(k , fid(kf , ans ) ) . `
+ + ` frag(fid(k , ans),ans ) : - keycompinfrag(_,fid(k , ans ) ) . ` [ [ repairs - construction . ] ] repairs construction .
+ + + + + + + + + + + + + + + + + + + + + + ` 1 < = { activefrag(f):frag(f , ans ) } < = 1 : - frag ( _ , _ ) . `
+ + ` 1
< = { activeatom(a):incompek(a , k ) } < = 1 : - activefrag(f ) , keycompinfrag(k , f ) . ` + + ` ignoredsub(s ) : - activefrag(f ) , subinfrag(s , f ) , insubeq(a , s ) , not activeatom(a ) . ` [ [ new - query . ] ] new query .
+ + + + + + + + + + + ` q`@xmath368`(s , z , w , d ) : - safeans(z , w , d ) . ` + ` q`@xmath368`(f , z , w , d ) : - frag(f , ans(z , w , d ) ) , not activefrag(f ) . `
+ ` q`@xmath368`(f , z , w , d ) : - activefrag(f ) , subinfrag(s , f ) , not ignoredsub(s ) , frag(f , ans(z , w , d ) ) . `
we report in this appendix some additional plots . in particular , we provide @xmath369 detailed plots for the overhead of _ pruning_w.r.t . safe answer computation ; @xmath370 scatter plots comparing , execution by execution , _
pruning_with _ bb_and _ mrt _ ; and , @xmath371 an extract of @xcite concerning the overhead measured for the mip - based approach for easing direct comparison with our results .
we report in the following the detailed plots concerning the overhead of _ pruning_w.r.t . the computation of safe answers .
the results are reported in three plots grouping queries per complexity class in figures [ fig : ratiosafe ] .
one might wonder what is the picture if the asp - based approaches are compared instance - wise .
an instance by instance comparison of _ pruning_with _ bb_and _ mrt _ , is reported in the scatter plots in figure [ fig : scatter ] . in these plots a point @xmath372 is reported for each query , where @xmath373 is the running time of _ pruning _ , and @xmath374 is the running time of _ bb_and _ mrt _ , respectively in figure [ fig : scatter : abc ] and figure [ fig : scatter : mrt ] .
the plots also report a dotted line representing the secant ( @xmath375 ) , points along this line indicates identical performance , points above the line represent the queries where the method on the @xmath373-axis performs better that the one in the @xmath374-axis and vice versa .
figure [ fig : comp ] clearly indicates that _ pruning_is also instance - wise superior to alternative methods . | consistent query answering over a database that violates primary key constraints is a classical hard problem in database research that has been traditionally dealt with logic programming .
however , the applicability of existing logic - based solutions is restricted to data sets of moderate size .
this paper presents a novel decomposition and pruning strategy that reduces , in polynomial time , the problem of computing the consistent answer to a conjunctive query over a database subject to primary key constraints to a collection of smaller problems of the same sort that can be solved independently .
the new strategy is naturally modeled and implemented using answer set programming ( asp ) .
an experiment run on benchmarks from the database world prove the effectiveness and efficiency of our asp - based approach also on large data sets . to appear in theory and practice of logic programming ( tplp ) , proceedings of iclp 2015 .
inconsistent databases , primary key constraints , consistent query answering , asp |
although the measurement of photon - number statistics gives an incomplete description of the optical state , it can be helpful to provide fundamental information on the nature of any optical field and hence to discriminate between light of different kinds , either in the classical or non - classical domain .
measuring the photon - number statistics is a difficult task as it requires a complete knowledge of the entire detection process . in some situations , the evaluation of
the first two statistical momenta can be enough for characterizing light statistics .
in particular , mean value and variance , combined into the fano factor , can discriminate between different statistics , even if sometimes the evaluation of higher order momenta is necessary . in this paper
we demonstrate that it is possible to determine the photon - number statistics by means of a direct intensity measurement performed with linear photodetectors endowed with internal gain . in order to bypass the problem of calibrating the detector ,
we devise a measurement technique that takes advantage of the linearity properties of the detector and is based on the measurement of the same state at different mean numbers of photons . in this way
we simultaneously obtain the calibration of the detection chain and the reconstruction of the photoelectron statistics .
our technique has the additional advantage of being applicable to fields in the mesoscopic intensity regime ( less than 1000 photons ) , which is scarcely explored . the competing techniques for determining the photon - number distributions are for instance : tomography with homodyne detection @xcite , measurement with internal - gain photodetectors able to resolve peaks of different photoelectron numbers , plus analysis of peak integrals in order to reconstruct the photoelectron number distribution @xcite . drawbacks of the latter approach are : the low - limit in the mean number of detected photons due to the low quantum efficiency ; the need of independent calibration of the detector ; the need of a guess on the statistics to which the data must be fitted , as there is no direct indication from the measurement @xcite .
if the measurements are performed with linear photodiodes lacking internal gain , the drawbacks is the low - limit to the mean number of detected photons ( p - i - n photodiodes can detect down to one thousand photons upon external amplification @xcite ) .
a further alternative is represented by on / off measurements combined with maximum likelihood reconstruction @xcite .
section [ sec : teo ] presents the theoretical model we will use to interpret the experimental data , section [ sec : exp ] describes the experimental setup used for the measurements and section [ sec : analysis ] presents the technique for analyzing the data and the experimental results .
finally we draw some conclusions .
we can model the detection process as a two step process : photodetection by the photocathode and amplification ( both internal to the detector and external ) ( see fig . [
f : scheme ] ) .
we start from the observation that a real photodetector , having quantum efficiency @xmath0 , behaves like a perfect detector ( @xmath1 ) preceded by a beam spitter ( bs ) with transmittance @xmath2 @xcite .
the photodetection process can thus be described as the convolution of the photon statistics with the bernoullian action of the bs @xcite .
the statistics of the number of photoelectrons emitted by the photocathode , @xmath3 , is thus linked to the statistics of the number of photons , @xmath4 , by @xmath5 if we limit our analysis to the first two momenta of the distributions , the links between the statistics of photons and photoelectrons are given by @xcite @xmath6 where @xmath7 is the mean value and @xmath8 is the variance . the same notation is adopted for photoelectrons . a complete description of the amplification process requires the detailed knowledge , for each particular detector , of the amplification mechanism .
however for the present discussion we limit ourselves to those detectors and to those operating regimes in which the amplification process can be simply considered as linear : we adopt a strongly simplifying approach in which the spread of the single photoelectron peak in the detector output is negligible as compared to its mean value .
the relation linking the statistics of photoelectrons to that of the voltage outputs , even at the end of the electronic chain external to the detector , is then @xmath9 being @xmath10 the amplification / conversion coefficient given by the overall amplification process and signal - processing electronics .
quite obviously , eq . ( [ eq : outs ] ) shows that in general the distribution of the output voltage values is different from that of the photoelectrons .
for the first two momenta of the distributions , the experimental final outputs are given by @xcite @xmath11 where the symbols are defined as above .
the most difficult task to be performed for reconstructing @xmath12 from @xmath13 is the experimental determination of the parameter @xmath10 in eq .
( [ eq : outs ] ) . this task can be tackled by following a procedure based on repeated measurements of the same state of light at different values of the quantum efficiency by inserting neutral filters in front of the detector . in particular
we evaluate the following quantity @xmath14}}{{\alpha \eta \bar{n } } } = \alpha \eta f + \alpha \left ( { 1 - \eta } \right)=\alpha \eta q + \alpha , \label{eq : fano}\end{aligned}\ ] ] where @xmath15 is the fano - factor that we have re - written as @xmath16 , @xmath17 being the mandel q - factor @xcite .
we observe that @xmath18 is linearly dependent on @xmath19 and that @xmath10 can be obtained as the limit of @xmath18 at @xmath20 . by multiplying and dividing by @xmath21
we get @xmath22 which now expresses @xmath18 as a function of the mean value of the measured output voltage .
note that @xmath23 varies because of the variation of @xmath19 and not because of the change in the measured state ( @xmath24 remains the same ) .
the expression in eq .
( [ eq : fanoq ] ) is general and all the information about the specific state under measurement is contained in the angular coefficient @xmath25 , which is zero for poissonian light , positive for classical super - poissonian light and negative for nonclassical sub - poissonian light . by measuring the same light state at different values of @xmath23 and evaluating @xmath18 ,
we can directly obtain the value of @xmath10 and the ratio @xmath24 .
of course , different light states may have , for some values of the parameters , the same value of @xmath24 and this prevents the possibility of the unique determination of the light statistics .
nevertheless , if an assumption can be made on the statistics of the measured light , the method allows one to check it immediately .
the expected expressions of @xmath18 for the states presented in this paper will be given in section [ sec : analysis ] , where the results are discussed .
in fig . [ f : setup ] we show the adopted experimental setup .
the light from the specific source to be investigated is collected by a multimode fiber ( 100 @xmath26 m core diameter , oz optics , canada ) . as the detector we used either a photomultiplier tube ( pmt , 8850 , burle industries , maximum quantum efficiency @xmath27 at 400 nm ) or a hybrid photodiode module ( hpd , h8236 - 40 , hamamatsu , japan , maximum quantum efficiency @xmath28 at 550 nm ) .
both detectors are endowed with partial photon resolving capability and are linear over a wide range of intensities .
the current outputs of the detectors are integrated ( sgi , stanford research systems ) , sampled and digitized ( adc , national instruments ) .
the final outputs are then recorded by a computer . as a preliminary check of the performances of the detectors , we simultaneously measured the light at the two outputs of a 50% beam splitter ( bs in fig .
[ f : setupbis ] ) dividing a ps - pulsed coherent beam at 523 nm .
the light was obtained from the second harmonics of a nd : ylf continuous - wave mode - locked laser amplified at 500 hz repetition rate ( high - q laser production ) . in fig .
[ f : response ] we show a typical pulse - height spectrum of the pmt ( @xmath29 ) and of the hpd ( @xmath30 ) .
the zero value of the output voltage @xmath31 is set at the mean value of the response to dark of each detector , which was measured independently ( see _ e.g. _ the horizontal scale in figs .
[ f : response ] ( @xmath29 ) and ( @xmath30 ) ) .
a standard procedure for analyzing a pulse - height - spectrum is to find a suitable fit obtained as a multiple convolution of the function best fitting the single - photon peak @xcite .
the procedure allows the reconstruction of a limited number of peaks ( that depend on the resolving capabilities of the detectors ) and fails in analyzing more intense fields when the pulse - height spectrum does not show well recognizable peaks .
the pulse - height spectrum of the detectors lose the peak structure as soon as the field intensity becomes mesoscopic , still remaining proportional to the field intensity . in fig .
[ f : linearity ] we show several pulse height spectra recorded by the pmt ( @xmath29 ) and by the hpd ( @xmath30 ) in the same configuration of fig . [ f : setupbis ] , in which neutral density filters are inserted on the beam path before the beam splitter . in fig .
[ f : linearity ] ( @xmath32 ) we plot the mean values of the output voltage as a function of the transmittance of the filters : the difference between the two curves reflects the difference in the overall quantum efficiency of the two detectors at 523 nm .
the linearity will play a key role in the analysis procedure described in section [ sec : analysis ] .
we measured different light fields characterized by different statistics and at wavelengths that were chosen so as to match the highest quantum efficiencies of the detectors at best .
the aim of the measurements is to identify the type of light field , to determine the overall amplification / conversion coefficient @xmath10 and to reconstruct the statistics of the photoelectrons .
the steps of the experimental procedure are : 1 ) to measure the same field at different attenuations of the neutral density filters ( the actual position of the filters depends on the specific measurement ) ; 2 ) to evaluate the mean value and the variance of the output voltage ; 3 ) to plot the experimental values of @xmath18 ( see eq .
( [ eq : fano ] ) ) as a function of the mean value @xmath23 ; 4 ) to fit the plot to a straight line , thus obtain the value of the @xmath10 factor , and to find , according to one of the theoretical models , the parameters of the photoelectron distribution ; 5 ) to use the fitting parameters to recover the photoelectron statistics .
first of all we consider the measurements performed on ps - pulsed coherent light at 523 nm obtained as above ( see fig .
[ f : setup ] ( @xmath29 ) ) . for a coherent field
we expect a poissonian photon - number distribution @xmath33 for which mean value and variance are @xmath21 .
note also that the photoelectron distribution obtained by applying eq .
( [ eq : phel ] ) to eq .
( [ eq : poiss ] ) remains poissonian , so that @xmath34 with the substitution @xmath35 .
the fano factor for the final output voltage becomes @xmath36 independent of the mean value @xmath23 .
the experimental results obtained with the pmt are shown in fig .
[ f : fanoexp ] ( full circles ) .
the linear fit gives @xmath37 v. the second kind of measurements is on the multimode thermal light given , as sketched in fig .
[ f : setup ] ( @xmath30 ) , by a blue portion ( 420 nm ) of the down conversion parametric fluorescence produced by a type - i bbo crystal pumped by the third harmonics pulses ( 349 nm ) of the same nd : ylf laser @xcite .
the expected photon - number distribution is given by the convolution of @xmath26 independent thermal modes @xcite @xmath38 for which the mean value is @xmath21 and the variance is @xmath39 . as in the case of coherent light , the photoelectron distribution obtained by applying eq .
( [ eq : phel ] ) to eq .
( [ eq : multit ] ) remains multimode thermal and again @xmath34 .
the fano factor for the final output voltage turns out to be @xmath40 that is linear in the mean value .
the experimental results obtained with the pmt are shown in fig .
[ f : fanoexp ] ( squares ) .
their fit to eq .
( [ eq : fanother ] ) gives @xmath41 and @xmath42 v. note that , as these measurements were made in the same experimental conditions as those on the coherent field , the two values obtained for @xmath10 are very similar to each other .
we now use the fitting parameters to recover the photoelectron statistics .
we divide the voltage output by the value of @xmath10 to convert it to photoelectrons and then re - bin the obtained distributions in unitary bins . in fig .
[ f : reconstr ] we show the photoelectron distributions , @xmath43 for coherent light ( @xmath29 ) and for multimode thermal light ( @xmath30 ) reconstructed from some of the data sets used to obtain the calibration .
the dots in the figure are the theoretical curves , @xmath12 ( poissonian and multimode thermal , respectively ) , evaluated at the measured mean values , by using the parameters , @xmath10 or @xmath10 and @xmath26 , obtained from the fits to eqs .
( [ eq : fanocoh ] ) and ( [ eq : fanother ] ) , respectively . in the figures the values of the fidelity @xcite of the reconstructed distributions
, @xmath44 are also displayed indicating the good quality of the reconstruction .
the same kind of measurements were repeated by using the hpd . to optimize the detection efficiency we exploited the light produced by the second harmonics of the nd : ylf laser . according to the setup depicted in fig .
[ f : setup ] ( c ) , the laser light was passed through a rotating ground glass diffuser to obtain a pseudo - thermal speckle pattern @xcite .
the light , selected with a pin - hole so as to transmit a number of coherence areas , was then delivered to the detector by a multimode fiber .
the photon - number statistics of the detected field follows the multimode thermal statistics in eq .
( [ eq : multit ] ) . in fig .
[ f : rechpd ] we plot the reconstructed photoelectron distribution for the multimode pseudo - thermal field along with the theoretical curve evaluated from the fit of the fano factor , @xmath45 v and @xmath46 ( see inset of the figure ) . again
the fidelity of the reconstruction is very good .
we have demonstrated that it is possible to implement a self - consistent procedure to recover the distribution of photoelectrons that avoids calibration of the photodetectors .
the procedure employs the evaluation of the fano factor at different values of the mean photon numbers of the field to be characterized .
the procedure has been satisfactorily tested on coherent and multimode thermal light .
the next step will be to investigate less trivial optical states , such as those generated by the mixing of different fields @xcite , in order to extend the validity of the method .
the authors thank sergio cova ( politecnico di milano ) for fruitful discussions .
+ present address of a. agliati is : quanta system , solbiate olona ( va ) , italy . | we demonstrate the possibility of a self - consistent characterization of the photon - number statistics of a light field by using photoemissive detectors with internal gain simply endowed with linear input / output responses .
the method can be applied to both microscopic and mesoscopic photon - number regimes .
the detectors must operate in the linear range without need of photon - counting capabilities . |
the solutions of linear elliptic and parabolic equations , both with cauchy and dirichlet boundary conditions , have a probabilistic interpretation , which not only provides intuition on the nature of the problems described by these equations , but is also quite useful in the proof of general theorems .
this is a very classical field which may be traced back to the work of courant , friedrichs and lewy @xcite in the 20 s . in spite of the pioneering work of mckean @xcite
, the question of whether useful probabilistic representations could also be found for a large class of nonlinear equations remained an essentially open problem for many years .
it was only in the 90 s that , with the work of dynkin@xcite @xcite , such a theory started to take shape .
for nonlinear diffusion processes , the branching exit markov systems , that is , processes that involve diffusion and branching , seem to play the same role as brownian motion in the linear equations .
however the theory is still limited to some classes of nonlinearities and there is much room for further mathematical improvement .
another field , where considerable recent advances were achieved , was the probabilistic representation of the fourier transformed navier - stokes equation , first with the work of lejan and sznitman@xcite , later followed by extensive developments of the oregon school@xcite @xcite @xcite . in all cases
the stochastic representation defines a process for which the mean values of some functionals coincide with the solution of the deterministic equation .
stochastic representations , in addition to its intrinsic mathematical relevance , have several practical implications : \(i ) they provide an intuitive characterization of the equation solutions ; \(ii ) they provide a calculation tool which may replace , for example , the need for very fine integration grids at high reynolds numbers ; \(iii ) by associating a stochastic process to the solutions of the equation , they provide an intrinsic characterization of the nature of the fluctuations associated to the physical system . in some cases the stochastic process is essentially unique , in others there is a class of processes with means leading to the same solution . the physical significance of this feature is worth exploring . a field where stochastic representations have not yet been developed ( and where for the practical applications cited above they might be useful ) is the field of kinetic equations for charged fluids . as a first step towards this goal ,
a stochastic representation is here constructed for the solutions of the poisson - vlasov equation . the comments in the final section point towards future work , in particular on how a stochastic representation may be used for a characterization of fluctuations , alternative to existing methods .
this is what we call _ the stochastic principle_.
consider a poisson - vlasov equation in 3 + 1 space - time dimensions @xmath0 with @xmath1 @xmath2 being a background charge density . passing to the fourier
transform @xmath3 with @xmath4 and @xmath5 , one obtains @xmath6 @xmath7 being the fourier transform of @xmath8 .
changing variables to @xmath9 where @xmath10 is a positive continuous function satisfying @xmath11 leads to @xmath12 with @xmath13 .
eq.([2.6 ] ) written in integral form , is @xmath14 for convenience , a stochastic representation is going to be written for the following function @xmath15 with @xmath16 a constant and @xmath17 a positive function to be specified later on .
the integral equation for @xmath18 is @xmath19 with @xmath20 and @xmath21 eq.([2.9 ] ) has a stochastic interpretation as an exponential process ( with a time shift in the second variable ) plus a branching process .
@xmath22 is the probability that , given a @xmath23 mode , one obtains a @xmath24 branching with @xmath25 in the volume @xmath26 .
@xmath27 is computed from the expectation value of a multiplicative functional associated to the processes .
convergence of the multiplicative functional hinges on the fulfilling of the following conditions : \(a ) @xmath28 \(b ) @xmath29 \(c ) @xmath30 condition ( c ) is satisfied , for example , for @xmath31 indeed computing @xmath32 one obtains @xmath33 this integral is bounded by a constant for all @xmath34 , therefore , choosing @xmath35 sufficiently small , condition ( c ) is satisfied .
once @xmath17 consistent with ( c ) is found , conditions ( a ) and ( b ) only put restrictions on the initial conditions and the background charge .
now one constructs the stochastic process @xmath36 . because @xmath37 is the survival probability during time @xmath38 of an exponential process with parameter @xmath16 and @xmath39 the decay probability in the interval @xmath40 , @xmath41 in eq.([2.9 ] ) is obtained as the expectation value of a multiplicative functional for the following backward - in - time process : starting at @xmath42 , a particle lives for an exponentially distributed time @xmath43 up to time @xmath44 . at its death a coin @xmath45 ( probabilities @xmath46 )
is tossed .
if @xmath47 two new particles are born at time @xmath44 with fourier modes @xmath48 and @xmath49 with probability density @xmath50 . if @xmath51 only the @xmath52 particle is born and the process also samples the background charge at @xmath53 .
each one of the newborn particles continues its backward - in - time evolution , following the same death and birth laws .
when one of the particles of this tree reaches time zero it samples the initial condition .
the multiplicative functional of the process is the product of the following contributions : - at each branching point where two particles are born , the coupling constant is @xmath54 - when only one particle is born and the process samples the background charge , the coupling is @xmath55 - when one particle reaches time zero and samples the initial condition the coupling is @xmath56 the multiplicative functional is the product of all these couplings for each realization of the process @xmath57 , this process being obtained as the limit of the following iterative process @xmath58 } + g_{2}\left ( \xi _ { 1},\xi _ { 1}^{^{\prime } } , s\right ) \\ & & \times x^{\left ( k\right ) } \left ( \xi _ { 1}-\xi _ { 1}^{^{\prime } } , \xi _ { 2}+s% \frac{\xi _ { 1}}{\gamma \left ( \left| \xi _ { 2}\right| \right ) } , \tau -s\right ) x^{\left ( k\right ) } \left ( \xi _ { 1}^{^{\prime } } , 0,\tau -s\right ) \mathbf{1}_{\left [ s<\tau \right ] } \mathbf{1}_{\left [ l_{s}=0\right ] } \\ & & + g_{1}\left ( \xi _ { 1},\xi _ { 1}^{^{\prime } } \right ) x^{\left ( k\right ) } \left ( \xi _ { 1}^{^{\prime } } , 0,\tau -s\right ) \mathbf{1}_{\left [ s<\tau \right ] } \mathbf{1}_{\left [ l_{s}=1\right ] } \end{aligned}\ ] ] then , @xmath59 is the expectation value of the functional .
@xmath60 for example , for the realization in fig.1 the contribution to the multiplicative functional is @xmath61 and @xmath62 with the conditions ( a ) and ( b ) , choosing @xmath63 and @xmath64 the absolute value of all coupling constants is bounded by one .
the branching process , being identical to a galton - watson process , terminates with probability one and the number of inputs to the functional is finite ( with probability one ) . with the bounds on the coupling constants , the multiplicative functional
is bounded by one in absolute value almost surely .
once a stochastic representation is obtained for @xmath65 , one also has , by ( [ 2.8 ] ) , a stochastic representation for the solution of the fourier - transformed poisson - vlasov equation .
the results are summarized in the following : * * theorem 2.1 * * _ _ - there is a stochastic representation for the fourier - transformed solution of the poisson - vlasov equation _
_ @xmath66 _ _ for any arbitrary finite value of the arguments , provided the initial conditions at time zero and the background charge satisfy the boundedness conditions ( a ) and ( b ) . _
_ as a corollary one also infers an existence result for ( arbitrarily large ) finite time .
notice that existence by the stochastic representation method requires only boundedness conditions on the initial conditions and background charge and not any strict smoothness properties .
in the past , the fluctuation spectrum of charged fluids was studied either by the bbgky hierarchy derived from the liouville or klimontovich equations , with some sort of closure approximation , or by direct approximations to the n - body partition function or by models of dressed test particles , etc .
( see reviews in @xcite @xcite ) .
alternatively , by linearizing the vlasov equation about a stable solution and diagonalizing the hamiltonian , a canonical partition function may be used to compute correlation functions @xcite . however , one should remember that , as a model for charged fluids , the vlasov equation is just a mean - field collisionless theory .
therefore , it is unlikely that , by itself , it will contain full information on the fluctuation spectrum .
kinetic and fluid equations are obtained from the full particle dynamics in the 6n - dimensional phase - space by a chain of reductions . along the way , information on the actual nature of fluctuations and turbulence may have been lost .
an accurate model of turbulence may exist at some intermediate ( mesoscopic ) level , but not necessarily in the final mean - field equation .
when a stochastic representation is constructed , one obtains a process for which the mean value is the solution of the mean - field equation .
the process itself contains more information .
this does not mean , of course , that the process is an accurate mesoscopic model of nature , because we might be climbing up a path different from the one that led us down from the particle dynamics .
nevertheless , insofar as the stochastic representation is qualitatively unique and related to some reasonable iterative process , it provides a surrogate mesoscopic model from which fluctuations are easily computed .
this is what we refer to as _ the stochastic principle_. at the minimum , one might say that the stochastic principle provides another closure procedure . | a stochastic representation for the solutions of the poisson - vlasov equation is obtained .
the representation involves both an exponential and a branching process . the stochastic representation , besides providing an alternative existence proof and an intuitive characterization of the solutions ,
may also be used to obtain an intrinsic definition of the fluctuations . |
it has been reported in the last few years the discovery of counterrotating cores in elliptical galaxies ( franx & illingworth 1988 ; bender et al . 1994 ; rix & white 1992 ) and the presence of extended retrograde motions in the discs of early type spirals ( rix et al .
1992 ; merrifield & kuijken 1994 ; bertola et al .
this phenomenon is also present in the sb galaxy ngc 7331 , where the bulge rotates opposite to the disc ( prada et al .
the origin of the stellar counterrotation has been extensively studied and generally attributed to the accretion of a small satellite galaxy or gradual infall of gas in a retrograde orbit ( e.g. merrifield & kuijken 1994 ; thakar & ryden 1996 ) . however ,
other alternatives are possible in terms of dynamical instabilities ( e.g. wozniak & pfenniger 1997 ) , that make it still unclear whether the causes of counterrotation are of external or internal origin . up to now no observational evidence exists of extended stellar counterrotation in barred galaxies , whose discovery would represent a challenge to understand the significance of counterrotation throughout the hubble sequence . having this in mind
, we decided to conduct a photometric and spectroscospic study of the sab(r)a galaxy ngc 5728 ( de vaucouleurs et al .
1991 ) . as a result of this analysis
it will be shown here that extended retrograde motions can also be present in barred systems . in combination with previous observations ,
our results demonstrate that counterrotation is present all along the hubble sequence ; the study of this phenomenon could thus sheds light to understand the mechanisms and the evolution of galaxy formation .
the paper is organized as follows : section 2 gives details of the observations and data reduction .
section 3 presents a brief description of the structure of the galaxy .
section 4 describes the analysis of the stellar line - of - sight velocity distribution ( losvd ) .
section 5 discuss on the nature of the kinematical components present in this galaxy , and finally section 6 gives the discussions and conclusions of this work .
long - slit spectroscopic observations of ngc 5728 were performed on several runs with the double arm isis spectrograph on the 4.2 m wht at la palma . in july 1996 ,
the slit , 1.3@xmath5 wide , was placed along p.a .
30@xmath6 , close to the bar major - axis ( p.a .
33@xmath6 , schommer et al .
we used the red arm of isis with a 1200 @xmath7 mm grating centered at 8750 including the near ir caii triplet .
the dispersion was 0.38 @xmath8pixel@xmath9 .
two consecutive 1800 s exposures were recorded . as velocity template we observed the kiii hr 5826 star .
the seeing was 1.4@xmath5 .
a similar setup was used in a previous run ( february 1995 ) , this time a 1@xmath5 slit wide was placed along p.a .
86@xmath6 , and 23@xmath6 .
a 600 l@xmath10 mm grating was used giving a dispersion of 0.79 @xmath8pixel@xmath9 .
two 1800 s frames were taken centered at 8750 for p.a .
86@xmath6 , and one 1800 s exposure for p.a .
the g8 giant hr 7753 star was observed as velocity template in this run .
the seeing was 0.9@xmath5 . in both
runs the slit length was 4@xmath11 and centered on the apparent optical nucleus of ngc 5728 .
the detector was a @xmath12 tek ccd giving 0.36@xmath5 pixel@xmath9 .
the debias , flatfielding , wavelength calibration and sky subtraction was done in the standard way using the iraf package .
an i - band ccd image of ngc 5728 was taken with the brocam ccd camera on the 2.5 m not at la palma for the analysis of the morphology and comparison with the stellar kinematics .
these observations were done in august 1996 with a seeing of 0.7@xmath5 .
the camera uses a @xmath12 tek ccd with 0.176 @xmath13pixel@xmath9 . in total , we recorded a 40 minutes exposure . figure 1 ( plate 1 ) presents this image ; the bar major - axis and the nuclear features inside the spheroidal component of the galaxy are indicated .
hereafter , we will refer to p.a .
30@xmath6 as the bar major - axis .
an isophotal analysis of the i - band image of ngc 5728 was done using galphot ( jrgensen , franx , & kjaergaard 1992 ) to determine the surface brightness , ellipticity , position angle and the c4 fourier coefficient along the semi - major axis radius ( see figure 2 ) .
the structure of the galaxy can be summarized as follows : a nuclear feature that appears elongated along p.a .
86@xmath6 with a semi - major axis radius of 3.5@xmath5 and a maximum ellipticity of 0.5 ; the stellar bar with a p.a .
33@xmath6 , a length of @xmath360@xmath5 , maximum ellipticity of 0.7 . beyond the end of the bar ,
the ellipticity drops to 0.55 and the disk dominates .
a star forming ring is located at @xmath34@xmath5 ( see schommer et al . 1988 ; wilson et al .
interior to this ring the hst images by wilson et al .
( 1993 ) reveal the nuclear feature as the presence of bar - like continuum knots ( called abcd , named from east to west , beeing a and b very blue ) .
this optical continuum light within the ring can be interpreted in terms of a nuclear stellar bar ( shaw et al .
1993 ; wozniak et al .
1995 ) , as star formation triggered by the radio ejecta or as scattered light from the active nucleus ( see wilson et al .
1993 for a discussion ) . the nature of this nuclear feature will be discussed in the following sections . in the central @xmath320@xmath5
, we have a spheroidal component misaligned with respect to the bar major - axis ( see figure 1 ) .
this is also shown in the ellipticity profile at @xmath310@xmath5 which has a plateau of @[email protected] and an isophotal twist of @xmath38@xmath6 .
the near - ir caii triplet absorption features were used to determine the stellar losvd .
the analysis was performed by means of the two - dimensional unresolved gaussian decomposition algorithm described in prada et al .
( 1996 ) , and prada , gutirrez , & mckeith ( 1998 ) .
the resolution used for the observations presented here , is 2 pixels and 1.4 pixels in the spectral and spatial axes respectively for p.a . 23@xmath6 and p.a .
86@xmath6 , and 3 pixels and 2.2 pixels respectively for the bar major - axis spectra . figure 3 shows on the left the spectra in several positions along the bar major - axis in the spectral region analyzed , i.e. 8450 - 8700 , together with the modeled spectra , obtained by convolution of the derived losvd with the spectrum of the velocity template .
the corresponding losvd are also plotted on the righthand side of figure 3 .
1-sigma error - bars for each losvd were obtained by monte carlo simulations in the following way : first , we consider that the difference between the real and the modeled spectra gives an estimation of the error - bar in each point of the real spectra .
then , we simulate spectra considering that the astronomical signal corresponds to the modeled spectra and the noise follows a gaussian distribution with the errorbar mentioned above .
we run again our algorithm for each simulation obtaining the distribution of the losvd and taking the dispersion of this distribution as an estimation of the uncertainty in each velocity bin of the losvd . finally , we average over all velocity bins at each position .
these error - bars constitute the statistical uncertainty relative to the fit between the original and the recovered spectra and do nt take into account any possible effect due for example to the selection of the stellar templates or modeling the shape of the galaxy s continuum .
we would like to mention that caii in emission is seen in some seyfert 1 galaxies phisically associated with the blr , however , there is no strong indication , generally , that the ca@xmath15 zone is within a separate dynamical entity such as a disk ( see persson 1988 ) .
he also find that the physical conditions are so extreme in the ca@xmath15 zone that the calcium emission can not be a common feature of the spectra of seyfert 1 galaxies and it it would be quite rare to see it in a seyfert 2 .
ngc 5728 is classified as seyfert 2 by philips et al .
( 1983 ) , therefore we consider that the caii in emission does not affect our results at all .
the losvd in the central @xmath16 along the bar major - axis shows two distinct components with their velocity means at opposite sides of the systematic velocity , except in the central 2@xmath5 where the two components are not present . beyond a radius of @xmath312@xmath5 at either side of the nucleus
the losvd consists of a single component .
the counterrotation is also confirmed along p.a .
23@xmath17 where the better seeing during this run allowed to see the two components in the losvd in the central 2@xmath5 ( see figure 4 ) .
while the losvd at p.a . 23@xmath17 and the bar major axis are very similar in the central positions , there are some differences at distances @xmath18 arcsecs from the center .
although the position angles are close we believe that at least partially this could be a consequence of the kinematics of the elongated physical structures associated with the barred component .
we have parameterized the losvd profiles by fitting one , and where possible two gaussians ; this is enough to describe the major kinematical features .
figure 5 shows the results of this analysis plotting the radial velocities as a function of the radius along the bar major - axis , p.a .
23@xmath6 and p.a .
these can be described as a @xmath1-shaped velocity profile in the central 20@xmath5 along the bar major - axis with @xmath2@xmath370@xmath2010 km s@xmath9 and a maximum velocity of @xmath3170 km s@xmath9 at @xmath205@xmath5 .
superimposed on this cold component is a fainter counterrotating component that extends from 2@xmath5 out to 11@xmath5 at either side of the nucleus along the bar major - axis with a @xmath2=95@xmath2010 km s@xmath9 , and a slow rotation reaching @xmath390 km s@xmath9 at @xmath205@xmath5 . in the central 2@xmath5
, a single gaussian was fitted to the losvd with @xmath2=165@xmath2015 km s@xmath9 .
a similar velocity pattern is also present along p.a .
at this position angle we fit two gaussians to the losvd in central 2@xmath5 ( see figure 5 ) .
beyond @xmath320@xmath5 radius the velocity linearly rises with radius and the velocity dispersion is 55@xmath2010 km s@xmath9 .
the kinematics in this region can be attributed to the stellar bar ( see next section ) .
the velocities along p.a .
86@xmath17 basically describe the overall kinematics seen along the bar major - axis , only in the central 4@xmath5 the losvd appears asymmetric .
this might suggest the presence of the counterrotating component seen at this position angle , since the fit of a single gaussian of @xmath2@xmath3160@xmath2015 km s@xmath9 to the losvd agrees with the velocity dispersion of a single gaussian fit to the losvd in the very center along the bar major - axis .
however , higher spectral resolution data is needed to confirm the presence of the two components along p.a .
86@xmath17 . in summary
, we have found in the central parts of the 11 kpc ( we adopt a scale of 1@xmath21 190pc for h@xmath22=75 km s@xmath9mpc@xmath9 ) bar of ngc 5728 a cold ( v/@[email protected] ) prograde @xmath1-shaped velocity component which coexist in the central 4 kpc with a dynamically hot ( v/@[email protected] ) counterrotating component .
a comparison between the photometric and the kinematic components found in the center of the bar in ngc 5728 , should shed some light onto the nature and dynamics of such structures . for this purpose ,
we have compared the surface brightness profile for a cut along p.a .
30@xmath6 obtained from the i - band image with the relative fluxes of the velocity components found in our spectroscopic analysis along this position ( see figure 6 ) .
on the basis of its kinematical and morphological properties we propose to fit the surface brightness profiles of the three velocity components inside 70@xmath5 with three components .
first , we model the bar by a flat bar profile ( @xmath23 , which describes the bar major - axis surface brightness profiles of early type barred galaxies ( elmegreen et al .
the parameters found for ngc 5728 yield a characteristic scale - length of @xmath24 , @xmath25 for the region between 30@xmath5 and 60@xmath5 . for the counterrotating component we propose an @xmath26 law for the fit due to the fact that this is a hot slowly rotating component and can be associated with the spheroidal component present in the central 20@xmath5 of the galaxy ( see section 3 ) .
we get a very good fit with an effective radius of 21.8@xmath5 for the region between 1@xmath5 and 9@xmath5 . finally , the brighter component in this region , i.e. the @xmath1-shaped velocity profile has been fitted by an exponential profile of 4@xmath5 scale - length .
the sum of the three modeled surface brightness profiles agrees very well with the overall radial surface brightness i - band image cut along the bar major - axis ( see figure 6 ) .
the question now is whether the counterrotating core and/or the @xmath1-shaped velocity profile represent the kinematics of the bar in the central regions .
the @xmath1-shaped velocity component has the same v/@xmath2 as the bar kinematics outside a radius of 20@xmath5 , but its surface brightness profile does not follow the flat bar surface brightness profile .
the reason for this might be the contamination by the nuclear feature and the star - forming ring inside the 5@xmath5 radius . following the discussion by wilson et al .
( 1993 ) , we agree with them that the nuclear feature correspond to a blue component that dominates the light in the very central regions ( see figure 5 ) , but does not contribute to the gravitational potential as indicated by its kinematics ( see section 4 ) .
the reason why we state that the @xmath1-shaped component is associated to the barred kinematics is that the gaseous velocity field in ngc 5728 ( see rubin 1980 ; schommer et al .
1988 ) shows also an @xmath1-shaped feature and follows the stellar kinematics at radii larger than 20@xmath5 .
the gaseous kinematics has been interpreted by schommer et al .
( 1988 ) in terms of models of gas motions in non - axisymmetric potentials and they claim that the star - forming ring and the @xmath1-shaped velocity field may be related to an inner lindblad resonance .
the gaseous kinematics also shows an expansion of @xmath3400 km s@xmath9 in the central 3.5@xmath5 but it does not show any counterrotating signature .
however , the @xmath1-shaped velocity profile can also be interpreted as a very fast rotation due to a heavy mass concentration in the center of the bar in ngc 5728 . from the measured velocities in the very center
, we can estimate an mass of @xmath34 10@xmath27 m@xmath28 inside a radius of 300 pc for an inclination of 48@xmath6 ( schommer et al .
this mass could be related to its seyfert nucleus .
several scenarios have been proposed to explain kinematical structures similar to the ones found in ngc 5728 : dynamical instabilities , the accretion of a satellite or gas infall with retrograde orbits . in the first case a non - axisymmetric structure such as
a bar modifies the stellar orbits , introducing retrograde motions in its center , as shown in numerical simulations by wozniak & pfenniger ( 1997 ) . in their models ,
the presence of retrograde orbits ( almost circular and perpendicular to the major - axis of the bar ) inside the corotation radius , can explain local counterrotation or even could support a counterrotating core .
one can expect that , if the mass on such orbits increases , a critical mass ratio can be reached beyond which the dynamics of the counterrotating component are decoupled from the direct bar . in this scenario the presence of the counterrotating core in ngc 5728 could be interpreted as a secondary retrograde bar , and its presence could thus be intrinsic to the primary bar . indeed , n - body simulations by friedli ( 1996 ) have shown that a counterrotating secondary stellar bar can be stable .
however , we can not rule out an external origin for the presence of retrograde motions in the center of ngc 5728 as shown by the models of thakar & ryden ( 1996 ) .
they explain the counterrotation of the disks of spirals by the accretion of small satellites or gas with negative angular momentum .
we conclude that the data presented in this paper reveal for the first time the presence of extended retrograde motions in barred galaxies which , together with previous discoveries in ellipticals , early - type spirals and sb galaxies ( see bertola & corsini 1997 for a review ) , seems to indicate that counterrotation is a phenomenon present all along the hubble sequence .
the scenarios mentioned above could explain these , or alternatively a more general physical mechanism that allows for galaxy transformation processes along the hubble sequence might be needed . in any case , the relevance of the phenomenon is clear and requires for a deeper understanding , more statistics and two - dimensional spectroscopy .
this research is based on observations obtained at the wht , operated by the ing and the not at the observatorio del roque de los muchachos ( la palma , spain ) of the iac .
we want to thank a. manchado for taking the not image and e. prez , h. zhao and the referee for useful comments on the manuscript .
bender , r. , saglia , r. p. & gerhard , o. e. 1994 , mnras 269 , 785 bertola , f. , cinzano , p. , corsini , e. m. , pizzella , a. , persic , m. & salucci , p. 1996 , apj , 458 , l67 bertola , f. , & corsini , e.m .
1997 , in _ galaxy interactions at low and high redshift _ , iau symposium 186 , kyoto , in press de vaucouleurs , g. , de vaucouleurs , a. , corwin , h. g.,jr , buta , r. j. , paturel , g. & fouqu , p. 1991
, third reference catalog ue of bright galaxies , springer verlag elmegreen , b. , elmegreen , d. m. , chromey , f. r. , hasselbacher , d. a. , & bissell , b. a. 1996 , anj , 111 , 2233 franx , m. & illingworth , g. 1988 , apj , 327 , l55 friedli , d. 1996 , a&a , 312 , 761 jrgensen , i. , franx , m. & kjaergaard , p. 1992 , a&a s , 95 , 489 merrifield , m. r. & kuijken , k. 1994 , apj , 432 , 575 persson , s. e. 1988 , apj , 330 , 751 phillips , m. m. , charles , p. a. , & baldwin , j. a. 1983 , apj , 266 , 485 prada , f. , gutirrez , c. m. , peletier , r. f. & mckeith , c. d. 1996 , apj , 463 , l9 prada , f. , gutirrez , c. m. & mckeith , c. d. 1998 apj , 495 , 765 rix , h .- w . & white , s. d. m. 1992 , mnras , 254 , 389 rix , h - w .
franx , m. fisher , d. & illingworth , g. 1992 , apj , 400 , l5 rubin , v. 1980 , apj , 238 , 808 shaw , m. , combes , f. , axon , d. j. & wright , g. s. 1993 , a&a , 273 , 31 schommer , r. a. , caldwell , n. , wilson , a. s. , baldwin , j. a. , phillips , m. m. , williams , t. b. & turtle , a. j. 1988 , apj , 324 , 154 thakar , a. r. & ryden , b. s. 1996 , apj , 461 , 55 wilson , a. s. , braatz , j. a. , heckman , t. m. , krolik , j. h. , & miley , g. k. 1993 , apj , 419 , l61 wozniak , h. , friedli , d. , martinet , l. , martin , p. & bratschi , p. 1995
, a&a , 111 , 115 wozniak , h. & pfenniger , d. 1997 , a&a,317,14 | we present a detailed study of the stellar kinematics in the barred galaxy ngc 5728 based on i - band photometry and long - slit spectroscopic observations in the region of the near - ir ca ii triplet .
the analysis of the stellar line - of - sight velocity distribution ( losvd ) has revealed , in the central regions of the bar , the presence of a cold ( v/@xmath02.5 ) prograde @xmath1-shaped velocity component that coexists in the central 4 kpc with a fainter and hotter ( v/@[email protected] ) counterrotating component . beyond 4 kpc from the nucleus
the losvd shows the stellar bar kinematics .
the comparison of the radial surface brightness profile of the velocity components with that obtained from an i - band image shows that the counterrotating core follows a r@xmath4 profile , while the @xmath1-shaped component does not follows the flat bar surface brightness profile .
several possible scenarios accounting for such kinematic signatures found in the center of the bar in ngc 5728 are discussed .
the data presented in this paper shows for the first time the presence of extended retrograde motions in barred systems which , together with previous discoveries seems to indicate that the stellar counterrotation is a phenomenon present all along the hubble sequence .
# 1*comment by rp : # 1 * |
exact diagonalizations are restricted to small spin systems ( up to about @xmath1 spins with the present day computers ) .
some problems as the study of critical regimes are out of reach of such an approach , on the contrary we think that this approach might help in answering other subtle questions as for example symmetry breaking , order out of disorder , qualitative nature of the ground - state and of the first excitations of two - dimensional ( 2d ) spin-1/2 magnets . we have used it extensively to study the competition of long range order ( lro ) versus `` quantum disorder '' at @xmath0 . up to now , by looking to the spectra of model spin systems , we have been able to characterize four kinds of behavior : nel lro ( collinear or not ) , dimer lro and two different spin - liquids . in this paper
we will give a brief picture of these four generic cases of 2d magnetism , underline the more efficient numerical criteria to determine the nature of a given unknown system and try to extract from the present numerical knowledge what are the conditions for the appearance of the four different phases .
the key point of our approach is the analysis of the symmetries and finite size scaling of the low lying eigen - levels of a given hamiltonian on small samples with periodic ( or eventually twisted ) boundary conditions .
the evolution of the spectrum with the size and shapes of the sample has to be thoroughly understood before conclusion : it may be a lengthy and sometimes subtle process but it is usually quite valuable .
some examples will be quoted in the following sections .
there is no mathematical proof that nel long range order exists at @xmath0 in 2d spin-1/2 systems .
but there is a very large consensus that it should be the case for the heisenberg nearest neighbor hamiltonian on bipartite lattices @xcite .
exact diagonalizations and/or monte - carlo calculations were performed for the heisenberg model on the square lattice@xcite , on the triangular lattice@xcite and on the honeycomb lattice@xcite . on each of these lattices
the heisenberg model displays long range order with a sublattice magnetization reduced to @xmath2 of the classical value on the square lattice ( @xmath3 on the triangular lattice@xcite , @xmath4 on the honeycomb lattice@xcite ) .
these values of the order parameter might be in error by a few percent : in these antiferromagnets with linear goldstone modes , the scaling law for the magnetization of samples of linear size @xmath5 is an expansion in @xmath6 ( i.e. @xmath7 in 2d)@xcite . for the sizes encountered in exact diagonalizations ,
the asymptotic @xmath7 law is never reached @xcite and the extrapolation to the thermodynamic limit remains uncertain : nevertheless the agreement between the scaling behaviors of diagonalization results and spin wave results gives a strong support to the above - mentioned conclusions .
a stronger confirmation of nel lro is obtained by the scaling of the spectrum itself , which is an illuminating illustration of the _ mechanism of symmetry breaking_. in order to understand in a simple way how a continuous symmetry breaking manifests itself on finite size spectra , let us consider a piece of solid .
our experience of the macroscopic solid is the following : it breaks the translation symmetry and its center of mass ( c.o.m . )
is localized in space . on the other hand
, we can sort its degrees of freedom in two parts : first the c.o.m .
coordinates and the internal variables .
the hamiltonian of such a problem breaks in two commuting parts : the c.o.m . kinetic energy and the hamiltonian of the internal degrees of freedom ( in a solid , in a first approximation
this second part is a phonon hamiltonian ) . for a finite size sample of n atoms , with mass m , this reads : @xmath8 the total hamiltonian does not break translation symmetry , the total momentum @xmath9 is a constant of the motion and its eigenvalues good quantum numbers . due to the commutation property of the sub - parts of the hamiltonian ,
the eigenstates are direct products of eigenstates of the c.o.m .
kinetic energy and phonons eigenstates . from these very simple observations
it follows that the spectrum of eq . 1 as a function of the eigenvalues of the square of the momentum has the appearance described in fig
a straight line is drawn across the eigen - energies of the c.o.m .
hamiltonian with @xmath10 phonon ( respectively @xmath11 phonon ... ) .
let us first concentrate on the vacuum of phonons : the associated eigen - levels collapse to the absolute ground - state as @xmath12 when n goes to infinity .
let us now consider the eigenlevels immediately above this family : they correspond to the softest phonon excitation and are distant from the vacuum family by the energy of this phonon , that is an energy @xmath13 .
when the size goes to infinity , _ if the dimension of the solid is equal or larger than two _ , the levels forming the vacuum of phonons collapse to the absolute ground - state faster than the softest phonon ! this allows to form a wave packet out of these levels and localize the c.o.m . with a cost in energy lower than the softest phonon : the solid phase with its specific macroscopic properties ( symmetry breaking and stiffness )
can exist in 2d ( at @xmath0 ) , whereas it can not exist quantum mechanically in 1d .
let us now turn to the nel ordered magnet .
for simplicity we will first consider a collinear magnet : the macroscopic state of the magnet at @xmath0 is defined by the euler angles of the order parameter .
they are the equivalent of the c.o.m .
coordinates in the solid problem .
the conjugate variables of these euler angles are the components of the total spin and their free precession is described by the hamiltonian : @xmath14 this is the equation of motion of a rigid rotor .
the moment of inertia of this rotor is written here in terms of the intensive homogeneous spin susceptibility @xmath15 , @xmath16 is the number of spins .
if the system has lro and rigidity , the motion of the internal variables is governed by an hamiltonian of ( antiferromagnetic ) magnons and the total hamiltonian reads : @xmath17 when @xmath18 is @xmath19 invariant , @xmath20 is a constant of motion and its eigenvalues @xmath21 are good quantum numbers .
it is thus interesting to display the spectrum of this ordered antiferromagnet as a function of @xmath21 .
its generic appearance is shown in fig .
2 . one recognizes the ( pisa-)tower of states associated with the vacuum of magnons @xmath22 ( these states are called qdjs for quasi degenerate joint states in refs .
@xcite ) , and the tower of states associated with the softest magnon @xmath23 .
the semi - classical nel ground - state with an order parameter pointing in the @xmath24 direction is a wave packet of eigen - levels with different total spin belonging to the qdjs @xmath22 .
mixing different s values up to @xmath25 allows the localization of the order parameter .
if the levels of the @xmath22 family collapse to the ground - sate faster than the softest magnon , the classical picture survives to quantum fluctuations and the system exhibits lro : as for the solid , that might be the case at @xmath0 for 2d magnets . for @xmath26 , thermal fluctuations
destroy the order ( mermin - wagner theorem ) .
notice that the entropy of the semi - classical ground - state scales as @xmath27 , and the entropy per spin of the ordered ground - state is zero in the thermodynamic limit .
the levels associated to the vacuum of magnons have special properties with respect to the lattice symmetry group : these properties reveal the space symmetry breaking of the nel state .
let us take as an example the collinear order on the honeycomb lattice ( fig .
3 ) : it does not break the translational symmetry of the bravais lattice ( thus only the @xmath28 momentum appears in the qdjs ) .
but it breaks both the inversion symmetry with respect to the center of an hexagon and the reflexion with respect to an axis going through nearest neighbor hexagon centers , and it is invariant in the product of these two symmetries .
thus both the even and odd irreducible representations of these two symmetry groups appear in the qdjs ( symbols @xmath29 , @xmath30 ) , but they are always associated so that @xmath31 which insures the invariance in the product of the two symmetries , and @xmath32 which insures the invariance of the semi - classical ground - state in the product of a reflexion times a spin - flip .
this is the simplest situation for an antiferromagnet . on the square lattice ( which is equally bipartite )
nel order is collinear and eq .
3 remains valid , but as the translation symmetry is broken , both the @xmath33 and @xmath34 vectors appear among the qdjs .
a more complex order parameter implies a richer structure of the qdjs : in the case of the heisenberg model on the triangular lattice the order parameter is no longer a vector but a trihedron ( or equivalently a rotation matrix ) .
eq.3 is then more complex , involving both the in - plane and out of plane susceptibilities , and there is an internal generation of an extra quantum number ( the projection of the total spin on the helicity of the magnet ) .
the free precession hamiltonian is that of a symmetric top @xcite and @xmath35 states collapse to the ground - state in the thermodynamic limit .
( quantum mechanically the qdjs associated to p - sublattice nel order are adiabatically connected to the states obtained by addition of p spins of size @xmath36 . ) a thorough analysis of the symmetry breaking associated to nel order on the triangular lattice is done in @xcite .
unexpectedly the mechanism of `` order by disorder '' can be detected by looking at the finite size scaling of relatively small samples @xcite .
we investigated this mechanism in the @xmath37 model on the triangular lattice .
for @xmath38 the classical ground - state exhibits a large degeneracy associated to different kinds of 4-sublattice nel order . for small sizes ( @xmath39 )
the quantum fluctuations are weak and the vacuum of magnons involve all the levels associated with addition of four momenta . in the @xmath40 sample quantum fluctuations are already strong enough to lift the degeneracy of this family :
the levels characteristic of the 2-sublattice nel order remain the lowest ones in the spectrum , the other ones drift towards the continuum of excitations .
it is a nice illustration of the fact that long wave length quantum fluctuations tend to restore the symmetries of the original hamiltonian .
the quantum ground - state is more symmetric than the classical one .
order parameters , spin susceptibilities @xcite , stiffnesses@xcite can be extracted from the numerical data : they are interesting pieces of data to be compared to the spin waves calculations at finite sizes , but as it has been discussed in the beginning of this section , the leading term of their finite size scaling is @xmath41 and therefore it is more significant to look to the tower of qdjs which collapse as @xmath42 .
it is also simpler to look at the symmetries of the low lying levels and deduce the symmetry of the order parameter than to determine it by direct computations . as a consequence when looking at the disappearance of nel lro due to extra frustration or quantum fluctuations it is preferable to pinpoint the range of parameters where the tower of states is destroyed than search for the vanishing of the order parameter@xcite
the second class of long - range ordered systems does not imply @xmath19 symmetry breaking but only space symmetry breaking .
we will call these systems valence bond crystals ( vbc ) : ( with z the coordination number of the lattice ) . in such circumstances
there exists a whole class of hamiltonians which have a unique highly symmetric ground - state where each site shares a singlet with any neighbor connected to it . ]
their ground - states are characterized by regular arrays of singlet states on simple bonds or more complicated subsets ( 4-site plaquettes ... ) . in the thermodynamic limit
the ground - state of a vbc is degenerate , with a finite degeneracy directly related to the symmetry breaking .
the vbc on the @xmath37 chain ( for @xmath43 ) is doubly degenerate ( it breaks the one - step translation symmetry of the lattice ) .
the simplest vbcs that could be encountered on a triangular lattice have a 12-fold degeneracy ( a factor 4 comes from the translation symmetry breaking and a factor 3 from the breaking of @xmath44 ) .
it is a common belief that a vbc has a gap to all excitations ( except indeed at critical points ) . in 2d , the more convincing evidence of such a phase , with a spontaneous discrete symmetry breaking has been seen on the @xmath37 model on the square lattice ( see @xcite and refs . therein ) . in that case
the vbc breaks translational symmetry but probably not @xmath45 : it is probably a long range order of @xmath46 plaquettes .
we have a new example of a vbc in the spectrum of the @xmath37 model on the honeycomb lattice .
when the second neighbor coupling is strong enough the frustration destroys the collinear nel long range order and gives birth to a pattern of valence bonds ( with a @xmath47 symmetry breaking associated to the three spatial directions for the dimerized bonds ) .
our predictions is based on the scaling of the low lying singlets of the spectra , it is in agreement with the dimer - dimer correlations functions measured in the first singlet levels supposed to be degenerate in the thermodynamic limit@xcite .
the nature of the first excitations of these models ( coherent or incoherent magnons ) remains to be elucidated .
besides these situations with symmetry breaking and lro in a local observable , we have found two other behaviors , probably generic , with no obvious symmetry breaking , no long range order in a local order parameter .
we call these situations `` spin - liquids '' and will successively describe both of them .
let us first describe very quickly the simplest one which has the following properties ( an extensive description of the properties of this phase is given in @xcite ) : * a gap to any excitations * all spin - spin correlations are short ranged * a four - fold degeneracy of the ground - state on a 2-torus we have observed this phase for the first time in the multiple - spin exchange problem .
this effective hamiltonian has been introduced by eyring and thouless to describe the magnetism of quasi localized fermions ( electrons in the wigner crystal on one hand , @xmath48 atoms in solid phase on the other ) .
the hamiltonian of the multiple - spin exchange model is given by @xmath49 where @xmath50 are the @xmath51-spin exchange tunneling probabilities ( exchange coefficients ) , @xmath52 and @xmath53 are the @xmath51-spin permutation operators and their inverse .
the alternate sign in the summation over @xmath51 in eq .
[ hamilt ] comes from the permutation of fermions . in general , the exchange coefficients decrease with increasing @xmath51 .
the two - spin exchange term gives the heisenberg hamiltonian up to a constant : @xmath54 where @xmath55 and @xmath56 are spins localized at site @xmath57 and @xmath58 , respectively .
the three - spin exchange operator is equal to a sum of two - spin exchange operators @xcite .
thus , the two and three - spin exchange terms are described ( up to a constant ) by an heisenberg hamiltonian with an effective coupling constant @xmath59 .
cyclic four - spin exchange gives birth to new physics : it is quartic in spin operators and has built - in frustration . as a result the ground - state of this 4-spin exchange hamiltonian on a single plaquette is doubly degenerate with total spin @xmath46 and @xmath60 ! at the classical level this hamiltonian displays various antiferromagnetic phases@xcite , which are wiped out by quantum fluctuations , leading in a large range of parameters to the spin - liquid with the above - mentioned properties @xcite .
the four - fold degeneracy of the ground - state on a 2-torus is a topological property which was predicted more than ten years ago@xcite .
contrarily to oshikawa@xcite we do not think that it is the signature of long range order in a local order parameter ( see discussion in misguich@xcite ) .
very few things are clearly stated on the excitations of this spin - liquid : * plateaus are expected in the magnetization curves both for zero and half the saturation magnetization .
* from the finite size spectra it might be suspected that the triplet excitations form a `` continuum '' above the spin - gap and that an @xmath46 bound - state exists just below the spin - gap . up to now , a clear signature of deconfined spinons has not been found in this spin - liquid .
futher investigations appear necessary . as a further hint that might be of importance : short range correlations are slightly ferromagnetic .
in fact this spin - liquid phase appears in the vicinity of a ferromagnetic instability , as can be seen both in the case of the multiple - spin exchange hamiltonian on the triangular lattice , but also for the @xmath37 model on the honeycomb lattice@xcite .
experimentally , this phase might be observed in the low density 2d layers of solid @xmath48 where there is evidence that the coupling constants might be in the good range ( see @xcite and references therein ) and possibly in the wigner crystal@xcite .
the last situation that we will now discuss is by some details more puzzling .
it has been obtained by depleting the triangular lattice , removing one spin out of four : the lattice which is obtained is called the kagom lattice .
some possible physical realizations are @xmath61 or various jarosites @xcite .
the model has been extensively studied in the classical limit @xcite and it displays an extensive local degeneracy ( i.e. an extensive number of modes of zero energy ) .
some authors have pleaded in favor of a selection of a particular ground - state by the `` order out disorder '' mechanism @xcite but the finite size scaling of spin-1/2 spectra contradicts this assumption @xcite .
the system is certainly a spin - liquid in the following sense : * all the correlations ( spin - spin , dimer - dimer , chiral - chiral , ... ) are very small and decay with distance @xcite . *
the system has a gap for magnetic excitations . even if it is not sure that the largest size ( n=36 ) is much larger than all correlation lengths samples seems to be in a cross - over regime of scaling , but the energy per spin still evolves as @xmath62 ! ] , the finite size scaling studies point to a small spin - gap : probably of the order of one tenth of the coupling constant and quite likely larger than one twentieth of the coupling constant @xcite . *
the spin - gap is filled by a continuum of singlets and the number of these singlets increases exponentially with the system size @xmath16 @xcite as shown in fig.4 .
the understanding of the low lying continuum of singlets is still incomplete but important progresses have been done recently @xcite .
mambrini and mila @xcite have shown that the low lying singlet levels could be related to a family of short range dimer coverings of the lattice .
it must be emphasized that short range dimer coverings on this lattice is not necessarily synonymous with exponential decreasing correlations and gapful excitations : the recent results of mambrini and mila prove that this is indeed not the case and the comparison between these results and those of zeng and elser @xcite point to the fact that the non - orthogonality of the dimer coverings plays a major role in this respect .
this is an important step forward and we would like now a simple picture of the first excitations .
ten years ago , rokhsar and kivelson have studied a model of hard - core dimers on the square lattive @xcite .
the phase diagram of their model displays two vbc phases separated by a critical point . at the transition point between the two phases
the spectrum of excitations becomes gapless .
the low lying modes are the goldstone modes associated to a quasi lro in the orientation of the dimers .
let us assume that this hypothesis is qualitatively valid on the kagom lattice and derive its consequences for the number of states below the spin - gap .
assuming that these goldstone modes are non - interacting , it is easy to extract from their dispersion law ( @xmath63 ) , the internal energy and entropy versus temperature and thus the number of states below the first triplet excitation .
the logarithm of this number of states scales as @xmath64 .
in fig.4 we compare the numerical results with this predicted law ( assuming @xmath65 or @xmath66 ) . fixing the unknown constants from the measured value for @xmath67 , we see that the prescription is an order of magnitude off at @xmath1 : the @xmath65 mode description gives @xmath68 singlet levels in the spin - gap , whereas their real number is @xmath69 ! with the present pieces of information we must conclude that the description of the low lying singlets as goldstone modes associated to a quasi lro in the dimer covering is not supported by the numerics .. due to the intrinsic quantum noise of the numerical density of states , it is impossible with this last criterium to really discriminate between the different hypotheses , but the density of states implied by a mode description @xmath70 seems less probable than a pure exponential or power law ( the @xmath71 criterium is approximately twice as large ) . ]
a simple picture of these first excitations is still lacking ! on the other hand , the exponential increase in the number of eigenlevels in a finite band of energy implies a finite entropy per spin at t=0 .
the same phenomena appearing in each spin sector might be at the origin of the anomalous glassy behavior `` without chemical disorder '' seen in various compounds ( see @xcite and refs therein ) .
less speculative , the presence of a finite entropy in the singlet channel at low temperature is consistent with the observation of ramirez and coworkers@xcite that the low temperature specific heat of @xmath61 is essentially insensitive to large magnetic fields ( much larger than the temperature)@xcite .
this picture is also compatible with the muons experiments of uemura _ et al _ @xcite and the elastic spin diffusion measurement of lee _ et al _ @xcite .
to conclude we must add that these systems seem able to support unconfined spinons excitations with non zero chern numbers@xcite .
up to very recently , we were thinking that this second kind of spin - liquid was associated to special lattices as the kagom or pyrochlore lattices , which at the classical level display an infinite local degeneracy .
some recent results bring new pieces of information : we have observed kagom - like spectra on the triangular lattice when the 3-sublattice nel order is destabilized by a frustrating 4-spin exchange@xcite and we have no evidence of an infinite local degeneracy at this point .
we now conjecture that the interesting feature common to these two cases is the appearance of the `` kagom - like '' phase after destabilization of a non - collinear long - range order ( either due to quantum fluctuations -when going from the triangular to the kagom lattice by weakening of bonds@xcite- , or to competing frustrating interactions in the multiple - spin exchange problem@xcite ) .
contrarily , destabilization of a collinear order ( like in section ii ) leads to dimer lro .
is this `` kagom - like behavior '' representative of a true new phase or of a quantum critical point ? this remains an open question . in the two studied cases we found these characteristics in a rather wide range of parameters : it is nevertheless impossible to exclude a finite size effect .
if criticality is to be viewed as a situation where there is no intrinsic length scale nor energy scale and a slowing down of all dynamics , it remains unclear if the proliferation of low lying singlets in this model could not be a signature of criticality ( and the above assumption on the scaling of the density of states much too naive ! ) .
as a last remark we would like to discuss some questions around the words and concepts of order and disorder .
the established expression of `` order by disorder '' ( or `` order out of disorder '' ) expresses the fact that quantum fluctuations ( as well as thermal fluctuations for which it was primarily introduced by villain ) stabilize the simpler and the more symmetric ordered state amongst the allowed @xmath0 classical ground - states . but more generally the effect of long wave - length quantum fluctuations on a ( semi-)classical ground - state might lead to still simpler and more symmetric situations which do not break @xmath19 ( section iii ) and even any space symmetries ( sections iv and v ) . in such an issue the word `` quantum disorder '' has been used ( by opposition to nel long range order ) but indeed it should be used with great caution : the ground - state of the spin - liquid of section iv is highly symmetric and very far from a disordered system .
only the `` kagom - like '' state might be qualified as `` quantum disordered '' ( section v ) . to measure this delicate issue one may look at the @xmath0 entropy per spin of the ground - state . in the thermodynamic limit
, this quantity scales as @xmath72 for a nel ordered state with @xmath73 sublattices , as @xmath74 in the valence bond crystals ( sec .
iii ) and in the fully gapped spin - liquid ( sec .
iv ) , and it is @xmath75 in the `` kagom - like '' spin - liquids . in conclusion
exact diagonalizations have allowed a characterization of 4 different and ( probably ) generic ground - states of quantum antiferromagnets in 2d .
the two kinds of spin - liquids conjectured by p.w .
anderson in 1973@xcite have been found ( the second with the continuum of singlets seems nevertheless rather different from the initial conjecture ) .
possible physical realizations of these spin - liquids exist .
there remain many questions on the first excitations of these spin - liquids .
some progress on this subject can probably be done in computing dynamical structure functions with exact diagonalizations schemes , but it would also be useful to devise new analytical approximate tools to look at these questions : exact results on the ground - state and low lying levels of the spectra may hopefully help in achieving this goal !
v. n. kotov , j. oitmaa , o. sushkov , and z. weihong , phys .
b * 60 * , 14613,14616 ( 1999 ) ; l. capriotti and s. sorella , phys . rev .
lett . * 84 * , 3173 ( 2000 ) ; o. p. sushkov , j. oitmaa and z. weihong , condmat/0007329 ; m. s. l. du croo de jongh , j. m. j. van leeuwen and w. van saarloos , condmat/0002116 .
this discussion was opened by the large @xmath16 study : n. read and s. sachdev , phys .
lett . * 66 * , 1773 ( 1991 ) ; s. sachdev and n. read , phys .
* 77 * , 4800 ( 1996 ) . | we describe the four kinds of behavior found in two - dimensional isotropic quantum antiferromagnets .
two of them display long range order at @xmath0 : the nel state and the valence bond crystal .
the last two are spin - liquids .
properties of these different states are shortly described and open questions are underlined .
pacs numbers : 75.10.jm ; 75.50.ee ; 75.40.-s |
every knot @xmath10 can be parametrized by polynomials on the open set @xmath11 , where @xmath12 for some point @xmath13 .
this was shown by shastri @xcite using the weierstrass approximation theorem . however , finding a parametrization for a specific knot @xmath14 is difficult .
the standard method has been to first find a parametrization @xmath15 of a regular plane projection of @xmath16 , and to then find a third polynomial @xmath17 giving the correct over- and under - crossings at the nodes of the projection .
consequently , the list of knots for which explicit polynomial parametrizations have been found to date is finite . a set of knots @xmath18 is _ topologically infinite _ , or _
t - infinite _ , if and only if @xmath18 contains a sequence @xmath19 , @xmath20 , such that @xmath19 and @xmath21 are distinct knot types when @xmath22 .
otherwise , @xmath18 is _ t - finite _ ,
i.e. , elements of @xmath18 represent a finite number of knot types .
the results presented in this paper include the following .
* let @xmath2 denote the degree-@xmath3 chebyshev polynomial of the first kind over a field @xmath0 of characteristic zero .
given positive integers @xmath5 , define their _ pairwise greatest common divisor _ to be @xmath23 _ proposition [ embedding ] _ asserts that @xmath24 is an algebraic line embedding if and only if @xmath25 .
* _ proposition [ reduced ] _ shows that , for a fixed pair of positive integers @xmath9 with @xmath26 , the infinite set @xmath27 represents a finite number of algebraic equivalence classes .
this implies that @xmath28 is t - finite over the field @xmath29 of real numbers . *
_ proposition [ t - infinite ] _ gives an explicit polynomial parametrization for each member of the t - infinite family of alternating @xmath9-knots , using chebyshev polynomials of both the first and second kind .
the paper is organized in the following way .
* section 2 : the chebyshev polynomials @xmath2 * section 3 : the curves @xmath8 * section 4 : line embeddings @xmath7 * section 5 : nodal curves and crossing sequences * section 6 : open problems the literature on polynomial line embeddings in 3-space is scant . following the appearance of the epimorphism theorem in the mid-1970s , which characterizes embeddings of the line in a plane , abhyankar considered the family of embeddings of the form @xmath30 for positive integers @xmath31 @xcite .
abhyankar conjectured that these embeddings are not rectifiable when none of the three integers belongs to the semigroup generated by the other two .
however , craighero showed that @xmath32 and @xmath33 can be rectified @xcite ; see @xcite , 4.3.1 , for another proof .
subsequently , bhatwadekar and roy gave families of integers @xmath31 for which @xmath34 can be rectified @xcite .
the general case remains open , including @xmath35 .
the lines defined by @xmath34 are topologically trivial over the field of real numbers . in a well - known paper @xcite ,
vassiliev observed that any `` non - compact '' knot can be represented by a polynomial embedding .
shastri @xcite gave a proof of this fact and provided specific polynomial embeddings for the trefoil and figure-8 knots .
thus , the gordon - lueke theorem from topology implies that there exist algebraic embeddings of @xmath36 in @xmath37 which are _ not _ rectifiable .
it is an important open question whether algebraic embeddings of @xmath38 in @xmath39 can always be rectified by algebraic automorphisms ; kaliman @xcite showed that they can be rectified by holomorphic automorphisms .
most work on the subject beyond that described above has been aimed at finding parametrizations for specific knots , or determining minimal degrees for such parametrizations @xcite . in their recent preprint @xcite , durfee and oshea survey results on polynomial knots , including parametrizations for several specific knots . the results from _ sect .
4 _ of this paper were reported previously in the two research reports @xcite of the second author .
[ [ preliminaries . ] ] preliminaries .
+ + + + + + + + + + + + + + the ground field @xmath0 is any field of characteristic zero , though the field of real numbers @xmath40 is used in discussing knots . the polynomial ring in @xmath3 variables @xmath41 is denoted by @xmath42}={{\bf k}}[x_1, ... ,x_n]$ ] .
the subalgebra generated by @xmath43 $ ] is denoted by @xmath44 $ ] .
affine @xmath3-space over @xmath0 is @xmath45 .
let @xmath46 be a morphism of @xmath0-varieties , and let @xmath47\to{{\bf k}}[t]$ ] be the corresponding map of * k*-algebras .
the following statements are equivalent ( see @xcite , cor .
b.2.6 , or @xcite , lemma 4 ) .
* @xmath4 is an embedding . * @xmath4 is injective , and @xmath48 for all @xmath49 .
* @xmath50 is surjective .
@xmath51 is the group of algebraic automorphisms of @xmath52 $ ] , or equivalently , of @xmath45 .
two embeddings @xmath53 are said to be _ algebraically equivalent _ if and only if there exists @xmath54 with @xmath55 .
an embedding @xmath46 is _ rectifiable _ if and only if @xmath4 is algebraically equivalent to the standard embedding defined by @xmath56 .
the famous epimorphism theorem of abhyankar and moh @xcite and suzuki @xcite asserts that every algebraic embedding of @xmath57 in @xmath58 is algebraically equivalent to the standard embedding .
we use the following equivalent version .
( epimorphism theorem ) assume @xmath0 is a field of characteristic zero . if @xmath59 $ ] are such that @xmath60 $ ] , then either @xmath61 or @xmath62 . many other proofs of this result have subsequently appeared .
of particular note is the paper of rudolph @xcite , which gives an elegant proof of the epimorphism theorem using knot theory .
shastri @xcite gives the embedding @xmath63 defined by @xmath64 , and shows that @xmath65 is the trefoil knot .
in particular , if @xmath66 $ ] is @xmath67 , then @xmath68 .
while @xmath4 itself is already quite simple , we see that @xmath69 can be simplified : @xmath70 .
thus , if @xmath71 is defined by @xmath72 , then @xmath73 we recognize that the defining polynomials for the trefoil in this re - parametrization are the ( monic ) chebyshev polynomials of degree 3,4 , and 5 .
it is therefore natural to ask if other combinations of chebyshev polynomials define embeddings , and if so , what knots they parametrize over the field of real numbers . in general ,
note that two kinds of equivalence of algebraic lines in @xmath74 are considered in this paper .
first , the notion of algebraic equivalence is defined above , and is valid for any ground field .
in addition , when the ground field is @xmath40 , we consider topological equivalence , by which we mean the existence of a homeomorphism from @xmath37 to itself which carries one embedded line to the other .
thus , a knot is not distinguished from its mirror image , as with isotopic equivalence .
clearly , in the case of algebraic lines in @xmath37 , algebraic equivalence implies topological equivalence .
[ [ note - added - in - proof . ] ] note added in proof .
+ + + + + + + + + + + + + + + + + + + + the paper _ chebyshev knots _ @xcite of koseleff and pecker was posted on the arxiv preprint server in december 2008 .
their paper also studies knots parametrized by chebyshev polynomials , and several of their results parallel those found in our paper . in some cases , their more topological
approach enables them to give proofs of results we had conjectured , most notably : ( 1 ) for @xmath75 , @xmath76 parametrizes the @xmath77 torus knot , and ( 2 ) for relatively prime @xmath78 and @xmath79 , @xmath80 parametrizes the alternating @xmath9-knot ( which they call an alternate harmonic knot ) . on the other hand , the focus of our paper is on algebraic aspects , for example , working over a field of characteristic zero when possible , with empasis on algebraic equivalence / non - equivalence of embedded lines .
given @xmath83}$ ] , there exist @xmath84}$ ] such that @xmath85 , where @xmath86 . in particular , for any integer @xmath87 , let @xmath88 $ ] and @xmath89 $ ] be such that @xmath90 note that @xmath81 and @xmath82 are homogeneous of degree @xmath3 .
note also that replacing @xmath91 by @xmath92 gives @xmath93 [ basic ] assume @xmath94 and @xmath95 . * @xmath96 and@xmath97 * @xmath98 and@xmath99 * @xmath100 * @xmath101 and@xmath102 _ proof . _
statement ( a ) follows immediately by comparing @xmath103 with @xmath104 . for ( b ) , we have : @xmath105 part ( c ) is a special case of ( b ) when @xmath106 , since @xmath107 .
finally , part ( d ) follows immediately by comparing @xmath108^n$ ] with @xmath109 .
@xmath110 define @xmath112\to{{\mathbb{z}}}[t]$ ] by @xmath113 and @xmath114 .
define @xmath115 $ ] by @xmath116 then @xmath2 and @xmath117 are the chebyshev polynomials of the first and second kind , respectively . in particular , note that @xmath118 , @xmath119 , @xmath120 , and @xmath121 . as with trigonometric functions , there are countless identities involving chebyshev polynomials .
those needed in this paper are given in the following lemma .
[ identities ] given @xmath94 with @xmath95 : * @xmath122 * @xmath123 * @xmath124 * @xmath125 * @xmath126 * @xmath127 * @xmath128 * @xmath129 * @xmath130 _ proof . _
_ lemma [ basic](a ) _ implies @xmath131 in addition , since @xmath132 , _
lemma [ basic](b ) _ implies @xmath133 parts ( a ) and ( b ) are thus proved , and part ( c ) follows by adding these two equalities .
similarly , _ lemma [ basic](a ) _ implies @xmath134 in addition , since @xmath132 , _
lemma [ basic](b ) _ implies @xmath135 parts ( d ) and ( e ) are thus proved , and part ( f ) follows by adding these two equalities . in order to prove ( g ) , for each @xmath87 let @xmath136}$ ] be such that @xmath137 and @xmath138 , noting that @xmath139 and @xmath140 . since @xmath141 $ ] for all @xmath87 , _
lemma [ basic](c ) _ implies @xmath142 therefore , by _
lemma [ basic](d ) _ , it follows that @xmath143 by symmetry , @xmath144 as well , so ( g ) is proved . for ( h )
, it follows from _
lemma [ basic](d ) _ that @xmath145 the remaining equality of ( h ) follows by symmetry .
finally , to prove part ( i ) we need the following three identities : @xmath146 the first of these is implied by part ( a ) when @xmath147 , and the second follows from ( d ) when @xmath148 . for the third , set @xmath149 in part ( d ) to obtain @xmath150 .
thus , @xmath151 we now prove ( i ) by induction , the cases @xmath152 being clear .
assume @xmath75 and @xmath153 whenever @xmath154 .
then @xmath155 this completes the proof of the lemma .
@xmath110 note that as a special case , identities ( a ) and ( d ) above give the familiar recursion formulae : @xmath156 and @xmath157 from this , it is clear that @xmath158 , and that the leading coefficient of @xmath2 is @xmath159 .
in addition , @xmath2 is an even function for even @xmath3 , and an odd function for odd @xmath3 . by induction , these also show that @xmath160 and @xmath161 . the chebyshev polynomials @xmath162 share many properties with the monomials @xmath163 .
they form a basis for @xmath164 $ ] as a @xmath0-vector space . moreover , up to a certain mild equivalence , these are the only families of univariate polynomials which commute with each other by composition : @xmath165 see @xcite .
[ roots ] let @xmath3 be a non - negative integer
. * for all @xmath166 , @xmath167 * the roots of @xmath2 are @xmath168 , @xmath169 , and each of these is a simple root .
* the roots of @xmath117 are @xmath170 , @xmath169 , and each of these is a simple root .
* given @xmath171 , @xmath172 if and only if @xmath173 .
_ let @xmath136}$ ] be defined as in the preceding proof , i.e. , @xmath137 and @xmath138
. then @xmath174 it follows that , for any real number @xmath175 , @xmath176 this proves ( a ) .
next , note that @xmath177 is a set of @xmath3 distinct real numbers , and from part ( a ) , each number in this set is a root of @xmath162 . since @xmath158 , part ( b ) follows .
a similar argument is used to show ( c ) . for ( d ) , let @xmath178 be given , @xmath172 , and choose @xmath166 such that @xmath179 .
then by part ( a ) , it follows that @xmath173 .
conversely , note that @xmath2 can have no critical points outside the interval @xmath180 $ ] , since @xmath130 and the roots of @xmath181 lie in the interval @xmath180 $ ] . therefore , since @xmath182 , @xmath2 is increasing for @xmath183 , and since @xmath160 , this implies @xmath184 for @xmath183 . finally , since @xmath185
, it follows that @xmath186 when @xmath187 .
given positive integers @xmath78 and @xmath79 with @xmath188 , define @xmath189 $ ] by @xmath190 .
consider the degree function on @xmath191 $ ] defined by @xmath192 and @xmath193 .
given @xmath194 $ ] , let @xmath195 denote the highest homogeneous summand of @xmath196 relative to this grading . clearly , if @xmath197 for @xmath198 $ ] of positive degree , then @xmath199 , where @xmath200 and @xmath201 are of positive degree . however , @xmath202 , and it is well known that this polynomial is irreducible when @xmath78 and @xmath79 are coprime
. therefore , @xmath69 is irreducible .
let @xmath203 be the irreducible curve defined by @xmath204 .
since @xmath205 , it follows that @xmath203 is parametrized by @xmath206 .
note that by _ lemma [ identities](i ) _ , @xmath207 . by _
lemma [ roots](c ) _ , it follows that @xmath208 for all @xmath209 .
[ curves ] assume @xmath78 and @xmath79 are positive integers with @xmath26 , and let @xmath203 be the plane curve defined by @xmath210 . * the singularities of @xmath203 consist of @xmath211 nodes . *
the nodes of @xmath203 are precisely points of the set @xmath212 * given the node @xmath213 , @xmath214 where @xmath215 , @xmath216 , and @xmath217 satisfy @xmath218 .
* @xmath219 _ proof . _
if @xmath220 , then the singular points of @xmath203 are defined by the system @xmath221 by _ lemma [ identities](k ) _ and _ lemma [ roots](c ) _ , the roots of @xmath222 are @xmath223 @xmath224 , and the roots of @xmath225 are @xmath226 , @xmath227 .
it follows that the set of singular points of @xmath203 is @xmath228 .
let @xmath229 be a point of @xmath230 , where @xmath231 as above .
then @xmath232 since @xmath233
. therefore , @xmath234 , and @xmath230 is precisely the set of singular points of @xmath203 .
it remains to show that @xmath235 is an ordinary double point of @xmath203 . for
@xmath236 and @xmath237 , let @xmath238}$ ] be such that @xmath239 , and write @xmath240 , where @xmath241 is homogeneous of degree @xmath242 in @xmath243 and @xmath244
. then @xmath245 in particular , @xmath246 note that @xmath247 and @xmath248 .
since the roots of @xmath249 and @xmath250 are simple , it follows that @xmath251 and @xmath252 . therefore , @xmath253 factors as the product of two distinct linear forms .
it follows that @xmath235 is a node of @xmath203 .
so parts ( a ) and ( b ) are proved . in order to prove part ( c ) , let @xmath254 , and set @xmath255 for @xmath256 defined above ( @xmath257 ) . if @xmath258 , then @xmath259 thus , either @xmath260 ( positive case ) or @xmath261 ( negative case ) . in the first case , the fact that @xmath218 implies @xmath262 , which is impossible since @xmath263 .
likewise , the second case yields @xmath264 , which is impossible since @xmath265 .
therefore , @xmath266 . on the other hand , @xmath267 , since for @xmath257 , @xmath268 this last equality uses the fact that @xmath233 .
so part ( c ) is proved . for part ( d ) ,
reasoning as above gives @xmath269 since @xmath270 , it follows that @xmath271 .
similarly , @xmath272 implies that @xmath273 .
therefore , @xmath274 the equality now follows by comparing cardinalities of these two sets .
@xmath110 the much - studied curves @xmath275 may be viewed as a degeneration of the curves @xmath276 .
the lin - zaidenberg theorem implies that , when @xmath277 , the curve @xmath275 has a unique embedding in the plane up to algebraic equivalence ; see @xcite . in the case @xmath278 and @xmath3 are distinct prime numbers , abhyankar and sathaye @xcite generalized this as follows . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ( uniqueness theorem ) _ if @xmath278 and @xmath3 are distinct prime numbers , and if @xmath279 with @xmath280 , then @xmath196 has only one place at infinity and @xmath196 has a unique plane embedding , up to algebraic equivalence .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ therefore , the curves @xmath281 admit only one planar embedding , up to algebraic equivalence , when @xmath278 and @xmath3 are distinct primes .
in the 1954 article @xcite ( in czech ) , fiedler and granat studied certain rational curves with a maximum number of nodes , including the family defined by @xmath282 , @xmath75 . , @xmath283 and @xmath284,width=480 ]
given positive integers @xmath5 , define the _ pairwise greatest common divisor _ of @xmath5 by @xmath285 [ embedding ] let @xmath0 be a field of characteristice zero .
given integers @xmath286 , let @xmath4 denote the morphism @xmath24 . then @xmath4 is an embedding if and only if @xmath287 . _ proof .
_ assume first that @xmath4 is an embedding .
set @xmath288 , and write @xmath289 and @xmath290 for integers @xmath291 and @xmath292
. then @xmath293={{\bf k}}[t_a(t_d),t_b(t_d),t_k]\subset{{\bf k}}[t_d , t_k]\,\ , , \ ] ] which implies that the morphism @xmath294 is an embedding . by the epimorphism theorem , either @xmath295 or @xmath296 .
if @xmath295 , write @xmath297
. then @xmath298={{\bf k}}[t_k]$ ] , which implies @xmath299 , a contradiction . therefore , @xmath296 .
. then @xmath301={{\bf k}}[t_d]$ ] , which implies @xmath302 . in the same way , @xmath303
. therefore , @xmath287 .
conversely , assume that @xmath287 . then at most one of @xmath5 is even , and we may assume that @xmath78 is odd
. then it suffices to show the existence of positive integers @xmath304 such that @xmath305 and @xmath306 . in this case ,
@xmath307\,\ , , \ ] ] and consequently @xmath4 is an embedding .
it remains to show the existence of such integers @xmath291 , @xmath292 , and @xmath308 . since @xmath26
, there exist positive integers @xmath309 and @xmath310 such that @xmath311 .
thus , for any integer @xmath312 , @xmath313 consider the linear congruence @xmath314 , or equivalently , @xmath315 since @xmath78 is odd and @xmath287 , it follows that @xmath316 , and thus @xmath317 .
choose @xmath312 such that @xmath318 and @xmath319 .
then the integers @xmath320 satisfy the required conditions .
given integers @xmath5 with @xmath25 , the proof above gives an algorithm for finding integers @xmath304 such that @xmath321 .
note that if @xmath322 , then @xmath323 $ ] and @xmath7 is an embedding . combining this with _ prop .
[ embedding ] _ gives a complete description of which triples @xmath324 yield embeddings .
we next investigate when two such embeddings are algebraically equivalent .
* example 1 .
* over the field @xmath40 , the graph of the embedding @xmath325 is given in _
[ fig:3,4,7]_. from this , it is evident that the image of @xmath4 is a trivial knot .
that @xmath4 is a trivial embedding is seen as follows .
define @xmath326 by @xmath327 .
since @xmath328 , it follows that @xmath329 .
the fact that @xmath330 is an algebraically trivial ( rectifiable ) embedding is a consequence of the following lemma .
note that rectifiability is _ a priori _ a much stronger property for an embedded real line than that of being topologically trivial .
, width=192 ] [ trivial ] if @xmath5 are positive integers and @xmath331 , then @xmath7 is a trivial embedding . _
_ suppose @xmath332 , and define @xmath333 by @xmath334 .
then @xmath335 @xmath110 an _ elementary involution _
@xmath336 is any polynomial map having one of the following forms : @xmath337 where @xmath338}$ ] and @xmath339}$ ] .
define an equivalence relation on the set of ordered triples of positive integers by @xmath340 if and only if there exists a sequence of elementary involutions @xmath341 @xmath342 such that @xmath343 .
let @xmath6 be positive integers such that @xmath26 .
the _ remnant _ of @xmath78 and @xmath79 is @xmath344 here , @xmath345 denotes the semigroup of positive integers generated by @xmath78 and @xmath79 .
it is well known that when @xmath26 , then the complement of @xmath345 in the set of positive integers is finite , and its largest element is @xmath346 ( the frobenius number of @xmath78 and @xmath79 ) .
since @xmath347 , it follows that @xmath348 .
the ordered triple @xmath324 of positive integers is a _ reduced triple _ if and only if either * ( a ) * @xmath331 , or * ( b ) * @xmath349 , @xmath26 , and @xmath350 . [ reduced ] let @xmath5 be positive integers with @xmath287 . then there exists a reduced triple @xmath351 such that @xmath352 . in particular , the embeddings @xmath7 and @xmath353 are algebraically equivalent .
we may assume @xmath354 .
the proposition obviously holds if @xmath324 is a reduced triple , so assume @xmath324 is non - reduced .
then @xmath355 and @xmath356 .
write @xmath357 for positive integers
@xmath291 and @xmath292 .
then @xmath358 define the elementary involution @xmath359 .
then @xmath360 , which implies @xmath361 .
in addition , @xmath362 implies @xmath363 .
observe that the total degree of the triple has been reduced , i.e. , @xmath364 .
since the degree can only be reduced a finite number of times , we eventually arrive at a reduced triple @xmath351 equivalent to @xmath324 .
@xmath110 [ 2-case ] if @xmath365 is an embedding , then it is a trivial embedding .
_ by the preceding proposition , @xmath366 for a reduced triple @xmath351 .
if @xmath367 , then the result follows from _
[ trivial]_. so assume @xmath368 .
the proof of the proposition shows that @xmath369 , which implies @xmath370 .
therefore , @xmath371 , and @xmath372 and @xmath14 must both be odd .
but then @xmath373 , a contradiction .
so this case can not occur .
if @xmath374 is an embedding , then it is a trivial embedding . _
_ assume @xmath375 . in the proof of _ prop .
[ reduced ] _ , the first reduction replaces the triple @xmath376 with @xmath377 , which is equivalent to @xmath378 .
note that the triple @xmath378 maintains the form @xmath379 .
therefore , this process produces a reduced triple @xmath380 equivalent to @xmath376 .
it follows that @xmath381 .
the result now follows by _
lemma [ trivial]_. @xmath110 in this section , we assume @xmath382 .
the results of the preceding section indicate that every embedding of the form @xmath7 is algebraically equivalent , via a sequence of elementary involutions , to one for which @xmath350 . thus ,
for a fixed pair of integers @xmath78 and @xmath79 with @xmath383 and @xmath26 , the set of reduced triples @xmath324 is finite .
moreover , _
lemma [ trivial ] _ and _ cor .
[ 2-case ] _ show that we need only consider cases in which @xmath384 .
_ table 1 _ shows the remnant @xmath385 for relatively prime pairs @xmath6 such that @xmath386 and @xmath387 . recall that @xmath211 is the number of nodes in the projection of @xmath7 onto the @xmath388-plane , and that @xmath389 .
in particular , @xmath390 .
* table 1 .
remnant for small degree pairs * + [ cols="^,^,<",options="header " , ] three knots from _ table 3 _ are illustrated in _
[ fig : alt - group]_. , @xmath391 , and @xmath392 knots , width=480 ] given coprime @xmath78 and @xmath79 , it is natural to look for a polynomial @xmath17 such that @xmath393 parametrizes the alternating @xmath9-knot .
in fact , @xmath17 can be any polynomial which alternates in sign at the nodes of @xmath8 ; a necessary condition is that @xmath394 .
however , we seek a closed form for @xmath17 as a function of @xmath78 and @xmath79 .
we construct such @xmath17 explicitly using chebyshev polynomials of the second kind , @xmath395 . as noted in _
lemma [ roots](c ) _ , the roots of @xmath395 are @xmath396 thus , given positive integers @xmath278 and @xmath3 , @xmath397 and @xmath181 each divides @xmath398 .
_ lemma [ identities](h ) _ give the identity @xmath399 if @xmath277 , then @xmath400 , which implies that @xmath401 is a polynomial of degree @xmath402 , and an even function , with roots @xmath403 in addition , each of these is a simple root , since roots of the polynomials @xmath117 are simple .
the reader will note that the definition of the polynomials in ( 2 ) is quite similar to that of the cyclotomic polynomials @xmath404 .
[ t - infinite ] let @xmath278 and @xmath3 be relatively prime positive integers , @xmath405 , and define @xmath406 then @xmath63 is an embedding , and @xmath65 is the alternating @xmath407-knot .
_ set @xmath408 and let @xmath409 be the roots of @xmath69 , where @xmath410 . by _ prop .
[ curves](d ) _ , the projection of @xmath65 onto the @xmath388-plane has nodes at precisely the roots of @xmath411 . since the roots of @xmath69 are simple , it follows that @xmath412 for each consecutive pair @xmath413 @xmath414 .
suppose @xmath415 for @xmath416 , where @xmath417 . then by the parity property for nodal curves ( _ prop .
[ parity ] _ ) , @xmath291 and @xmath292 are of opposite parity .
it follows that @xmath418 .
in particular , @xmath419 .
this implies that @xmath4 is one - to - one , since if @xmath420 is a point of self - intersection for the image of @xmath4 , then @xmath421 is a node of @xmath422 .
likewise , @xmath423 is nowhere vanishing , since this is already true for @xmath424 .
therefore , @xmath4 is an embedding , and the projection of @xmath425 onto the @xmath388-plane is an alternating projection . @xmath110
* example . *
consider the case @xmath426 and @xmath427 .
using the identity ( 1 ) above , and the fact that @xmath428 , we obtain @xmath429 the alternating @xmath430-knot is thus parametrized as follows .
@xmath431 _
table 2 _ suggests that the knot @xmath7 is the alternating @xmath9-knot whenever @xmath432 , though we have not been successful in proving this .
the difficulty in doing so is that , unlike @xmath433 above , the sign of @xmath434 does not alternate at the nodes of @xmath8 . in this section ,
let @xmath436 denote the nodal curve defined by @xmath437 , @xmath75 . by _ prop .
[ curves ] _ , the number of nodes of @xmath438 is @xmath439 .
define a crossing sequence @xmath440 , @xmath441 , as follows .
first , @xmath442 for the remaining elements , set @xmath443 the sign pattern for the first one - third of this sequence is either @xmath444 or @xmath445 .
if @xmath3 is odd , this pattern repeats twice , and if @xmath3 is even , the pattern alternates with its negation . for @xmath75 , the curve @xmath438 together with crossing sequence @xmath440 , as defined above , define the @xmath77 torus knot . _
_ given a 2-tangle @xmath446 , a pair of points @xmath447 is a _
central pair _ for @xmath446 if each belongs to distinct components of @xmath446 and lie between the crossings in a standard projection . for @xmath75 ,
let @xmath448 be the @xmath77 torus knot .
for @xmath449 , @xmath448 can be constructed from @xmath450 as follows .
let @xmath451 be a 2-tangle , and let @xmath447 be a central pair .
cut @xmath450 at @xmath235 and @xmath452 , and insert an additional 2-tangle @xmath453 .
the resulting knot is @xmath448 .
we now prove the proposition by induction on @xmath3 , the case @xmath148 having been established in _
table 2_. for @xmath75 , assume @xmath438 is equivalent to @xmath448 .
let @xmath454 be the 2-tangle whose two crossings correspond to @xmath455 and @xmath456 in the crossing sequence @xmath440 , and let @xmath447 be a central pair for @xmath446 ; see _ fig .
[ fig : splicing](a)_. cut @xmath438 at @xmath235 and @xmath452 , and insert a 2-tangle @xmath453 , as per _ fig .
[ fig : splicing](b ) _ and _ ( c)_. now perform a twist as indicated in _ fig . [
fig : splicing](d ) _ , where @xmath457 represent corresponding points in _ fig .
[ fig : splicing](c ) _ and _ ( d)_. the resulting knot is @xmath458 . since @xmath458 is obtained from @xmath438 by inserting a 2-tangle at the central pair of the preceding 2-tangle , beginning with the trefoil @xmath459 , it follows by induction that , for all @xmath75 , @xmath438 is equivalent to @xmath448 . @xmath110 , width=384 ] _ table 2 _ suggests that @xmath76 parametrizes the @xmath77 torus knot . this has been verified through @xmath460 , though we have not succeeded in proving this for all @xmath3 .
the paper @xcite shows that the @xmath77 torus knot can be parametrized in degrees @xmath461 , where the degree-3 component is @xmath462 .
various patterns can be observed in _ table 2 _ , suggesting several conjectures for the family of knots defined by @xmath7 for reduced triples @xmath324 .
finally , we note that the standard three - term skein relations used to define invariant polynomials are similar to the recursive relation used to define chebyshev polynomials . in fact , the conway polynomial for the @xmath469 torus knot or link is precisely the ( monic ) chebyshev polynomial of the first kind of degree @xmath470 .
it would be of interest to know if there is a connection between the invariant polynomials of a knot and the equations which define its parametrization .
10 [ 2 ] http://www.ams.org/mathscinet-getitem?mr=#1[#2 ] [ 2]#2 s. abhyankar , _ lectures on expansion techniques in algebraic geometry _ , tata inst .
lectures on math . and phys . , vol .
57 , tata inst .
res . , bombay , 1977 . | working over a field @xmath0 of characteristic zero , this paper studies line embeddings of the form @xmath1 , where @xmath2 denotes the degree @xmath3 chebyshev polynomial of the first kind . in _ section 4
_ , it is shown that ( 1 ) @xmath4 is an embedding if and only if the pairwise greatest common divisor of @xmath5 is 1 , and ( 2 ) for a fixed pair @xmath6 of relatively prime positive integers , the embeddings of the form @xmath7 represent a finite number of algebraic equivalence classes .
_ section 2 _ gives an algebraic definition of the chebyshev polynomials , where their basic identities are established , and _ section 3 _ studies the plane curves @xmath8 .
_ section 5 _ establishes the parity property for nodal curves , and uses this to parametrize the family of alternating @xmath9-knots over the real numbers . |
instantons @xcite are well known to represent tunnelling transitions in non - abelian gauge theories between degenerate vacua of different topology .
these transitions induce processes which are _ forbidden _ in perturbation theory , but have to exist in general @xcite due to adler - bell - jackiw anomalies . correspondingly
, these processes imply a violation of certain fermionic quantum numbers , notably , @xmath7 in the electro - weak gauge theory and chirality ( @xmath8 ) in ( massless ) qcd .
an experimental discovery of such a novel , non - perturbative manifestation of non - abelian gauge theories would clearly be of basic significance .
a number of results has revived the interest in instanton - induced processes during recent years : * first of all , it was shown @xcite that the generic exponential suppression of these tunnelling rates , @xmath9 , may be overcome at _ high energies _ , mainly due to multi - gauge boson emission in addition to the minimally required fermionic final state . *
a pioneering and encouraging theoretical estimate of the size of the instanton ( @xmath10 ) induced contribution to the gluon structure functions in deep - inelastic scattering was recently presented in ref .
the summation over the @xmath10-induced multi - particle final state was implicitly performed by starting from the optical theorem for the virtual @xmath11 forward amplitude .
the strategy was then to evaluate the contribution to the functional integral coming from the vicinity of the instanton - antiinstanton ( @xmath12 ) configuration in euclidean space , to analytically continue the result to minkowski space and , finally , to take the imaginary part .
while the instanton - induced contribution to the gluon structure functions turned out to be small at larger values of the bjorken variable @xmath13 , it was found in ref .
@xcite to increase dramatically towards smaller @xmath13 . *
last not least , a systematic phenomenological and theoretical study is under way @xcite , which clearly indicates that deep - inelastic @xmath14 scattering at hera now offers a unique window to experimentally detect qcd - instanton induced processes through their characteristic final - state signature .
the searches for instanton - induced events have just started at hera and a first upper limit of @xmath15 nb at @xmath16 confidence level for the cross - section of qcd - instanton induced events has been placed by the h1 collaboration @xcite .
new , improved search strategies are being developped @xcite with the help of a monte carlo generator ( qcdins 1.3 ) @xcite for instanton - induced events .
the central question is , of course , whether instanton - induced processes in deep - inelastic scattering can both be reliably computed and experimentally measured . in particular , whether contributions associated with the non - perturbative vacuum structure can be controlled in the same way as perturbative short - distance corrections , in terms of a hard scale @xmath17 . in the work of refs .
@xcite on deep - inelastic scattering , the integrals over the instanton size @xmath0 were found to be infrared ( ir ) divergent , like in a number of previous instanton calculations in different areas .
yet , the authors claimed that this problem does not affect the possibility , to isolate in deep - inelastic scattering a well - defined , ir - finite and sizable instanton contribution in the regime of small qcd - gauge coupling , on account of the ( large ) photon virtuality @xmath5 .
the ir - divergent pieces of the @xmath10-size integrals were supposed to be factorizable into the parton distributions , which anyway have to be extracted from experiment at some reference scale .
on the other hand , also ir - finite instanton contributions to certain observables in momentum space have been found in the past @xcite . in this ideal case ,
the size of the contributing instantons is limited by the inverse momentum scale @xmath1 of the experimental probe , as one might intuitively expect .
no _ ad hoc _ cutoff or assumption about the behaviour of large , overlapping instantons need be introduced . a main issue of the present work is to shed further light on these important questions around the ir - behaviour associated with the instanton size in deep - inelastic scattering .
this paper represents the first of several papers in preparation @xcite , containing our theoretical results on @xmath10-induced processes in the deep - inelastic regime . for clarity ,
let us reduce here the realistic task of evaluating the @xmath10-induced cross - sections of the chirality violating multi - particle processes ( illustrated in fig .
[ f1 ] ) @xmath18 to the detailed study of the _ simplest _ one , without additional gluons and with just one massless flavour ( @xmath19 ) , @xmath20 the price is , of course , that this process only represents a small fraction of the total @xmath10-induced contribution to the gluon structure functions . however , there are also a number of important virtues : the present calculation provides a clean and explicit discussion of most of the crucial steps involved in our subsequent task @xcite to calculate the _ dominant _ @xmath10-induced contributions .
while unessential technical complications have been eliminated here , a generalization to the realistic case with gluons and more flavours in the final state is entirely straightforward @xcite .
we shall explicitly calculate the corresponding fixed angle cross - section and the contributions to the gluon structure functions in leading semi - classical approximation within standard instanton perturbation theory ( sect .
gauge invariance is kept manifest along the calculation and we may compare at various stages with the appropriate chirality - conserving process , calculated in leading order of perturbative qcd . as a central result of this paper and unlike ref .
@xcite , we find _ no _ ir divergencies associated with the integration over the instanton size @xmath0 , which can even be perfomed analytically .
we are able to demonstrate explicitly that the typical hard scale @xmath17 in deep - inelastic scattering provides a _ dynamical _ infrared cutoff for the instanton size , @xmath21 .
additional gluons in the final state will not change this conclusion , as is briefly outlined in sect .
thus , deep - inelastic scattering may indeed be viewed as a distinguished process for studying manifestations of qcd - instantons .
let us start with the matrix elements @xmath22 of the general , exclusive photon - parton reactions @xmath23 in terms of which we form the inclusive structure tensor @xmath24 of the parton @xmath25 , @xmath26 { \mathcal w}^{\mu \nu ( n ) } _ { p } ( q , p ) & = & \frac{1}{4\,\pi}\ , \int dps^{(n)}\ , { \mathcal t}^\mu ( q , p;k_{1},\ldots , k_{n } ) { \mathcal t}^{\nu\,\ast}(q , p;k_{1},\ldots , k_{n } ) , \label{wmunu_n}\end{aligned}\ ] ] and @xmath27 averaging over colour and spin of the initial state is implicitly understood in eq .
( [ wmunu_n ] ) ; the index @xmath28 is to label besides the final state partons also their spin and colour degrees of freedom . for exclusive @xmath29 processes , @xmath30 , the differential cross - section
is then expressed as @xmath31 \
, , \ ] ] where @xmath32 denotes the photon virtuality and @xmath33 the momentum transfer squared . in general , each final state , @xmath34 , contributes to the structure functions @xmath35 of the parton @xmath25 via the projections @xmath36
x\,{\mathcal w } _
{ p}^{\mu \nu\ , ( n ) } ( q , p ) \ , , \label{projf2 } \\[10pt ] { \mathcal f}^{(n)}_{l\ , { p } } ( x , q^2 ) & = & 4\,x^2\,\frac{p_\mu p_\nu}{p\cdot q } { \mathcal w } _ { p}^{\mu \nu\ , ( n ) } ( q , p ) \ , \label{projfl } \hspace{24pt } { \rm such\ that } \\[10pt ] { \mathcal f}_{i\ , { p}}(x , q^{2})&=&\sum_{n=1}^{\infty } { \mathcal f}^{(n)}_{i\ , { p}}(x , q^{2})\ , , \end{aligned}\ ] ] with @xmath37 denoting the bjorken variable of the photon - parton subprocess .
the spin averaged proton structure functions @xmath38 and @xmath39 , appearing ( in the one photon exchange approximation ) in the unpolarized inclusive lepto - production cross - section as @xmath40 , \ ] ] are expressed via a standard convolution in terms of the parton structure functions @xmath41 and corresponding parton densities @xmath42 , @xmath43 here , @xmath44 is the center - of - mass ( c.m . )
energy of the lepton - hadron system .
the corresponding bjorken variables are defined as usual @xmath45 where @xmath46 is the four - momentum of the incoming proton ( lepton ) . as outlined in the introduction
, we shall consider in this paper only the contributions from the simplest instanton - induced , _ chirality - violating _ photon - gluon process , corresponding to one massless quark flavour @xmath47 , @xmath48 it will be very instructive to compare with the appropriate leading - order perturbative qcd amplitudes for the _ chirality - conserving _ process ( fig .
[ f2 ] ) , @xmath49 at the various stages of the instanton calculation .
therefore , for reference , let us summarize the well - known perturbative results next ( see any textbook on perturbative qcd , e.g. ref .
@xcite ) .
let us use two - component weyl - notation for the ( massless ) fermions involved , in order to facilitate the comparison with the instanton calculation later on .
the leading - order amplitude for the perturbative process ( [ pertproc ] ) ( fig .
[ f2 ] ) then reads , @xmath50 \nonumber & & e_{q}\,g_s\,t^{a}\ , \chi_{l}^{\dagger}(k_{2 } ) \left [ \overline{\sigma}_{\mu ^{\prime}}\frac{(q - k_{1})}{(q - k_{1})^{2 } } \overline{\sigma}_\mu -\overline{\sigma}_\mu \frac{(q - k_{2})}{(q - k_{2})^{2 } } \overline{\sigma}_{\mu ^{\prime } } \right ] \chi_{l}(k_{1})\ , \end{aligned}\ ] ] where the two - component weyl - spinors @xmath51 satisfy the weyl - equations , @xmath52 and @xmath53 in eqs .
( [ ampl_pt]-[compl ] ) and throughout the paper we use the abbreviations , @xmath54 where the familiar @xmath55-matrices we use the standard notations , in minkowski space : @xmath56 , @xmath57 , and in euclidean space : @xmath58 , @xmath59 , where @xmath60 are the pauli matrices . ]
satisfy , @xmath61 finally , in eq .
( [ ampl_pt ] ) , @xmath62 , are the su(3 ) generators , @xmath63 is the quark charge in units of the electric charge @xmath64 , and @xmath65 is the su(3 ) gauge coupling . with help of eqs .
( [ weyleq ] ) , ( [ com ] ) and the on - shell conditions @xmath66 , the gauge - invariance constraints , @xmath67 are easily checked .
next , we obtain the leading - order contribution of the process ( [ pertproc ] ) to @xmath68 by contracting eq .
( [ ampl_pt ] ) with the gluon polarization vector @xmath69 and taking the traces in eq .
( [ wmunu_n ] ) by means of relations ( [ compl ] ) and ( [ com ] ) .
averaging over the initial - state gluon polarization and colour amounts to an overall factor @xmath70 .
the final result for the projections needed in eqs .
( [ diffcross ] ) , ( [ projf2 ] ) , and ( [ projfl ] ) then reads @xmath71 \nonumber & & e_q^2\ , \frac{\alpha_s}{4\,\pi}\ , \frac{x}{2\,q^2}\ , \left [ \frac{u}{t}+\frac{t}{u}-2\,\frac{1-x}{x}\,\frac{q^4}{tu } \right ] \ , , \end{aligned}\ ] ] @xmath72 where @xmath73 . upon integrating eq .
( [ gmunupt ] ) over @xmath74 , we encounter the familiar collinear divergencies for @xmath75 . in order to isolate the hard contributions to the gluon structure functions ,
it is adequate to regularize the collinear singularities by introducing an infrared cutoff scale @xmath76 in the integration limits , @xmath77 . on account of eqs .
( [ projf2 ] ) , ( [ projfl ] ) , one then obtains the familiar resultswhen comparing with the literature , one has to remember that we considered only the production of a @xmath78 pair ( c.f .
( [ pertproc ] ) ) . in the full @xmath79 contribution to the gluon structure functions ,
the production of a @xmath80 pair has also to be included .
this amounts to multiplying eqs .
( [ f2pert ] ) , ( [ flpert ] ) , by a factor of 2 . ] in the bjorken limit , @xmath81 \nonumber & & x\ , \left [ p_{qg}(x)\ , \ln\left(\frac{q^{2}}{\mu _ { c}^{2}}\right ) + p_{qg}(x)\,\ln \left(\frac{1}{x}\right ) -\frac{1}{2 } + 3\,x\,(1-x ) \right ] \left[1+{\mathcal o}\left(\frac{\mu
_ { c}^{2}}{q^{2}}\right)\right ] \ , , \end{aligned}\ ] ] @xmath82 with the splitting function @xmath83 of course , the _ finite _ part of the structure function @xmath84 , that is everything except for the large logarithm , @xmath85 , is scheme dependent .
in this section , we turn to the central issue of this paper . we consider the simplest instanton - induced exclusive process , @xmath87 , and compute its contributions to the fixed angle differential cross - section and the gluon structure functions @xmath88 and @xmath89 , in leading semi - classical approximation . to this end , the respective green s function
is first set up according to standard instanton - perturbation theory in euclidean configuration space @xcite , then fourier transformed to momentum space , lsz amputated , and finally continued to minkowski space .
the basic building blocks are ( in euclidean configuration space and in the singular gauge ) : \i ) the classical instanton gauge field @xcite @xmath90 , @xmath91 depending on the various collective coordinates , the instanton size @xmath0 and the colour orientation matrices @xmath92 .
the @xmath93 matrices involve both colour ( @xmath94 ) and spinor ( @xmath95 ) indices , the former ranging as usual only in the @xmath96 upper left corner of @xmath97 su(3 ) colour matrices .
indices will , however , be suppressed , as long as no confusion can arise .
\ii ) the quark zero modes @xcite , @xmath98 and @xmath99 , @xmath100 and \iii ) the quark propagators in the instanton background @xcite , @xmath101 \nonumber & & \frac{1}{\sqrt{\pi_x\pi_y } } \left [ \frac{x - y}{2\pi^2(x - y)^4 } \left ( 1 + \rho^2\frac { \left [ u x \overline{y } u^\dagger\right ] } { x^2 y^2}\right ) + \frac{\rho^2\sigma_\mu } { 4\pi^2 } \frac{\left [ u x\,(\overline{x}-\overline{y } ) \sigma_\mu \overline{y } u^\dagger \right ] } { x^2(x - y)^2 y^4\pi_y } \right ] , \\[2ex ] \label{sibar } & & \overline{s}^{(i)}(x , y ) = \\[1.6ex ] \nonumber & & \frac{1}{\sqrt{\pi_x\pi_y } } \left [ \frac{\overline{x}-\overline{y}}{2\pi^2(x - y)^4 } \left ( 1 + \rho^2\frac { \left [ u x \overline{y } u^\dagger\right ] } { x^2 y^2}\right ) + \frac{\rho^2 \overline{\sigma}_\mu } { 4\pi^2 } \frac{\left [ u x \overline{\sigma}_\mu ( x - y ) \overline{y } u^\dagger \right ] } { \pi_x x^4(x - y)^2 y^2 } \right ] .\end{aligned}\ ] ] the relevant diagrams for the exclusive process of interest , eq . ( [ instproc ] ) , are displayed in fig .
[ f3 ] , in leading semi - classical approximation .
the amplitude is expressed in terms of an integral over the collective coordinates @xmath0 and the colour orientation @xmath93 , @xmath102 where @xmath103 \ , \left ( \rho\,\mu _ { r } \right)^{\beta_0+\frac{\alpha_s(\mu _ { r } ) } { 4\,\pi}\ , \left ( \beta_1 - 12\,\beta_0 \right)}\ , , \label{density}\ ] ] denotes the instanton density @xcite , with @xmath104 being the renormalization scale .
the form ( [ density ] ) of the density , with next - to - leading order expression for @xmath105 , @xmath106 \ , , \ ] ] is improved to satisfy renormalization - group invariance at the 2-loop level @xcite .
the constants @xmath107 and @xmath108 are the familiar perturbative coefficients of the qcd beta - function , @xmath109 and the constant @xmath110 is given by is known only to @xmath111-loop accuracy . in ref .
@xcite , only the ultraviolet divergent part of the @xmath112-loop correction to the instanton density has been computed . ]
@xmath113 with @xmath114 , @xmath115 , and @xmath116 , in the @xmath117-scheme . in our case
, we should of course take @xmath118 in eqs .
( [ beta ] ) and ( [ d ] ) . before analytic continuation ,
the amplitude @xmath119 entering eq .
( [ ampl_inst ] ) takes the following form in euclidean space , @xmath120 \chi_{l}(k_{1})\ , , \nonumber\end{aligned}\ ] ] with contributions @xmath121 from the diagrams on the left and right in fig .
[ f3 ] , respectively , @xmath122 , \label{tinst}\\ { { \mathcal v}_\mu ^{(u)}}(q ,- k_{2})&\equiv & \int d^{4}x\,{\rm e}^{-{{\rm i}}\,q\cdot x } \ , \left [ \lim_{k_{2}^{2}\to 0}\,({{\rm i}}k_{2})\,\overline{s}^{(i)}\,(-k_{2},x)\ , \sigma_\mu \,\kappa(x)\right]\ , , \label{uinst}\end{aligned}\ ] ] and generic notation for various fourier transforms involved , @xmath123 the lsz - amputation of the classical instanton gauge field @xmath90 in eq .
( [ ampi ] ) and the quark zero modes @xmath98 and @xmath99 in eqs .
( [ tinst ] ) and ( [ uinst ] ) , respectively , is straightforward @xcite , @xmath124\,u^{\dagger } \right]\ , , \label{lszgluon}\\ \lim_{k_{2}^{2}\to 0}\ , ( { { \rm i}}k_{2})^{\alpha\dot\alpha}\,\kappa^{i}_{\dot\alpha}(-k_{2})&= & 2\,\pi\,\rho^{3/2}\,u^{i}_{\ \beta}\,\epsilon^{\beta\alpha}\ , , \label{lszkappa}\\ \lim_{k_{1}^{2}\to 0}\,\overline{\phi}_{j}^{\dot\gamma}(-k_{1})\ , ( -{{\rm i}}\,\overline{k}_{1})_{\dot\gamma\delta } & = & 2\,\pi\,\rho^{3/2}\,\epsilon_{\beta\delta}\ , \left ( u^{\dagger}\right)^{\beta}_{\ j}\ , .
\label{lszbarphi } \end{aligned}\ ] ] on the other hand , the lsz - amputation of the quark propagators @xmath125 and @xmath126 in eqs .
( [ tinst ] ) and ( [ uinst ] ) , respectively , is quite non - trivial and has important physical consequences , as we shall see below .
we give here only the final result and refer the interested reader to appendix a where the details of the calculation can be found : @xmath127 \nonumber & & \frac{- 1}{\sqrt{\pi_x}}\ , { \rm e}^{{{\rm i}}\,k_1\cdot x}\ , \left
[ 1 + \frac{1}{2}\,\frac{\rho^2}{x^2}\ , \frac{\left [ u\,x\,\overline{k_1}\,u^\dagger \right]}{k_1\cdot x}\ , \left ( 1- { \rm e}^{-{{\rm i}}\,k_1\cdot x}\right ) \right ] \ , , \\[1.6ex ] \label{onshellsibar } \lefteqn { \lim_{k_2 ^ 2\to 0}\ , \left ( { { \rm i}}\,k_2\right)\ , \overline{s}^{(i)}\,(-k_2,x ) = } \\[0.8ex ] \nonumber & & \frac{- 1}{\sqrt{\pi_x}}\ , { \rm e}^{{{\rm i}}\,k_2\cdot x}\ , \left [ 1 + \frac{1}{2}\,\frac{\rho^2}{x^2}\ , \frac{\left [ u\,k_2\,\overline{x}\,u^\dagger \right]}{k_2\cdot x}\ , \left ( 1- { \rm e}^{-{{\rm i}}\,k_2\cdot x}\right ) \right ] \ , .\end{aligned}\ ] ] it should be noted that the first terms in eqs .
( [ onshellsi ] ) , ( [ onshellsibar ] ) , corresponding to the @xmath111 in square brackets , were argued to be present on general grounds already in ref .
the remaining terms , however , have not been given in the literature .
as we shall see below , they play a very important rle in ensuring electromagnetic gauge invariance .
the fourier transforms entering eqs .
( [ tinst ] ) and ( [ uinst ] ) , respectively , can now be done with the help of eqs .
( [ onshellsi ] ) and ( [ onshellsibar ] ) .
the result is ( see appendix b ) , @xmath128_{\gamma\alpha } } { q\cdot k_{1}}\ , f\left ( \rho\,\sqrt{q^{2 } } \right ) \right .
} \\[1.6ex ] \nonumber & & \left .
+ \left [ \frac{\left [ \epsilon\,\left(q - k_{1}\right ) \,\overline{\sigma}_\mu \right]_{\gamma\alpha } } { ( q - k_{1})^2 } -\frac{1}{2}\ , \frac{\left [ \epsilon\,k_{1}\,\overline{\sigma}_\mu \right]_{\gamma\alpha } } { q\cdot k_{1 } } \right]\,f\left(\rho\,\sqrt{\left ( q - k_{1}\right)^2}\right ) \right\}\ , , \nonumber \\[2.4ex ] \label{uvertex } \lefteqn { { \mathcal v}^{(u)\ \alpha\ , i}_{\mu \ } \,(q ,- k_{2 } ) = 2\,\pi\,{{\rm i}}\,\rho^{3/2}\ , u^i_{\ \gamma}\ , \left\ { \frac{1}{2}\ , \frac{\left [ \sigma_\mu \ , \overline{k_{2}}\,\epsilon\right]^{\alpha\gamma } } { q\cdot k_{2}}\ , f\left ( \rho\,\sqrt{q^{2}}\right ) \right . } \\[1.6ex ] \nonumber & & \left .
+ \left [ \frac{\left [ \sigma_\mu \ , \left(\overline{q}-\overline{k_{2}}\right)\,\epsilon \right]^{\alpha\gamma } } { ( q - k_{2})^2 } -\frac{1}{2}\ , \frac{\left [ \sigma_\mu \ , \overline{k_{2}}\,\epsilon\right]^{\alpha\gamma } } { q\cdot k_{2 } } \right]\,f\left(\rho\,\sqrt{\left ( q - k_{2}\right)^2}\right ) \right\}\ , , \nonumber \end{aligned}\ ] ] with the shorthand ( `` form factor '' ) , @xmath129 in terms of the bessel - k function , implying the normalization , @xmath130 the next step is to insert eqs .
( [ tvertex ] ) , ( [ uvertex ] ) , and ( [ lszgluon]-[lszbarphi ] ) into eq .
( [ ampi ] ) and to perform the integration over the colour orientation according to eq .
( [ ampl_inst ] ) by means of the relation , @xmath131 \ , .\end{aligned}\ ] ] after analytic continuation to minkowski space we find for the scattering amplitude , eq .
( [ ampl_inst ] ) , @xmath132 & & \chi_r^\dagger ( k_2 ) \left [ \left ( \sigma_{\mu ^{\prime}}\overline{p}-p\overline{\sigma}_{\mu ^\prime}\right ) v(q , k_1;\rho ) \overline{\sigma}_\mu -\sigma_\mu \overline{v}(q , k_2;\rho ) \left ( \sigma_{\mu ^\prime}\overline{p } - p\overline{\sigma}_{\mu ^\prime } \right ) \right ] \chi_l(k_1 ) , \nonumber\end{aligned}\ ] ] with the four - vector @xmath133 , @xmath134\rho\sqrt{-\left ( q - k\right)^2}\ , k_1\left(\rho\sqrt{-\left ( q - k\right)^2}\right ) \\[0.8ex ] \label{v } \mbox { } & & - \frac{k_{\lambda}}{2 q\cdot k}\rho\sqrt{-q^{2}}\ , k_1\left(\rho\sqrt{-q^{2}}\right ) .
\end{aligned}\ ] ] at this stage of our instanton calculation , the gauge - invariance constraints , eqs .
( [ gaugeinv ] ) , can easily be checked . while the relation @xmath135 holds trivially , the electromagnetic ( e.m . )
current conservation @xmath136 follows again from the relations ( [ com ] ) of the @xmath55-matrices , the weyl - equations ( [ weyleq ] ) and the on - shell conditions @xmath66 .
electromagnetic current conservation provides also for a non - trivial check of our result for the amputated quark propagators , which differs somewhat from the result quoted in ref .
@xcite : if we keep only the first terms in eqs .
( [ onshellsi ] ) , ( [ onshellsibar ] ) , corresponding to the @xmath111 in square brackets , e.m .
current conservation would only hold for a restricted set of momenta in phase space , namely for @xmath137 .
furthermore , we note one of the main results of this paper : the integration over the instanton size @xmath0 in eq .
( [ ampi2 ] ) is _
finite_. in particular , the _ good infrared _ behavior ( large @xmath0 ) of the integrand is due to the exponential decrease of the bessel - k function for large @xmath0 in eq .
( [ v ] ) . its origin , in turn , can be traced back to the `` feed - through '' of the factor @xmath138 , by which the amputated ( current ) quark propagators ( [ onshellsi ] ) and ( [ onshellsibar ] ) in the @xmath10-background essentially differ from the respective amputated free propagators . if the current - quark propagators in eqs . ( [ tinst ] ) and ( [ uinst ] )
are naively approximated by the free ones ( c.f .
( [ si ] ) , ( [ sibar ] ) ) , @xmath139 the result is both gauge variant and contains an ir - divergent piece in the @xmath0 integration .
we have thus demonstrated explicitly and to our knowledge for the first time that the typical hard scales ( @xmath140 ) in deep - inelastic scattering provide a dynamical ir cutoff for the instanton size ( at least in leading semi - classical approximation ) .
now we are ready to perform the final integration over the instanton size @xmath0 by inserting the instanton density , eq .
( [ density ] ) , into eq .
( [ ampi2 ] ) .
the result is : @xmath141 \nonumber & & -{{\rm i}}\frac{\sqrt{2}}{3}d\pi^{3}e_q \left ( \frac{2\,\pi}{\alpha_s(\mu _ { r } ) } \right)^{13/2 } { \exp}\left[-\frac{2\,\pi}{\alpha_s(\mu _ { r } ) } \right]\ , 2^{\,b}\gamma \left(\frac{b+1}{2}\right ) \gamma \left(\frac{b+3}{2}\right ) \sigma^a\,\times \\[2.4ex ] & & \chi_r^\dagger ( k_2 ) \left [ \left ( \sigma_{\mu ^{\prime}}\overline{p}-p\overline{\sigma}_{\mu ^\prime}\right ) v(q , k_1;\mu _ { r } ) \overline{\sigma}_\mu -\sigma_\mu \overline{v}(q , k_2;\mu _ { r } ) \left ( \sigma_{\mu ^\prime}\overline{p } - p\overline{\sigma}_{\mu ^\prime } \right ) \right ] \chi_l(k_1 ) , \nonumber\end{aligned}\ ] ] with the four - vector @xmath142 , @xmath143 \nonumber & & \frac{1}{\mu _ { r}}\,\left\ { \left [ \frac{\left ( q - k\right)_\lambda}{-(q - k)^2 } + \frac{k_{\lambda}}{2 q\cdot k } \right ] \left ( \frac{\mu _ { r}}{\sqrt{-\left ( q - k\right)^2}}\right)^{b+1 } - \frac{k_{\lambda}}{2 q\cdot k } \left ( \frac{\mu _ { r}}{\sqrt{-q^{2}}}\right)^{b+1 } \right\ } . \end{aligned}\ ] ] in eqs .
( [ ampifin ] ) and ( [ v ] ) , the variable @xmath144 is a shorthand for the effective power of @xmath145 in the instanton density , eq .
( [ density ] ) , @xmath146 the next steps consist in contracting the amplitude , eq .
( [ ampifin ] ) , with the gluon polarization vector @xmath69 and taking the modulus squared of this amplitude according to eq .
( [ wmunu_n ] ) . after applying eq .
( [ compl ] ) , the remaining spinor traces can be evaluated , in principle , by repeated use of eq .
( [ com ] ) .
for the actual calculation , however , we used form and , for an independent check , the hip package for maple .
the final result for the relevant projections ( c.f .
( [ diffcross ] ) , ( [ projf2 ] ) , and ( [ projfl ] ) ) of the contribution of the @xmath10-induced process ( [ instproc ] ) to the differential gluon structure tensor is found to be @xmath147 \label{gmunu } & & \frac{e_q^2}{16}\ , { \mathcal n}^{2}\ , \left ( \frac{2\,\pi}{\alpha_s(\mu _ { r } ) } \right)^{13}\ , { \exp}\left[-\frac{4\,\pi}{\alpha_s(\mu _ { r } ) } \right]\ , \left ( \frac{\mu _ { r}^{2}}{q^{2}}\right)^{b}\,\times \\[1.6ex ] \nonumber & & \frac{1-x}{q^{2}}\ , \left [ \left(\frac{q^{2}}{-t}\right)^{b+1 } + \left(\frac{q^{2}}{-u}\right)^{b+1 } + 2\,t\,u\ , \frac { \left ( \left(\frac{q^{2}}{-t}\right)^{\frac{b+1}{2 } } -1 \right ) \left ( \left(\frac{q^{2}}{-u}\right)^{\frac{b+1}{2 } } -1 \right ) } { ( t+q^{2})(u+q^{2 } ) } \right ] \ , , \\[2.4ex ] \nonumber \lefteqn { \frac{p_\mu \,p_{\nu}}{p\cdot q } \ , \frac{d{\mathcal w}^{\mu \nu\,{(2)}}_{g}}{dt } \ , \left ( \gamma^\ast + { \rm g}\rightarrow \overline{{\rm q}_l } + { \rm q}_r \right ) = } \\[1.6ex ] \label{pmupnu } & & \frac{e_q^2}{16 } { \mathcal n}^{2}\ , \left ( \frac{2\,\pi}{\alpha_s(\mu _ { r } ) } \right)^{13}\ , { \exp}\left[-\frac{4\,\pi}{\alpha_s(\mu _ { r } ) } \right ] \ , \left ( \frac{\mu _ { r}^{2}}{q^{2}}\right)^{b}\,\times \\[1.6ex ] \nonumber & & \frac{(1-x)^{2}}{q^{2}}\ , t\,u\ , \left [ \frac{\left(\frac{q^{2}}{-t}\right)^{\frac{b+1}{2 } } + \frac{u}{q^{2}}\frac{x}{1-x}}{t+q^{2 } } -\frac{\left(\frac{q^{2}}{-u}\right)^{\frac{b+1}{2 } } + \frac{t}{q^{2}}\frac{x}{1-x}}{u+q^{2 } } \right]^{2 } \ , .\end{aligned}\ ] ] here we have introduced the shorthand @xmath148 in eqs .
( [ gmunu ] ) and ( [ pmupnu ] ) , the @xmath149 symmetry is manifest . upon inserting eqs .
( [ gmunu ] ) , ( [ pmupnu ] ) into eqs .
( [ diffcross ] ) , ( [ projf2 ] ) , and ( [ projfl ] ) , we see that the contribution of the @xmath10-induced process ( [ instproc ] ) to the differential cross - section @xmath150 and the differential gluon structure functions , @xmath151 , is well - behaved as long as we avoid the ( collinear ) singularities for @xmath75 .
this is illustrated in fig .
[ f4 ] , where we compare the differential cross - sections , @xmath152 , of both the @xmath10-induced process , eq . ( [ instproc ] ) , and the perturbative process , eq . ( [ pertproc ] ) , where @xmath153 we note that the renormalization - scale dependence of the @xmath10-induced cross - section in fig . [ f4 ] is very small , due to the renormalization - group improved density ( [ density ] ) .
let us address at this point the important question concerning the range of validity of the present calculation .
specifically , let us examine the constraints emerging from the requirement of the dilute instanton gas approximation following refs . @xcite . along these lines
one finds that instantons with size @xmath154 are ill - defined semi - classically @xcite , corresponding to a breakdown of the dilute gas approximation . on the other hand , using the form of our @xmath0 integral in eqs .
( [ ampi2 ] ) , ( [ v ] ) , we may determine the average instanton size @xmath155 contributing for a given virtuality @xmath156 according to @xmath157 hence , with eq .
( [ rhoc ] ) and eq .
( [ beff ] ) , we find that the virtuality @xmath158 should obey @xmath159 in particular , our results in fig .
[ f4 ] ( top ) for the @xmath10-induced differential cross - section , @xmath160 , at @xmath161 and @xmath162 , should be taken seriously only for @xmath163 gev ( since here @xmath164 ) .
thus , like in the perturbative case , _ fixed - angle scattering processes at high @xmath5 are reliably calculable in ( instanton ) perturbation theory _
( at least in leading semi - classical approximation ) . next , we note that the contributions ( [ gmunu ] ) , ( [ pmupnu ] ) of the @xmath10-induced exclusive process ( [ instproc ] ) to the differential gluon structure functions are much more singular ( @xmath165 ) for @xmath166 than the perturbative ones ( @xmath167 ) .
this leads to a much stronger scheme dependence @xcite than in the perturbative case .
let us have a closer look at this feature .
we regularize the collinear divergence of the @xmath74 integral along the same lines as in perturbation theory , i.e. we restrict the integration to the interval @xmath168 . on account of eqs .
( [ projf2 ] ) , ( [ projfl ] ) , we then obtain for the hard contributions of the @xmath10-induced exclusive process ( [ instproc ] ) to the gluon structure functions , @xmath169 \,\left ( \frac{\mu _ { r}^{2}}{\mu _ { c}^{2}}\right)^{b } \\[1.6ex ] \label{f2inst } \mbox { } & & \times \frac{x\,(1-x)}{b}\ , \left[1+{\mathcal o}\left ( \frac{\mu _ { c}^{2}}{q^{2}}\right)\right ] \ , , \\[2.4ex ] \nonumber { \mathcal f}_{l\,g}^{(\overline{{\rm q}_l}{\rm q}_r)}\,(x , q^2;\mu _ { c}^{2 } ) & = & \frac{e_q^2}{2}\ , { \mathcal n}^{2}\ , \left ( \frac{2\,\pi}{\alpha_s(\mu _ { r } ) } \right)^{13}\ , { \exp}\left[-\frac{4\,\pi}{\alpha_s(\mu _ { r } ) } \right ] \,\left ( \frac{\mu _ { r}^{2}}{\mu _ { c}^{2}}\right)^{b}\ , \frac{\mu _ { c}^{2}}{q^{2}}\ , \\[1.6ex ] \label{flinst } \mbox { } & & \times \ , \frac{x\,(1-x)^{2}}{b-1}\ , \left[1+{\mathcal o}\left(\frac{\mu _ { c}^{2}}{q^{2}}\right)\right ] .\end{aligned}\ ] ] in the bjorken limit , @xmath170 , but with @xmath171 fixed , we find from eqs .
( [ f2inst ] ) and ( [ flinst ] ) , respectively , @xmath172 \nonumber & & \frac{e_q^2}{8}\ , { \mathcal n}^{2}\ , \left ( \frac{2\,\pi}{\alpha_s(\mu _ { r } ) } \right)^{13}\ , { \exp}\left[-\frac{4\,\pi}{\alpha_s(\mu _ { r } ) } \right ] \,\left ( \frac{\mu _ { r}^{2}}{\mu _ { c}^{2}}\right)^{b}\ , \frac{x\,(1-x)}{b } \ , , \\[2.4ex ] \label{flinstbj } \lefteqn { \lim_{q^{2}\to \infty}\ , { \mathcal f}_{l\,g}^{(\overline{{\rm q}_l}{\rm q}_r)}(x , q^2;\mu _ { c}^{2 } ) \equiv } \\[1.6ex ] \nonumber & & \lim_{q^{2}\to \infty}\,\left [ { \mathcal f}_{2\,g}^{(\overline{{\rm q}_l}{\rm q}_r ) } ( x , q^2;\mu _ { c}^{2 } )
-2\,x\,{\mathcal f}_{1\,g}^{(\overline{{\rm q}_l}{\rm q}_r ) } ( x , q^2;\mu _ { c}^{2})\right ]
= 0\ , .\end{aligned}\ ] ] hence , in this limit , the considered @xmath10-induced process gives a _ scaling _ contribution to @xmath3 and the analogue of the callan - gross relation , @xmath173 , holds . in particular
, this means that the _ same _ parton distribution can absorb the infrared sensitivity of both structure functions , @xmath174 and @xmath175 .
this is one of the prerequisites of factorization @xcite .
in this paper , we studied qcd - instanton induced processes in deep - inelastic lepton - hadron scattering .
the purpose of the present work was to shed further light on the important questions around the ir - behaviour associated with the instanton size . in order to eliminate unessential technical complications ,
we have reduced the realistic task of evaluating the @xmath10-induced cross - sections of the chirality violating multi - particle processes ( illustrated in fig .
[ f1 ] ) @xmath18 to the detailed study of the _ simplest _ one , without additional gluons and with just one massless flavour ( @xmath19 ) , @xmath20 we have explicitly calculated the corresponding fixed angle cross - section and the contributions to the gluon structure functions within standard instanton perturbation theory in leading semi - classical approximation ( sect . 3 ) .
to this approximation , _ fixed - angle scattering processes at high @xmath5 are reliably calculable_. in the bjorken limit , the considered @xmath10-induced process gives a _ scaling _ contribution to @xmath3 and the analogue of the callan - gross relation , @xmath173 , holds .
all along we focused our main attention on the ir behaviour associated with the instanton size .
gauge invariance was kept manifest along the calculation . as a central result of this paper and unlike ref .
@xcite , we found _ no _ ir divergencies associated with the integration over the instanton size @xmath0 , which can even be perfomed analytically .
we have explicitly demonstrated that the typical hard scale @xmath17 in deep - inelastic scattering provides a _ dynamical _ infrared cutoff for the instanton size , @xmath21 .
thus , deep - inelastic scattering may indeed be viewed as a distinguished process for studying manifestations of qcd - instantons . in ref .
@xcite , the @xmath10-induced contribution to deep - inelastic scattering of a virtual gluon from a real one @xcite , @xmath176 , served as a simplified rle model for the splitting into a ir - finite contribution ( @xmath177 ) and an ir - divergent term ( large @xmath0 ) .
as speculated by one of the authors @xcite , the occurence of the ir - divergent term could well have been due to the lacking gauge invariance of this model , associated with the off - shellness of one of the initial gluons .
in fact , one may enforce gauge invariance by replacing the instanton gauge field @xmath178 describing the virtual gluon by the familiar gauge - invariant operator @xmath179\ , g^{(i)}_{\mu \,\alpha}(x)\,[x , x+\infty],\ ] ] where @xmath180 is the instanton field - strength , and @xmath181=p \exp\left\ { { { \rm i}}\,g_{s } \int^\infty_0 d\lambda\,\ , e\cdot a^{(i)}(x+\lambda e ) \right\}\ , , \ ] ] is a gauge factor ordered along the lightlike line in the direction @xmath182 . in this case
the ir - divergent term is , indeed , absent @xcite .
since our present calculation is manifestly gauge - invariant , the absence of an ir divergent term fits well in line with these arguments .
a further main purpose of the present calculation was to provide a clean and explicit discussion of most of the crucial steps involved in our subsequent task @xcite to calculate the _ dominant _ @xmath10-induced contributions coming from final states with a large number of gluons ( and three massless flavours , say ) .
let us close with some comments on the generalization to the more realistic case with @xmath183 gluons in the final state , which is entirely straightforward @xcite . instead of eq .
( [ ampi2 ] ) , the corresponding amplitude involves , in leading semi - classical approximation , the additional factors from the @xmath183 gluons ( c.f .
eq . ( [ lszgluon ] ) ) , @xmath184 \nonumber & & { { \rm i}}\,e_q\,4\,\pi^2\,\left ( \frac{\pi^3}{\alpha_s}\right)^{\frac{n_g+1}{2}}\ , \int du \int\limits_0^\infty d\rho\,d(\rho , \mu _ { r } ) \,\rho^{2\,n_g } \\[1.6ex ] \nonumber & & \times \ , { \rm tr}\left [ \sigma^{a}\,u\,\left [ \epsilon_g(p)\cdot p-\epsilon_g(p)\,\overline{p } \right]\,u^{\dagger } \right]\ , \prod_{i=1}^{n_g } { \rm tr}\left [ \sigma^{a_i}\,u\,\left [ \epsilon_g(p_i)\,\overline{p_i}-\epsilon_g(p_i)\cdot p_i \right]\,u^{\dagger } \right ] \\[1.6ex ] \nonumber & & \times \left\ { \left [ u \chi_r^\dagger ( k_2 ) \epsilon \right ] \left[\epsilon v(q , k_1;\rho)\overline{\sigma}_\mu \chi_l(k_1)\ , u^\dagger \right ] \right . \\[1.6ex ] \nonumber \mbox { } & & - \left . \left [ u \chi_r^\dagger ( k_2 ) \sigma_\mu \overline{v}(q , k_2;\rho)\epsilon \right ] \left [ \epsilon\ , \chi_l(k_1)\ , u^\dagger \right ] \right\ } , \end{aligned}\ ] ] where the four - vector @xmath185 is again given by eq .
( [ v ] ) .
besides the enhancement by a factor of @xmath186 , each additional gluon gives rise to a factor of @xmath187 under the @xmath10-size integral .
the ir - finiteness of this integral is , however , not altered by the presence of the additional overall factor of @xmath188 , on account of the exponential cutoff @xmath189 $ ] from the bessel - k function in the `` form - factors '' contained in @xmath190 , eq .
( [ v ] ) .
we also note , that the amplitude ( [ ampi2 g ] ) satisfies e.m .
gauge invariance . in analogy to electro - weak @xmath191-violation @xcite
, one expects @xcite the sum of the final - state gluon contributions to exponentiate , such that the total @xmath10-induced @xmath192 g cross - section takes the form ( at large @xmath5 ) , @xmath193 \nonumber & \sim & \int_x^1 dx^\prime\int\limits^{q^2\frac{{{x^\prime}}}{x } } \frac{dq^{\prime 2}}{q^{\prime 2}}\,\ldots\ , \frac{1}{q^{\prime 2}}\ , \exp\left[-\frac{4\pi}{\alpha_s(q^\prime)}\,f(x^\prime ) \right]\ , , \end{aligned}\ ] ] where the so - called `` holy - grail function '' @xcite @xmath194 ( normalized to f(1)=1 ) is expected to decrease towards smaller @xmath195 , which implies a dramatic growth of @xmath196 for decreasing @xmath13 .
here we want to derive eqs .
( [ onshellsi ] ) and ( [ onshellsibar ] ) for the lsz - amputated quark propagators .
let us first consider the fourier transform of the quark propagator ( [ si ] ) which we write as @xmath197 where @xmath198 \nonumber & & 2\left [ x- ( -{{\rm i}}\partial_k)\right]\ , ( -{{\rm i}}\overline{\partial}_k)\,(-{{\rm i}}\partial_k)\ , \int \frac{d^4y}{(2\pi ) ^2 } \frac{{\rm e}^{{{\rm i}}k\cdot y } } { \left ( ( x - y)^2\right)^2 \left ( y^2+\rho^2\right)^{1/2 } \left ( y^2\right)^{1/2 } } \ , , \\[2.4ex ] \label{s2 } \lefteqn { s^{(2)}(x ,- k ) = } \\[1.6ex ] \nonumber & & 2\,\frac{\rho^2}{x^2}\left [ x- ( -{{\rm i}}\partial_k)\right ] u x ( -{{\rm i}}\overline{\partial}_k)\ , u^\dagger \int \frac{d^4y}{(2\pi ) ^2 } \frac{{\rm e}^{{{\rm i}}k\cdot y } } { \left ( ( x - y)^2\right)^2 \left ( y^2+\rho^2\right)^{1/2 } \left ( y^2\right)^{1/2 } } \ , , \\[2.4ex ] \label{s3 } \lefteqn { s^{(3)}(x ,- k ) = } \\[1.6ex ] \nonumber & & \frac{\rho^2}{x^2 } \sigma_\mu u x\ , \left [ \overline{x}- ( -{{\rm i}}\overline{\partial}_k)\right ] \sigma_\mu ( -{{\rm i}}\overline{\partial}_k)\ , u^\dagger \int \frac{d^4y}{(2\pi ) ^2 } \frac{{\rm e}^{{{\rm i}}k\cdot y } } { \left ( x - y\right)^2 \left ( y^2+\rho^2\right)^{3/2 } \left ( y^2\right)^{1/2 } } \ , .\end{aligned}\ ] ] our strategy to analyze the @xmath199 limit of eqs . ( [ s1]-[s3 ] ) starts by partially evaluating the master integral , @xmath200 by means of the feynman parametrization ( see e.g. @xcite ) , @xmath201 \nonumber & & \frac{\gamma ( \alpha + \beta + \gamma ) } { \gamma ( \alpha ) \,\gamma ( \beta ) \ , \gamma ( \gamma ) } \ , \int_0 ^ 1 da\,a \ , \int_0 ^ 1 db\ , \frac{(ab)^{\alpha -1 } ( a(1-b))^{\beta -1}\,(1-\alpha)^{\gamma -1 } } { \left [ ab\,a + a\,(1-b)\,b + ( 1-a)\,c\right]^{\alpha + \beta + \gamma } } \ , .\end{aligned}\ ] ] with the help of eq .
( [ feynman ] ) , it is possible to show that eq .
( [ masterint ] ) can be expressed as @xmath202 \nonumber & & \frac{2^{1-(\alpha + \beta + \gamma ) } } { \gamma ( \alpha ) \ , \gamma ( \beta ) \ , \gamma ( \gamma ) } \ , \int_0 ^ 1 da\,a^{\alpha + \beta -1}\,(1-a)^{\gamma -1 } \int_0 ^ 1 db\,b^{\alpha -1}\,(1-b)^{\beta -1}\ , { \rm e}^{{{\rm i}}\,k\cdot x\,ab } \\[1.6ex ] \nonumber & & \times \,\left ( \frac{\sqrt{x^2 ab(1-ab)+\rho^2 a(1-b)}}{\sqrt{k^2 } } \right)^{2-(\alpha + \beta + \gamma ) } \\[1.6ex ] \nonumber & & \times \,k_{2-(\alpha + \beta + \gamma ) } \left ( \sqrt{k^2}\sqrt{x^2 ab(1-ab)+\rho^2\ , a(1-b ) } \right ) .\end{aligned}\ ] ] next we insert eq .
( [ masterintfin ] ) into eqs .
( [ s1])-([s3 ] ) , perform the various derivatives , and expand the integrand with respect to @xmath199 .
finally , the remaining feynman parameter integrations are done .
after this procedure we find : @xmath203 \lim_{k^{2}\to 0}{s^{(2 ) } } ( x ,- k)(-{{\rm i}}\,\overline{k } ) & = & - \frac{1}{2}\frac{\rho^2}{x^2 } \frac{\left [ u x \overline{k } u^\dagger \right]}{k\cdot x } \left [ { \rm e}^{{{\rm i}}\,k\cdot x } - \frac{{{\rm i}}}{k\cdot x } \left(1-{\rm e}^{{{\rm i}}\,k\cdot x}\right)\right ] , \\[1.6ex ] \lim_{k^{2}\to 0}{s^{(3 ) } } ( x ,- k)(-{{\rm i}}\,\overline{k } ) & = & \frac{1}{2}\frac{\rho^2}{x^2 } \frac{\left [ u x\overline{k } u^\dagger \right]}{k\cdot x } \left [ 1 - \frac{{{\rm i}}}{k\cdot x } \left(1-{\rm e}^{{{\rm i}}\,k\cdot x}\right)\right ] .\end{aligned}\ ] ] thus , on account of eq . ( [ sumsi ] ) , the on - shell residuum of the quark propagator ( [ si ] ) is given by eq .
( [ onshellsi ] ) .
a similar reasoning leads to eq .
( [ onshellsibar ] ) for the residuum of the quark propagator ( [ sibar ] ) .
our task is to derive eqs .
( [ tvertex ] ) , ( [ uvertex ] ) , corresponding to the @xmath204- quark vertex @xmath205 in the leading - order @xmath10-induced amplitude .
we will concentrate on the derivation of eq .
( [ tvertex ] ) , since the derivation of eq .
( [ uvertex ] ) is completely analogous .
let us recall the definition of @xmath206 , but now with indices written explicitly , @xmath207 inserting eqs .
( [ phibar ] ) and ( [ onshellsi ] ) into eq .
( [ phivertex1 ] ) we obtain for the vertex , @xmath208^\delta_{\ \lambda}\ , \left [ \left ( u^\dagger\right)^\gamma_{\ m } + \frac{1}{2}\,\frac{\rho^2}{x^2}\ , \frac{\left [ x\,\overline{k}\,u^\dagger\right]^\gamma_{\ m } } { k\cdot x}\ , \left ( 1-{\rm e}^{-{{\rm i}}\,k\cdot x}\right ) \right]\ , . \nonumber \end{aligned}\ ] ] the matrix structure in eq .
( [ phivertex2 ] ) can be simplified using @xmath209^\delta_{\ \lambda}\ , \left [ x\,\overline{k}\,u^\dagger\right]^\gamma_{\ m } = x^2\,\epsilon_{\gamma\delta}\,\left ( u^\dagger\right)^\gamma_{\ m}\ , \left [ k\,\overline{\sigma}_\mu \right]^\delta_{\ \lambda}\ , , \ ] ] which follows from the transposition rules of the @xmath55-matrices .
thus eq .
( [ phivertex2 ] ) can be rewritten as @xmath210^\delta_{\ \lambda } + \frac{1}{2}\,\rho^2\ , \left [ k\,\overline{\sigma}_\mu \right]^\delta_{\ \lambda}\ , \frac{1}{k\cdot x}\ , \left ( 1-{\rm e}^{-{{\rm i}}\,k\cdot x}\right ) \right]\ , . \nonumber\end{aligned}\ ] ] the remaining @xmath211 integration in eq .
( [ phivertex3 ] ) can be done with the help of the following formulae ( @xmath212 is always understood ) , @xmath213 \nonumber & & -2\,\pi^2\,{{\rm i}}\,\frac{q - k}{\left ( q - k\right)^2}\ , \rho\,\sqrt{\left ( q - k\right)^2}\ , k_1\left(\rho\,\sqrt{\left ( q - k\right)^2}\right)\ , , \\[2.4ex ] \label{int2 } \lefteqn { \int d^4x\,{\rm e}^{-{{\rm i}}\,(q - k)\cdot x}\ , \frac{1}{\left ( x^2+\rho^2\right)^2\,\left ( k\cdot x\right ) } = } \\[1.6ex ] \nonumber & & 2\,\pi^2\,{{\rm i}}\ , \frac{1}{q\cdot k}\,\frac{1}{\rho^2}\ , \rho\,\sqrt{\left ( q - k\right)^2}\ , k_1\left(\rho\,\sqrt{\left ( q - k\right)^2}\right)\ , , \\[2.4ex ] \label{int3 } \lefteqn { \int d^4x\,{\rm e}^{-{{\rm i}}\,q\cdot x}\ , \frac{1}{\left ( x^2+\rho^2\right)^2\,\left ( k\cdot x\right ) } = } \\[1.6ex ] \nonumber & & 2\,\pi^2\,{{\rm i}}\ , \frac{1}{q\cdot k}\,\frac{1}{\rho^2}\ , \rho\,\sqrt{q^2}\ , k_1\left(\rho\,\sqrt{q^2}\right)\ , .\end{aligned}\ ] ] by means of these basic integrals , we obtain finally for the vertex , @xmath214^\delta_{\ \lambda } } { ( q - k)^2}\ , \rho\,\sqrt{\left ( q - k\right)^2}\ , k_1\left(\rho\,\sqrt{\left ( q - k\right)^2}\right ) \nonumber \\ \mbox{}&&- \frac{1}{2}\ , \frac{\left [ k\,\overline{\sigma}_\mu \right]^\delta_{\ \lambda } } { q\cdot k}\ , \left [ \rho\,\sqrt{\left ( q - k\right)^2}\ , k_1\left(\rho\,\sqrt{\left ( q - k\right)^2}\right ) -\rho\,\sqrt{q^2}\ , k_1\left(\rho\,\sqrt{q^2}\right ) \right ] \biggr\}\ , , \nonumber \end{aligned}\ ] ] in accordance with eq .
( [ tvertex ] ) .
we would like to acknowledge helpful discussions with v. braun and v. rubakov .
a. belavin , a. polyakov , a. schwarz and yu .
tyupkin , . g.
t hooft , ; ; ( erratum ) .
a. ringwald , ; o. espinosa , .
m. mattis , ; p. tinyakov , ; r. guida , k. konishi and n. magnoli , . i. balitsky and v. braun , .
a. ringwald and f. schrempp , desy 94 - 197 , hep - ph/9411217 , in : _ quarks
94 _ , proc .
viiith int .
seminar , vladimir , russia , may 11 - 18 , 1994 , eds .
d. grigoriev _ et al .
170 - 193 .
m. gibbs , a. ringwald and f. schrempp , desy 95 - 119 , + hep - ph/9506392 , in : _ proc .
workshop on deep - inelastic scattering and qcd _ , paris , france , april 24 - 28 , 1995 , eds . j .-
f . laporte and y. sirois , pp .
341 - 344 .
a. ringwald and f. schrempp , desy 96 - 125 , hep - ph/9607238 , to appear in : _ proc .
workshop dis96 on deep - inelastic scattering and related phenomena _ , rome ,
italy , april 15 - 19 , 1996 .
m. gibbs , t. greenshaw , d. milstead , a. ringwald and f. schrempp , `` search strategies for instanton - induced processes at hera '' , to appear in : proc .
_ future physics at hera _ , 1996 .
h1 collaboration , s. aid _
et al . _ ,
desy 96 - 122 , hep - ex/9607010 , submitted to nucl .
i. balitsky and v. braun , .
t. appelquist and r. shankar , .
for further related discussions , see : + n. andrei and d. gross , ; l. baulieu , j. ellis , m. gaillard and w. zakrzewski , ; ; v. novikov , m. shifman , a. vainshtein and v. zakharov , ; m. dubovikov and a. smilga , . s. moch , a. ringwald and f. schrempp , to be published .
a. ringwald and f. schrempp , desy 96 - 203 , to appear in : _
quarks 96 _ , proc .
seminar , yaroslavl , russia , may 5 - 11 , 1996 . r. field , _ applications of perturbative qcd _ , ( addison - wesley , new york , 1989 ) . l. brown , r. carlitz , d. creamer and c. lee , . c. bernard , .
a. vainshtein , v. zakharov , v. novikov and m. shifman , .
t. morris , d. ross and c. sachrajda , .
i. balitsky , m. beneke and v. braun , .
f. yndurin , _ the theory of quark and gluon interactions _ , ( springer - verlag berlin heidelberg , 1993 ) .
c. callan , r. dashen and d. gross , .
n. andrei and d. gross , .
m. shifman , a. vainshtein and v. zakharov , .
j. collins , in : _ perturbative quantum chromodynamics _ , edited by a. mueller , ( world scientific , singapore , 1989 )
. i. balitsky , pennsylvania state univers .
preprint , psu - th-146 ( 94/05 ) , hep - ph/9405335 , in : _ continuous advances in qcd _ , proc . ,
minneapolis , usa , feb
. 18 - 20 , 1994 , ed .
a. smilga , pp.167 - 194 . | instanton calculations in qcd are generically plagued by infrared divergencies associated with the integration over the instanton size @xmath0 . here
, we demonstrate explicitly that the typical inverse hard momentum scale @xmath1 in deep - inelastic scattering provides a dynamical infrared cutoff for the size parameter @xmath0 .
hence , deep - inelastic scattering may be viewed as a distinguished process for studying manifestations of qcd - instantons . for clarity , we restrict the explicit discussion to the simplest chirality - violating process , @xmath2 .
we calculate the corresponding fixed angle cross - section as well as the contributions to the gluon structure functions , @xmath3 and @xmath4 , within standard instanton perturbation theory in leading semi - classical approximation . to this approximation ,
fixed - angle scattering processes at high @xmath5 are reliably calculable . in the bjorken limit
, the considered instanton - induced process gives a scaling contribution to @xmath6 and the callan - gross relation holds . |
at the turn of the 20th century studies of heat capacity of solids at low temperatures played an important role in revealing the quantum character of nature .
@xcite this example vividly illustrates the power of heat capacity measurements in physics . as a matter of fact measurements of heat capacity reveal so great a deal of information about matter that calorimetry has become an indispensable tool for modern day research in chemistry , physics , materials science , and biology .
unfortunately , however , calorimetry is a relatively insensitive method , and it is particularly difficult to obtain an accurate absolute value of heat capacity of samples with minute masses . since new materials with interesting physical properties are usually not synthesized in quantity , it is of extreme necessity for the advance of materials science or condensed matter physics to secure a convenient means to measure absolute heat capacity of sub - milligram samples . on another front , the generalization of calorimetry into the dynamic regime has attracted wide attention in recent years .
@xcite although measurement of a thermodynamic quantity is the usual notion that is tied to calorimetry , it is possible to go beyond this traditional understanding and generalize heat capacity as a dynamic quantity .
the concept of dynamic heat capacity appears natural if one recalls that static thermodynamic quantities are time - averaged ( or ensemble - averaged ) . in other words , they are static not because they do not change in time , but because they change too rapidly on the experimental time scale . then , suppose that a system contains a dynamic process relaxing with a characteristic time which lies within our experimental time window , this will result in a time - dependent ( or frequency - dependent ) heat capacity depending on the time scale of measurements .
conversely , measurements of dynamic heat capacity of condensed matter would provide insights which may not be available to other dynamic probes .
thus , the development of a convenient dynamic calorimeter for solid samples appears to be of great necessity . in this paper
, we describe a new type of the ac calorimeter , termed _ peltier ac calorimeter _ ( pac ) , which fills both needs described above ; pac is not only a microcalorimeter capable of measuring heat capacity of sub - milligram samples but also a dynamic calorimeter with wide dynamic range .
for the heat capacity measurements of a minute sample adiabatic calorimetry @xcite does not appear to be suitable due to the so - called addenda problem ; in other words , calorimetry requires indispensable addenda ( heater and sensor ) to be put on a sample and the mass of the addenda may even be greater than that of the sample in the case of minute samples . although one may expect the same kind of problem in ac calorimetry , @xcite we have devised a way of avoiding the addenda problem in a new ac calorimeter by utilizing the peltier effect of extremely thin thermocouple wires .
it is also obvious that an ac calorimeter has a potential of being a dynamic one .
suppose that a voltage difference @xmath0 and/or a temperature difference @xmath1 exist across a metallic wire , the electric current @xmath2 and the heat current @xmath3 through the wire can be expressed as @xcite @xmath4 @xmath5 where @xmath6 is the resistance , @xmath7 the thermoelectric power , @xmath8 the peltier coefficient , and @xmath9 the thermal conductance .
the thermoelectric power and the peltier coefficient are related by @xmath10 .
now if an electric current is run through a thermocouple , consisting of two distinct metal wires , under an isothermal condition , then the junction acts as either a heat sink or a heat source depending on the current direction .
the heat current @xmath3 caused by this peltier effect is directly proportional to @xmath2 .
if an ac electric current at an angular frequency @xmath11 , @xmath12 , is applied to a thermocouple under an isothermal condition , an ac power oscillation at the same frequency will be induced at the junction by the peltier effect .
the amplitude @xmath13 of the ac power is equal to @xmath14 ; @xmath13 can be accurately determined , since @xmath15 is well tabulated for thermocouples and the ac current can be measured with high precision .
thus it is evident that the peltier effect of a thermocouple junction can be utilized as a power source for ac calorimetry .
various ac calorimetric techniques with non - contact energy sources such as chopped light or light emitting diode were developed previously .
@xcite however , it should be pointed out that these power sources can only supply heat and can not act as a heat sink . as a consequence of this ,
the average temperature of a sample is always above that of the heat bath in these traditional ac calorimetric methods .
( this so - called dc shift is equal to the average dc power divided by the thermal conductance between the sample and the bath . ) in addition , it is not easy in traditional ac calorimetry to determine an absolute value of heat capacity due to inaccuracy in the determination of input power and heat leak . on the other hand , the pac which utilizes the peltier effect as an ac power source is free from these difficulties : first , it is capable of both heating and cooling .
this means that there is no dc shift in temperature , and lack of the dc shift in turn has an important implication on the working frequency range of the ac calorimeter .
( see below . )
second , the power generated at a thermocouple junction can be measured with high accuracy .
third , the mass of the thermocouple junction attached to the sample is entirely negligible in most situations .
all these factors make the pac superior to previous methods in that the experimental setup is simple , it directly yields absolute values of heat capacity of sub - milligram samples , and it may be used as a dynamic calorimeter .
the implementation of the principle described in the previous section is rather straightforward ; the schematic diagram and the photograph of the peltier ac calorimeter we constructed is shown in fig . 1 .
we made thermocouple junctions by spot - welding with chromel and constantan wires of 25 @xmath16 ( or 12 @xmath16 ) in diameter .
a couple of thermocouple junctions ( tc1 ) were connected via copper wires to a function generator supplying an ac electric current , which was measured by a digital ammeter with 1 na sensitivity . using a very small amount of ge 7031 varnish for electrical insulation and good thermal contacts ,
one of the junctions was attached to one side of a sample ( typically of linear size less than 1 mm ) and the other to a copper block ( heat bath ) .
the heat bath was then attached to a closed - cycle he refrigerator and its temperature was controlled within @xmath17 2 mk stability in the range of 15420 k. a digital voltmeter was used to read the voltage difference across another thermocouple junction ( tc2 ) attached to the other side of the sample . from the voltage readings , we could measure the sample temperature with the sensitivity of 1.67 mk at 15 k and 0.14 mk at 420 k. since the mass of the thermocouple junctions and the varnish is completely negligible compared to the sample mass even for sub - milligram samples , the amplitude of the temperature oscillation @xmath18 at the imposed frequency ( measured by tc2 ) can be written as @xcite @xmath19 where @xmath20 is the sample heat capacity , @xmath21 the external or sample - to - bath relaxation time , @xmath22 the internal diffusion time in the sample , @xmath23 the thermal conductance of the link between the sample and the bath , and @xmath24 the thermal conductance of the sample . since only the thin thermocouple wires of diameter 12 or 25 @xmath25 m provide paths for heat conduction from the sample to the heat bath , @xmath23 is extremely small and negligible compared to @xmath24 . and the third term in the parenthesis of eq .
( [ eq_accal ] ) can be neglected .
thus , if one can select the frequency range of @xmath26 , heat capacity may be directly obtained from @xmath27 .
in addition to providing a means for measuring the absolute value of heat capacity of minute solid samples , the pac possesses a distinct ability to function as a dynamic calorimeter .
this capability of the pac stems from the fact that @xmath23 is extremely small as noted above ( heat conduction paths are provided by microns thick thermocouple wires only . ) and thus @xmath21 is exceedingly large .
large @xmath21 then allows a wide frequency range where the relationship @xmath28 holds .
this situation is contrasted to that of traditional ac calorimeters where one is required to have a reasonable size of @xmath23 , because otherwise the dc shift of the sample due to a dc power would become prohibitively large .
( remember that the dc shift is given by the dc power divided by @xmath23 . )
this size requirement of @xmath23 then places a limit to the working frequency range of the ac calorimeter by decreasing @xmath21 .
note that the origin of this limitation of the traditional ac calorimeter goes to the fact that the energy source is only capable of heating , but not cooling .
on the other hand , the energy source of the pac is able to both heat and cool , the average temperature of the sample is the same as the bath temperature , and the pac can afford to have exceedingly small @xmath23 and long @xmath21 .
the performance of the pac was tested with small pieces of synthetic sapphire ( @xmath29-@xmath30 ) , the standard material designated by nist .
@xcite we first examined the induced temperature oscillation in tc2 , in response to an oscillating electric current in tc1 , at 30 k , 150 k , and 320 k with a test sample of mass 0.54 mg and dimension [email protected]@xmath310.3 mm@xmath32 .
the frequency and amplitude of the electric current were 0.25 hz and 0.4 ma , respectively . from fig .
2(a ) and ( b ) , it is seen that nice temperature oscillations at the applied frequency are obtained at 150 k and 320 k. however , fig .
2(c ) reveals that a significant amount of the second harmonic appears at 30 k. it is readily clear that this second harmonic originates from joule heating in the thermocouple wires . to precisely determine @xmath18 which will yield heat capacity ,
the raw data were fitted to a sinusoidal function containing the fundamental and second harmonic terms .
2(d ) shows the power amplitude at the fundamental and second harmonics as a function of temperature .
the power amplitude at the fundamental were converted from the measured @xmath33 using the table for @xmath15 .
@xcite the joule heating part was obtained from the fitting procedure . above 50 k
, the joule heating effect is completely negligible ; however , there appears an increasing amount of the second harmonic as temperature decreases below 50 k. this is probably caused by the reduction in thermal resistance of the thermocouple wires at low temperatures , and thus part of the heat generated along the wires flows back toward the sample .
however , the existence of the second harmonic does not cause too much problem in the heat capacity measurements , since the governing equation for the present problem is linear and therefore one needs only to measure the signal at the fundamental frequency .
nevertheless , it is desirable to reduce the joule heating as much as possible to attain the sensitivity . the dynamic characteristics of the pac was then checked by measuring the frequency dependence of @xmath18 for the test sample ( 0.54 mg ) at a fixed amplitude of the applied current .
3 is the plot of the results obtained at three temperatures .
it is seen from the figure that @xmath18 is proportional to @xmath34 in the whole measured frequency range at high temperatures ( 150 k and 320 k ) , and the figure clearly illustrates that the pac is indeed a dynamic calorimeter . the data obtained at 30 k , however , shows a large deviation from the @xmath34 behavior at frequencies below 0.13 hz .
this is caused by the fact that the smallness of @xmath23 is compensated by a rapid decrease in heat capacity at low temperatures , and the external relaxation time @xmath21 becomes short enough to be comparable to the oscillation period at 0.13 hz .
this compensation effect is unavoidable and the dynamic capability of the pac is of limited use below roughly liquid nitrogen temperature .
it may be also noted that the deviation from the @xmath34 behavior is also expected at high frequencies , since our power source is an extremely local one and the internal diffusion time will interfere above a certain frequency .
this high cutoff should be size - dependent , and present no problem for small samples . in order to ascertain the capability of the pac as a microcalorimeter
, we carried out the heat capacity measurements for two test samples of @xmath29-@xmath30 with mass 0.54 mg and 2.25 mg .
the measuring frequency was set at 0.25 hz and the measured temperature range was from 15 k to 420 k. it is stressed that background subtraction , required in most calorimetric methods , is not necessary for the pac .
4 ( a ) is the plot of the heat capacity data for two sapphire samples .
the two sets of data coincide very well and display excellent reproducibility of the pac .
also plotted in the figure is the reference data ( @xmath35 ) of the same material .
@xcite the agreement is again excellent in the whole temperature range from 15 k to 420 k. in order to estimate the accuracy and precision of the pac , we plotted the residual heat capacity values , _
i.e. _ , ( @xmath36 ) in fig .
4(b ) . from the figure
, we estimate the absolute accuracy for heat capacity of a sub - milligram sample to be @xmath37% for the temperature range of 30150 k and @xmath38% for 150420k . at temperatures below 30 k ,
the absolute accuracy worsens due to very small values of heat capacity of @xmath29-@xmath30 and becomes even more than 10% .
however , the precision is better than 0.5% in the whole temperature range of 15420 k.
having demonstrated the capability of the pac as a microcaloriemter and dynamic calorimeter , we briefly outline the possible extensions of the pac , which are currently under development .
first of all , it should be noted that the pac would function as a microcalorimeter even at helium temperatures if cryogenic thermocouple wires of au - fe or cu - fe are used .
this replacement of thermocouples would have an additional effect of suppressing unwanted joule heating .
a more novel extension of the pac would be the peltier thermal microscope , which would enable to measure local thermophysical properties of matter at submicron length scales .
here it is proposed that the tip for a atomic force microscope is replaced by a thermocouple tip .
an important feature of our proposal is that the thermocouple tip here is not just a temperature sensor , but it plays dual roles of a heater and sensor .
for this purpose , we have shown that a single junction can indeed be used as both a heat source and sensor simultaneously .
@xcite the successful development of the peltier thermal microscope would be an exciting and important event for this so - called nano - age when the submicron local thermophysical properties are in great demand .
a. einstein : ann .
physik * 22 * , 180 ( 1907 ) .
y. h. jeong : thermochimica acta * 304/305 * , 67 ( 1997 ) .
w. nernst , ann .
physik * 36 * , 395 ( 1911 ) .
p. f. sullivan and g. seidel , phys
. rev . * 173 * , 679 ( 1968 ) .
h. b. callen : _ thermodynamics and an introduction to thermostatistics _ , john wiley & sons ( 1985 )
. i. hatta , pure & appl . chem . * 64 * , 79 ( 1992 ) .
n. overend , m. a. howson , i. d. lawrie , s. abell , p. j. hirst , c. changkang , s. chowdhury , j. w. hodby , s. e. inderhees , and m. b. salamon , phys .
b * 54 * , 9499 ( 1996 ) . k. f. sterrett , d. h. blackburn , a. b. bestul , s. s. chang , and j. horman , j. res .
* 69c * , 19 ( 1965 ) ; s. s. chang , proc .
seventh symp .
, a. cezairliyan ( ed . ) 75 , asme ( 1977 ) . or 12 @xmath39 are used .
a copper block plays a role of heat bath .
a function generator applies an oscillating current to ch - cn thermocouples ( tc1 ) in contact with the sample , and a digital voltmeter measures an ensuing voltage oscillation from another thermocouple ( tc2 ) attached to the sample .
( b ) the photograph of the peltier ac calorimeter .
the linear dimension of the sample is approximately 1 mm , and the diameter of the thermocouple wires is 25 @xmath39.,title="fig:",height=226 ] or 12 @xmath39 are used .
a copper block plays a role of heat bath .
a function generator applies an oscillating current to ch - cn thermocouples ( tc1 ) in contact with the sample , and a digital voltmeter measures an ensuing voltage oscillation from another thermocouple ( tc2 ) attached to the sample .
( b ) the photograph of the peltier ac calorimeter .
the linear dimension of the sample is approximately 1 mm , and the diameter of the thermocouple wires is 25 @xmath39.,title="fig:",height=226 ] [email protected] mm@xmath32 .
the broken line of ( b ) shows the electric current oscillation .
the solid lines indicate the fitting results with the fundamental and second harmonics , and @xmath40 represents the dc component .
( d ) the amplitude of the power oscillations at the fundamental ( peltier ) and second harmonics ( joule).,width=377 ] is plotted as a function of frequency .
the thick solid line represents the @xmath34 behavior .
note that the the @xmath34 law is well obeyed in the whole frequency range at high temperatures ; the pac is able to function as a dynamic calorimeter .
the arrow indicates the low cutoff frequency below which the deviation from the @xmath34 behavior appears at low temperatures.,width=377 ] | a new ac calorimeter , utilizing the peltier effect of a thermocouple junction as an ac power source , is described .
this peltier ac calorimeter allows to measure the absolute value of heat capacity of small solid samples with sub - milligrams of mass .
the calorimeter can also be used as a dynamic one with a dynamic range of several decades at low frequencies . |
principal component analysis ( pca ) is one of the oldest @xcite and most fundamental techniques in data analysis , enjoying ubiquitous applications in modern science and engineering @xcite .
given a data matrix @xmath1 of @xmath2 data points of dimension @xmath3 , pca gives a closed form solution to the problem of fitting , in the euclidean sense , a @xmath4-dimensional linear subspace to the columns of @xmath5 . even though the optimization problem associated with pca is non - convex ,
it does admit a simple solution by means of the singular value decomposition ( svd ) of @xmath5 .
in fact , the @xmath4-dimensional subspace @xmath6 of @xmath7 that is closest to the column span of @xmath5 is precisely the subspace spanned by the first @xmath4 left singular vectors of @xmath5 .
using @xmath6 as a model for the data is meaningful when the data are known to have an approximately linear structure of underlying dimension @xmath4 , i.e. they lie close to a @xmath4-dimensional subspace @xmath8 . in practice ,
the principal components of @xmath5 are known to be well - behaved under mild levels of noise , i.e. , the angle between @xmath6 and @xmath8 is relatively small and more importantly @xmath6 is optimal when the noise is gaussian @xcite .
however , in the presence of even a few outliers in @xmath5 , i.e. , points whose angle from the underlying ground truth subspace @xmath8 is large , the angle between @xmath8 and its estimate @xmath6 will in general be large .
this is to be expected since , by definition , the principal components are orthogonal directions of maximal correlation with _ all _ the points of @xmath5 .
this phenomenon , together with the fact that outliers are almost always present in real datasets , has given rise to the important problem of outlier detection in pca .
traditional outlier detection approaches come from robust statistics and include _ influence - based detection _ , _ multivariate trimming _ , @xmath9__-estimators _ _ , _ iteratively _ _ weighted recursive least squares _ and _ random sampling consensus _ ( ransac ) @xcite .
these methods are usually based on non - convex optimization problems , admit limited theoretical guarantees and have high computational complexity ; for example , in the case of ransac many trials are required .
recently , two attractive methods have appeared @xcite with tight connections to _ compressed sensing _
@xcite and _ low - rank representation _ @xcite .
both of these methods are based on convex optimization problems and admit theoretical guarantees and efficient implementations .
remarkably , the self - expressiveness method of @xcite does not require an upper bound on the number of outliers as the method of @xcite does .
however , they are both guaranteed to succeed only in the low - rank regime : the dimension @xmath4 of the underlying subspace @xmath8 associated to the inliers should be small compared to the ambient dimension @xmath3 . in this paper
we adopt a _ dual _ approach to the problem of robust pca in the presence of outliers , which allows us to transcend the low - rank regime of modern methods such as @xcite .
the key idea of our approach comes from the fact that , in the absence of noise , the inliers lie inside any hyperplane @xmath10 that contains the underlying linear subspace @xmath8 .
this suggests that , instead of attempting to fit directly a low - dimensional linear subspace to the entire data set , as done e.g. in @xcite , we can search for a hyperplane @xmath11 that contains as many points of the dataset as possible .
when the inliers are in general position inside the subspace , and the outliers are in general position outside the subspace , this hyperplane will ideally contain the entire set of inliers together with possibly a few outliers . after removing the points that do not lie in that hyperplane
, the robust pca problem is reduced to one with a potentially much smaller outlier percentage than in the original dataset .
in fact , the number of outliers in the new dataset will be at most @xmath12 , an upper bound that can be used to dramatically facilitate the outlier detection process using existing methods .
we think of the direction @xmath13 of the normal to the hyperplane @xmath11 as a _ dual principal component _ of @xmath5 , as ideally it is an element of @xmath14 .
naturally , one can continue by finding a second dual principal component by searching for a hyperplane @xmath15 , with @xmath16 , that contains as many points as possible from @xmath17 , and so on , leading to a _ dual principal component analysis _ of @xmath5 .
we pose the problem of searching for such hyperplanes as an @xmath18 cosparsity - type problem , which we relax to a non - convex @xmath0 problem on the sphere .
we provide theoretical guarantees under which every global solution of that problem is a dual principal component .
more importantly , we relax this non - convex optimization problem to a sequence of linear programming problems , which , after a finite number of steps , yields a dual principal component . experiments on synthetic data demonstrate that the proposed method is able to handle more outliers and higher dimensional subspaces than the state - of - the - art methods @xcite .
we begin by establishing our data model in section [ subsection : datamodel ] , then we formulate our dpcp problem conceptually and computationally in sections [ subsection : conceptualformulation ] and [ subsection : computationalformulation ] , respectively .
we employ a deterministic noise - free data model , under which the inliers consist of @xmath19 points @xmath20 \in \re^{d \times n}$ ] that lie in the intersection of the unit sphere @xmath21 with an unknown proper subspace @xmath8 of @xmath7 of unknown dimension @xmath4 . accordingly , the outliers consist of @xmath9 arbitrary points @xmath22 \in \re^{d \times m}$ ] that lie on @xmath21 .
the dataset , that we assume given , is @xmath23 \boldsymbol{\gamma } \in \re^{d \times l}$ ] , where @xmath24 and @xmath25 is some permutation , indicating that the partition of the columns of @xmath26 into @xmath5 and @xmath27 is unknown .
we further assume that the columns of @xmath26 are in _ general position _ in the following sense : first , any @xmath4-tuple of inliers and any @xmath3-tuple of outliers is linearly indepenent .
second , for any @xmath28 , of which at most @xmath29 come from @xmath5 , the hyperplane of @xmath7 spanned by @xmath30 does not contain any of the remaining points .
notice that in our data model we have made no assumption about the dimension of @xmath8 : indeed , @xmath8 can be anything from a line to a @xmath31-dimensional hyperplane .
ideally , we would like to be able to partition the columns of @xmath26 into those that lie in @xmath8 and those that do nt .
but under such generality , this is not a well - posed problem since @xmath5 lies inside every subspace that contains @xmath8 , which in turn may contain some elements of @xmath27 . in other words ,
given @xmath26 and without any other a - priori knowledge , it may be impossible to correctly partition @xmath26 into @xmath5 and @xmath27 .
instead , we formulate the following well - posed problem : [ prb : partition ] partition the columns of @xmath32 into two groups , such that one of the groups is a subset of @xmath26 with maximal cardinality , with respect to the property of lying inside a @xmath31-dimensional hyperplane of @xmath7 .
the usefulness of this formulation is that for large values of @xmath33 , where known methods for outlier detection in pca fail , one of the groups , say @xmath34 will contain the entire @xmath5 together with precisely @xmath35 columns of @xmath27 , while the other group , say @xmath36 , will contain the remaining @xmath37 columns of @xmath27 .
note that the first group is _ structured _ in the sense that it must lie in a hyperplane and so in general @xmath38 .
having the partition @xmath39 , we can reject the unstructured group @xmath36 and reconsider the robust pca problem on the group @xmath34 .
but now the number of outliers has decreased from @xmath40 to @xmath35 .
in fact , we can use the upper bound @xmath12 on the number of outliers to dramatically facilitate the outlier detection process using other existing methods .
a natural approach towards solving problem [ prb : partition ] is to solve @xmath41 the idea behind is that a hyperplane @xmath42 contains a maximal number of columns of @xmath26 if and only if @xmath43 is as sparse as possible . since is intractable ,
consider @xmath44 notice that the objective in is convex , while the constraint @xmath45 is non - convex , thus leading to a non - smooth and non - convex optimization problem .
[ prb : nonconvex ] when is every global solution @xmath46 of orthogonal to @xmath47 ? how can we efficiently solve ? in this paper , we propose to relax by a sequence of linear programs of the form @xmath48 where @xmath49 is some arbitrary vector and @xmath50 indicates normalization to unit @xmath51-norm .
we naturally ask : [ prb : convexrelaxations ] under what conditions does the sequence of converge to a vector @xmath52 that is orthogonal to @xmath47 ?
in this section , we aim to familiarize the reader with the state - of - the - art of outlier detection in modern single subspace learning ( section [ subsection : outlierpca ] ) , as well as give a brief overview ( section [ subsection : dl ] ) of existing work , that relates technically to the problems of interest of this paper , i.e. problems and .
one of the oldest and most popular outlier detection methods in pca is _ random sampling consensus ( ransac ) _
the idea behind ransac is simple : alternate between randomly sampling @xmath53 points from the dataset and computing a subspace model for these points , until a model is found that fits a maximal number of points in the entire dataset within some error @xmath54 .
ransac is usually characterized by high performance , when not both @xmath53 and the oultier percentage are large ; otherwise it requires a high computational time , particularly when @xmath4 is unknown and @xmath53 is allowed to vary , since exponentially many trials are required in order to sample outlier - free subsets , and thus obtain reliable models . moreover , its performance is very sensitive on in the input parameters @xmath53 and @xmath54 . among many other outlier detection methods ( see section [ section : introduction ] ) , in the remaining of this section
we will focus on the modern low - rank / sparse - representation theoretic methods of @xcite and @xcite , which we will later use experimentally to compare against our proposed method .
the first method @xcite , referred to as l21 , is a variation of the _ robust pca _ algorithm of @xcite , which computes a @xmath55-norm decomposition denotes the nuclear norm of a matrix , i.e. , the sum of its singular values , and @xmath56 is defined as the sum of the euclidean norms of the columns of a matrix . ] of the data matrix , instead of the @xmath57-decomopsition in @xcite .
more specifically , l21 solves the convex optimization problem @xmath58 it is shown in @xcite that , under certain conditions , the optimal solution to this problem is of the form @xmath59 \boldsymbol{\gamma}$ ] and @xmath60 \boldsymbol{\gamma}$ ] .
that is , the nonzero columns of the @xmath61 matrix give the inliers and the nonzero columns of the @xmath62 matrix give the outliers .
however , the theoretical conditions require the intrinsic dimension @xmath63 and the outlier percentage to be small enough .
the second method that we consider , referred to as se , is based on the _ self - expressiveness _ property of the data matrix , a notion popularized by the work of @xcite in the area of subspace clustering @xcite .
more specifically , if a column of @xmath26 is an inlier , then it can in principle be expressed as a linear combination of @xmath4 other columns of @xmath26 , which are inliers .
if the column is instead an outlier , then it will in principle be expressible as a linear combination of not less than @xmath3 other columns . to encourage each point to express itself as a linear combination of the smallest number of other data points ,
the following convex optimization problem is solved : @xmath64 if @xmath4 is small enough with respect to @xmath3 , an element is declared as an outlier if the @xmath0 norm of its coefficient vector in @xmath65 is large ; see @xcite for an explicit formula .
se admits theoretical guarantees @xcite and efficient admm implementations @xcite .
however , as it is clear from its description , it is expected to succeed only when @xmath4 is sufficiently small .
in contrast though to l21 , se has the remarkable property that it can , in principle , handle an arbitrary number of outliers .
problems of the form @xmath66 and variants of its relaxations have appeared on several occasions and in diverse contexts in the literature , but are much less understood than the now classic sparse @xcite and cosparse @xcite problems of the form @xmath67 respectively .
the main source of difficulty is that , in contrast to , obtaining tight convex relaxations of is a hard problem .
one of the first instances where was considered was in the context of blind source separation @xcite , where it was proposed to relax it with the problem @xmath68 this is still a non - convex problem , and a heuristic based on quadratic programming was proposed to solve it .
it was not until very recently , that the convex relaxation @xmath69 was proposed , with @xmath70 taken to be a row or a sum of two rows of @xmath71 , and theorems of correctness were given in the context of dictionary learning @xcite . notice that our proposed convex relaxations can be seen as a generalization of . in the context of finding the sparsest vector in a subspace , which is intrinsically related to dictionary learning
, an alternating direction minimization scheme was proposed in @xcite to solve a relaxation of the form @xmath72 remarkably , under some mild conditions , this was shown to converge with high probability to a global solution of @xmath73 the geometry of was further studied in a probabilistic framework in the recent @xcite , after replacing the @xmath0-norm with a smooth surrogate .
in this section we state and discuss our main theoretical results , regarding problems and . before doing so though
, we need to introduce additional notation and draw some interesting connections with the field of numerical integration on the sphere ( section [ subsection : integrationperspective ] ) . to begin with , for a vector @xmath45 , denote by @xmath74 the function @xmath75 .
then given a set of @xmath2 points @xmath76 , the quantity @xmath77 is a discrete approximation of the integral @xmath78 where @xmath79 is the uniform measure on @xmath21 and @xmath80 is the mean height of the unit hemisphere of @xmath7 , given in closed form by @xmath81 where the double factorial is defined as @xmath82 a useful fact is that @xmath80 is a decreasing function of @xmath3 and in fact tends to zero as @xmath3 goes to infinity .
now , observe that because of the symmetry of @xmath21 , the integral in does not depend on @xmath83 .
however , the _ integration error _ @xmath84 does depend both on the direction of @xmath83 as well as the distribution of the points @xmath85 on @xmath21 .
it is clear though , that the more uniformly the points are distributed , the smaller will be the dependence of the integration error on the direction of @xmath83 .
we note here that the notion of uniform point set distribution on the sphere is a non - trivial one . in a deterministic setting
, this is an active subject of study in the fields of combinatorial geometry and numerical integration on the sphere @xcite .
a widely used measure of the uniformity of a point set on the sphere is the so - called _ point set discrepancy _
@xmath86 of the set , which can be defined in terms of spherical harmonics as @xmath87 where @xmath88 is the dimension of the vector space of spherical harmonics of order @xmath89 , and @xmath90 is the @xmath91-th basis element .
it is then a fact that the integration error is small if and only if @xmath86 is small .
as before , for any @xmath45 we define a vector valued function @xmath92 by @xmath93 .
note that the image of @xmath94 is @xmath95 and that points that are orthogonal to @xmath83 are mapped to @xmath96 .
moreover , [ lem : vectorintegral ] @xmath97 .
this result suggests that the quantity @xmath98 can be interpreted as a discrete approximation of the integral @xmath99 and so the more uniformly distributed are the points @xmath85 , the closer @xmath100 is to the quantity @xmath101 .
the above discussion motivates defining the quantities @xmath102 and @xmath103 , to capture the uniformity of outliers and inliers , respectively : @xmath104 before we consider the _ discrete _ non - convex problem , it is instructive to examine its continuous counterpart @xmath105 where @xmath106 is the uniform measure on @xmath107 .
of course this problem is only of theoretical interest and serves in establishing a first intuition for the idea behind .
in fact , [ thm : continuousnonconvex ] any global solution to problem must be orthogonal to @xmath8 .
the proof of the above theorem follows easily from the symmetry of the sphere , since the first integral appearing in does not depend on @xmath83 , while the second integral depends only on the angle of @xmath83 from @xmath8 .
theorem [ thm : continuousnonconvex ] suggests that under sufficiently well - distributed point sets of inliers and outliers , any global solution to the discrete problem should _ also _ be orthogonal to the span of the inliers . before stating the precise result
, we need one last piece of notation : for a set @xmath108 \subset \cs^{d-1}$ ] and integer @xmath109 , define @xmath110 to be the maximum circumradius among all polytopes @xmath111 , where @xmath112 are distinct integers in @xmath113 $ ] , and @xmath114 indicates the convex hull operator .
[ thm : discretenonconvex ] suppose that the quantity @xmath115 satisfies @xmath116 for all positive integers @xmath117 such that @xmath118 .
then any global solution @xmath46 to will be orthogonal to @xmath47 . towards interpreting this result ,
consider first the asymptotic case where we allow @xmath19 and @xmath9 to go to infinity , while keeping the ratio @xmath119 constant
. under point set uniformity , i.e. under the hypothesis that @xmath120 and @xmath121 , we will have that @xmath122 and @xmath123 , in which case is satisfied .
this suggests the interesting fact that when the number of inliers is a linear function of the number of outliers , then will always give a normal to the inliers even for arbitrarily large number of outliers and irrespectively of the subspace dimension @xmath4 .
along the same lines , for a given @xmath119 and under the point set uniformity hypothesis , we can always increase the number of inliers and outliers ( thus decreasing @xmath103 and @xmath102 ) , while keeping @xmath119 constant , until is satisfed , once again indicating that is possible to yield a normal to the space of inliers irrespectively of their intrinsic dimension .
+ + + + + + in this section we consider the sequence of convex relaxations ; in particular , there are two important issues to be addressed .
first , note that relaxing the constraint @xmath124 in with a linear constraint @xmath125 as in , has already been found to be of limited theoretical guarantees @xcite .
so it is natural to ask whether the idea of considering a sequence of such relaxations @xmath126 has an intrinsic merit or not , irrespectively of the data distribution .
for example , if the data is _ perfectly well distributed _ , yet the sequence does not yield vectors orthogonal to the inlier space , then we will know that a - priori the method is limited .
fortunately , this is not the case : when the data is perfectly well distributed , i.e. when we restrict our attention to the continuous analog of , given by @xmath127 , \label{eq : convexrelaxationscontinuousraw}\end{aligned}\ ] ] then the sequence @xmath128 achieves the property of interest : [ thm : convexrelaxationscontinuous ] consider the sequence of vectors @xmath128 generated by recursion , where @xmath129 is arbitrary .
let @xmath130 be the corresponding sequence of angles from @xmath8 .
then @xmath131 , provided that @xmath132 .
this result suggests that relaxing @xmath124 with the sequence @xmath133 is intrinsically the right idea .
the second issue is how the distribution of the data affects the ability of this sequence of relaxations to give vectors orthogonal to @xmath8 .
the answer is given by theorem [ thm : discreteconvexrelaxations ] , which says that when the angle between @xmath49 and @xmath8 is large enough and the data points are well distributed , the sequence will consist of vectors orthogonal to the inlier space , for sufficiently large indices @xmath134 .
[ thm : discreteconvexrelaxations ] let @xmath135 be the angle between @xmath49 and @xmath8 .
suppose that condition on the outlier ratio @xmath119 holds true and consider the vector sequence @xmath136 generated by recursion .
then after a finite number of terms @xmath137 , for some @xmath109 , every term of @xmath136 will be orthogonal to @xmath138 , providing that @xmath139 first note that if is true , then the expression of always defines an angle between @xmath140 and @xmath141 .
second , theorem [ thm : discreteconvexrelaxations ] can be interpreted using the same asymptotic arguments as theorem [ thm : discretenonconvex ] . in particular , notice that the lower bound on the angle @xmath135 tends to zero as @xmath142 go to infinity with @xmath119 constant .
note also that this result does not show convergence of the sequence @xmath143 : it only shows that this sequence will eventually satisfy the desired property of being orthogonal to the space of inliers ; a convergence result remains yet to be established .
so far we have established a mechanism of obtaining an element @xmath13 of @xmath14 , where @xmath144 : run the sequence of linear programs until the function @xmath145 converges within some small @xmath146 ; then assuming no pathological point set distributions , any vector @xmath143 can be taken as @xmath13 .
there are two possibilities : either @xmath8 is a hyperplane of dimension @xmath147 or @xmath148 . in the first case , @xmath13 is the unique up to scale element of @xmath14 , which proves that in this case the sequence of in fact converges . in such a case , we can identify our subspace model with the hyperplane defined by the normal @xmath13 .
next , if @xmath148 , we can proceed to find a second element @xmath149 of @xmath14 that is orthogonal to @xmath13 and so on .
this naturally leads to the _ dual principal component pursuit _ shown in algorithm [ alg : dpcp ] .
@xmath150 ; @xmath151 ; @xmath152 ; @xmath153 ; @xmath154 [ dpcp : update ] ; @xmath155 ; @xmath156 ; @xmath157 ; @xmath158 ; a few comments are in order . in algorithm
[ alg : dpcp ] , @xmath159 is an estimate for the codimension @xmath160 of the inlier subspace @xmath47 .
if @xmath159 is rather large , then in the computation of each @xmath161 , it is more efficient to reduce the coordinate representation of the data by replacing @xmath26 with @xmath162 , where @xmath163 is the orthogonal projection onto @xmath164 , and solve the linear program in step [ dpcp : update ] in the projected space . notice further how the algorithm initializes
@xmath49 : this is effectively the right singular vector of @xmath165 , that corresponds to the smallest singular value . as it will be demonstrated in section [ section : experiments ] , this choice has the effect that the angle of @xmath49 from the inlier subspace is typically large , in particular , larger than the smallest initial angle required for the success of the principal component pursuit of .
in this section we investigate experimentally the proposed dpcp alg .
[ alg : dpcp ] . using both synthetic ( subsection [ subsection : synthetic ] ) and real data ( subsection [ subsection : real ] ) , we compare dpcp to the three methods se , l21 and ransac discussed in section [ subsection : outlierpca ] as well as to the method of eq . discussed in section [ subsection : dl ] , which we will refer to as svs ( _ sparsest vector in a subspace _ ) .
the parameters of the methods are set to fixed values , chosen such that the methods work well across all tested dimension and outlier configurations .
in particular , we use @xmath166 and @xmath167 ; see @xcite and @xcite for details . regarding dpcp , we fix @xmath168 , and unless otherwise noted , we set @xmath159 equal to the true codimension of the subspace .
to begin with , we evaluate the performance of dpcp in the absence of noise , for various subspace dimensions @xmath169 and outlier percentages @xmath170 .
we fix the ambient dimension @xmath171 , sample @xmath172 inliers uniformly at random from @xmath173 and @xmath9 outliers uniformly at random from @xmath21 .
we are interested in examining the ability of dpcp to recover a single normal vector ( @xmath174 ) to the subspace , by means of recursion .
the results are shown in fig .
[ figure : dpcp ] for @xmath175 independent trials .
[ figure : dpcp_theory ] shows whether the theoretical conditions of are satisfied or not . in checking these conditions , we estimate the abstract quantities @xmath176 by monte - carlo simulation .
whenever these conditions are satisfied , we choose @xmath177 in a controlled fashion , so that its angle @xmath135 from the subspace is larger than the minimal angle @xmath178 of , and then we run dpcp ; if the conditions are not true , we do not run dpcp and report a @xmath140 .
fig [ figure : dpcp_controlled ] shows the angle of @xmath179 from the subspace .
we see that whenever is true , dpcp returns a normal after only @xmath175 iterations .
fig [ figure : dpcp_random ] shows that if we initialize @xmath177 randomly , then its angle @xmath135 from the subspace becomes less than the minimal angle @xmath178 , as @xmath4 increases .
even so , fig .
[ figure : dpcp_randomsuccess ] shows that dpcp still yields a numerical normal , except for the regime where both @xmath4 and @xmath180 are very high . notice that this is roughly the regime where we have no theoretical guarantees in fig .
[ figure : dpcp_theory ] . fig .
[ figure : dpcp_smart ] shows that if we initialize @xmath177 as the right singular vector of @xmath181 corresponding to the smallest singular value , then @xmath182 is true for most cases , and the corresponding performance of dpcp in fig .
[ figure : dpcp_smartsuccess ] improves further . finally , fig .
[ figure : dpcp_angle ] plots @xmath178 .
we see that for very low @xmath4 this angle is almost zero , i.e. dpcp does not depend on the initialization , even for large @xmath180 . as @xmath4 increases though , so does @xmath178 , and in the extreme case of the upper rightmost regime , where @xmath4 and @xmath180 are very high , @xmath178 is close to @xmath183 , indicating that dpcp will succeed only if @xmath177 is very close to @xmath14 .
next , for the same range of @xmath180 and @xmath4 , and still in the absence of noise , we examine the potential of each of se , l21 , svs , ransac and dpcp to perfectly distinguish outliers from inliers .
note that each of these methods returns a _ signal _ @xmath184 , which can be thresholded for the purpose of declaring outliers and inliers . for se
, @xmath185 is the @xmath0-norm of the columns of the coefficient matrix @xmath65 , while for l21 it is the @xmath51-norm of the columns of @xmath62 . since ransac ,
svs and dpcp directly return subspace models , for these methods @xmath185 is simply the distances of all points to the estimated subspace model . in fig .
[ figure : separation ] we depict success versus failure , where success is interpreted as the existence of a threshold on @xmath185 that perfectly separates outliers and inliers .
as expected , the low - rank methods se and l21 can not cope with large dimensions even in the presence of @xmath186 outliers . as expected , ransac is very successful irrespectively of dimension , when @xmath180 is small , since the probability of sampling outlier - free subsets is high .
but as soon as @xmath180 increases , its performance drops dramatically . moving on , svs is the worst performing method , which we attribute to its approximate nature .
remarkably , dpcp performs perfectly irrespectively of dimension for up to @xmath187 outliers .
note that we use the true codimension @xmath159 of the subspace as input to dpcp ; this is to ascertain the true limits of the method . certainly , in practice only an estimate for @xmath159 can be used . as we have observed from experiments ,
the performance of dpcp typically does not change much if the codimension is underestimated ; however performance can deteriorate significantly if the true @xmath159 is overestimated .
moreover , we note that while se , l21 and svs are extremely fast , as they rely on admm implementations , dpcp is much slower , even if we use an optimizer such as gurobi @xcite .
speeding up dpcp is the subject of current research . finally , in fig .
[ figure : rocsynthetic ] we show roc curves associated with the thresholding of @xmath185 for varying levels of noise and outliers .
when @xmath4 is small , fig .
[ figure : roc_synthetic_ld ] shows that se , l21 and dpcp are equally robust giving perfect separation between outliers and inliers , while svs and ransac perform poorly .
interestingly , for large @xmath4 ( fig .
[ figure : roc_synthetic_ld ] ) , dpcp gives considerably less false positives ( fp ) than all other methods across all cases , indicating once again its unique property of being able to handle large subspace dimensions in the presence of many outliers . in this subsection
we consider an outlier detection scenario in pca using real images .
the inliers are taken to be all @xmath188 face images of a single individual from the extended yale b dataset @xcite , while the @xmath9 outliers are randomly chosen from caltech101 @xcite .
all images are cropped to size @xmath189 as was done in @xcite . for a fair comparison ,
we run se on the raw @xmath190-dimensional data , while all other methods use projected data onto dimension @xmath191 . since it is known that face images of a single individual under different lighting conditions lie close to an approximately @xmath192-dimensional subspace @xcite , we choose the codimension parameter of dpca to be @xmath193 . we perform @xmath175 independent trials for each individual across all @xmath194 individuals for a different number of outliers @xmath195 and report the ensemble roc curves in fig .
[ figure : rocfaces ] . as is evident , dpca is the most robust among all methods .
we presented _ dual principal component pursuit ( dpcp ) _ , a novel @xmath0 outlier detection method , which is based on solving an @xmath0 problem on the sphere by linear programs over a sequence of tangent spaces on the sphere .
dpcp is able to handle subspaces of as low codimension as @xmath196 in the presence of as many outliers as @xmath187 .
future research will be concerned with speeding up the method as well as extending it to multiple subspaces and other types of data corruptions , such as missing entries and entry - wise errors .
this work was supported by grant nsf 1447822 . | we consider the problem of outlier rejection in single subspace learning .
classical approaches work directly with a low - dimensional representation of the subspace .
our approach works with a dual representation of the subspace and hence aims to find its orthogonal complement .
we pose this problem as an @xmath0-minimization problem on the sphere and show that , under certain conditions on the distribution of the data , any global minimizer of this non - convex problem gives a vector orthogonal to the subspace .
moreover , we show that such a vector can still be found by relaxing the non - convex problem with a sequence of linear programs
. experiments on synthetic and real data show that the proposed approach , which we call dual principal component pursuit ( dpcp ) , outperforms state - of - the art methods , especially in the case of high - dimensional subspaces . |
explicit time evolution of quantum many - particle systems out of equilibrium has generated intense interest due to unprecedented possibilities for experimentally accessible non - equilibrium situations , opened up by developments in laser - cooled atomic clouds and in mesoscopic systems .
one theme has been the response to an instantaneous ` quench ' , where a physical parameter is suddenly changed to a different value . while instantaneous quenches are more convenient to analyze , a change of parameter can of course be performed at any rate . at the other extreme from the instantaneous quench
, one can make the parameter sweep _ adiabatically _ , in which case the system reaches the ground state of the final hamiltonian .
the kibble - zurek theory describes finite - rate ramps , neither instantaneous nor adiabatic , across phase transitions .
@xcite while the original interest concerned thermal phase transitions , there has been a recent surge of interest in finite - rate traversals of _ quantum _ phase transitions in lattice hamiltonians ( refs . and
references therein ) .
one important context is the bose - hubbard hamiltonian , where finite - rate ramps of the hubbard interaction @xmath0 have been analyzed , @xcite motivated by an influential experiment quenching across the mott - superfluid transition .
@xcite in this work , we analyze finite - rate ramps in a bose - hubbard dimer . for far - from - equilibrium issues where few standard theoretical techniques exist , it is of obvious interest to look at finite clusters because of near - exact solvability . one can hope to understand the non - equilibrium dynamics in some detail .
such detailed results for clusters clearly provide an invaluable background for the emerging field of non - equilibrium dynamics in macroscopic ( many - site ) quantum systems . in addition
, the bosonic dimer is an important model system by itself , and its physics is relevant in diverse contexts . in recent years
several experiments have achieved two - site bosonic systems in cold - atom setups .
@xcite the quenches we study could be implemented in such a setup , _
e.g. _ , through slower ramps in the experiment of ref . .
the bose - hubbard dimer can be mapped onto a single - spin hamiltonian , @xcite very similar to that governing single - molecule magnetic experiments , where finite - rate ramps have received some attention . @xcite
finite - rate quenches have also been reported with a josephson junction arrangement , @xcite which is closely related to a bosonic dimer . for @xmath1 bosons in two sites , the hamiltonian is @xmath2 with @xmath3 and @xmath4 .
quenches of the interaction @xmath0 in the dimer have been considered in refs . .
however , for the dimer , quenches of the site bias parameter @xmath5 are also of natural interest .
instantaneous quenches of @xmath5 from an imbalanced to a symmetric situation , _
i.e. _ , @xmath6 , have been studied as a quantum realization of _ self - trapping_. @xcite classical self - trapping , described by the two - mode gross - pitaevskii equation , is the persistence of imbalance in the state despite the final hamiltonian being unbiased . @xcite in the quantum case , the relative number imbalance @xmath7 oscillates after the instantaneous quench , with decay and long - time revivals of the oscillations . for small @xmath8 ,
the imbalance oscillations are around zero . however , for large enough values of @xmath8 and @xmath9 , after a fast quench the mean value of the oscillating @xmath10 does not relax to zero within any reasonable time scale .
@xcite the astronomically long relaxation times ( `` quantum self - trapping '' ) can be understood in terms of extremely small energy splittings in the eigenvalue spectrum of the final hamiltonian . for the instantaneous quench ,
the various relationships between timescales of dynamical features and energy scales of the @xmath11 spectrum has been discussed in some detail in the literature .
@xcite ( color online . )
( a ) and ( b ) form of quench analyzed in this work .
lower panels illustrate the approach to adiabaticity with increasing @xmath12 , for ( @xmath1,@xmath13,@xmath8)=(4,0.2,9 ) .
( c ) overlap with ground state of final hamiltonian .
curves from lowest to highest final values correspond to @xmath14 10 , 40 , 80 , 150 .
( d ) evolution of energy : curves from top to bottom are @xmath14 0.01 , 10 , 40 , 80 .
dotted horizontal line is the ground state energy of the final hamiltonian .
the final excess energy over this dotted line is the residual energy , which decreases with @xmath12 ( e ) . ]
we use @xmath15 , measuring energy [ time ] in units of @xmath0 [ @xmath16 .
we focus mostly on the regime @xmath17 .
we will analyze finite - rate quenches of @xmath5 , of the form @xmath18 as illustrated in figs .
[ fig_energyevolution]a , [ fig_energyevolution]b .
we consider the whole range from @xmath19 ( instantaneous ) to @xmath20 ( adiabatic ) .
the initial ( @xmath21 ) state is taken to be the ground state of the initial hamiltonian with @xmath22 .
the initial asymmetry @xmath8 is taken to be large enough that the bosons are initially concentrated almost entirely on site 1 , _
i.e. _ , the wavefunction is dominated by @xmath23 .
the final ground state is dominated by @xmath24 but the system does not reach this ground state unless the quench is truly adiabatic . fig .
[ fig_energyevolution]c demonstrates how larger-@xmath12 quenches are more nearly adiabatic , through the temporal evolution of overlaps with the @xmath11 ground state . in kibble - zurek theory , one uses the _ defect density _ in the final ordered state to quantify the deviation from adiabaticity . in few - site clusters ,
defects are not easily defined , nor are phase transitions or ordering .
however , there is a natural quantity that serves an analogous role , namely , the final energy after the quench , @xmath25 . in an adiabatic sweep ,
the final energy is the ground state energy @xmath26 of the final hamiltonian .
the excess energy over @xmath26 ( _ residual energy _ ) measures the deviation from adiabaticity .
@xcite the interpolation between instantaneous and adiabatic limits is illustrated through energy evolution in fig .
[ fig_energyevolution]d , and through residual energies in fig .
[ fig_energyevolution]e .
our main results concern the dependence of the residual energy , @xmath27 , on the quench time @xmath12 . for near - instantaneous quenches ( small @xmath12 )
, the energy deviation from the instantaneous limit is found to scale as @xmath28 .
we analyze larger-@xmath12 quenches through a multi - crossing landau - zener @xcite scenario , and derive concise expressions for @xmath29 in an intermediate-@xmath12 regime as well as in the near - adiabatic ( very large @xmath12 ) limit .
while we focus on @xmath29 , our analysis can in principle be adapted to treat other observables .
since instantaneous @xmath5-quenches are associated with the self - trapping phenomenon , we also ask whether and how self - trapping survives when the ramp rate is finite .
our analysis for large @xmath12 relies on the avoided level crossing structure of the problem .
[ fig_crossings_excessenergy]a - c shows that the level crossing structure is only relevant for small @xmath13 .
our analytic treatment for slow quenches is focused on this parameter region , @xmath30 ( fig .
[ fig_crossings_excessenergy]a ) .
therefore , we can regard the quench problem as one of traversing a sequence of well - separated avoided level crossings
. this scenario would clearly not be applicable for situations such as those in figs .
[ fig_crossings_excessenergy]b , [ fig_crossings_excessenergy]c .
( color online . ) ( a - c ) energy spectra for @xmath31 bosons . for small @xmath13 and even @xmath1
, the lowest state goes through @xmath32 avoided level crossings on either side of @xmath11 , at @xmath33 , with @xmath34 .
( d ) @xmath35 bosons .
( e ) residual energy ; dots are exact values and line is theory . ] we start with the simplest case of two bosons .
there is only one level crossing encountered during the quench , at @xmath36 , where the @xmath37 and @xmath38 states are mixed ( fig . [ fig_crossings_excessenergy]d ) .
the hamiltonian within this space is @xmath39 . the landau - zener formula @xcite gives the probability of excitation at this level crossing to be @xmath40 , where @xmath41 .
at the end of the quench , the energy has a contribution of weight @xmath42 from the ground state ( energy @xmath43 ) and a contribution of weight @xmath44 from the two higher levels ( energy @xmath45 ) .
thus the residual energy is @xmath46 .
this reproduces numerical calculations for large @xmath12 ( fig .
[ fig_crossings_excessenergy]e ) .
for larger numbers of bosons , the quantum state can take various paths to @xmath11 , in a multi - crossing situation such as that shown in fig .
[ fig_crossings_excessenergy]a .
fortunately , at small @xmath13 the higher crossings are simple to treat because the energy splitting at these points are of higher than linear order in @xmath13 .
thus for @xmath47 one can regard these as real crossings rather than avoided crossings , so that the excitation probabilities are unity .
one therefore has to consider only the excitation probabilities at the @xmath32 crossings involving the lowest energy state .
the excitation probabilities are @xmath48\ ] ] at the @xmath49-th crossing encountered during the quench ; here @xmath50 is the bosonic factor relevant for the coupling between states @xmath51 and @xmath52 .
any weight going into the upper level at the ( @xmath53)th crossing goes straight on to the final energy @xmath54 , since we neglect further deflection at the higher crossings .
the final ( @xmath11 ) energies are @xmath55 , with @xmath56 . from this picture ,
the final energy is found to be @xmath57 for moderate values of @xmath1 for which exact numerical evolution is feasible up to large times , we find this relationship to work very well at small @xmath13 ( _ e.g. _ , fig .
[ fig_alltausmalltau]a ) .
there are two situations , corresponding to distinct physical pictures of the excitation process , where eq .
reduces to compact forms .
first , when @xmath12 is large enough that one can use the landau - zener formula , but small enough that the @xmath58 are close to unity , the excitation at the first two crossings deplete the weight , since @xmath59 .
thus only the top two final levels ( @xmath60 and @xmath61 ) contribute to the final energy . expanding @xmath62 to linear order ,
one gets for this intermediate regime @xmath63 here @xmath64 is the residual energy for the instantaneous case ( @xmath19 ) , namely @xmath65 .
we have used @xmath66 .
( in an instantaneous quench , the weight would all go to the highest final level . )
one could also extend the above analysis by including the third crossing and expanding up to @xmath28 .
obviously , including further levels and expanding the exponentials up to higher orders in @xmath12 , one eventually gets back the full expression .
note that , although we have obtained a linear intermediate behavior in @xmath12 , this is a nonperturpative result since it is is based on landau - zener probabilities .
second , at very large @xmath12 the @xmath67 are small , so that @xmath68 , because of the bosonic factors @xmath50 . neglecting @xmath69 ,
one obtains @xmath70 only the lowest and highest energy levels contribute in this case .
( color online . ) ( a ) ( @xmath1,@xmath13,@xmath8 ) = ( 10,0.1,300 ) .
thick line : exact numerical values .
the three dashed lines are the analytical results for small , intermediate and large @xmath12 ( eqs .
[ eq_smalltau2 ] , [ eq_intermediate1 ] , [ eq_largetau2 ] ) .
[ eq_largetau1 ] is dotted line interpolating between [ eq_intermediate1 ] , [ eq_largetau2 ] , but is barely visible because the exact curve almost coincides .
( the exact curve has some additional oscillations . )
( b ) quadratic behavior , shown here for cases we do not treat analytically , large @xmath13 ( upper ) and @xmath71 ( lower ) .
( c ) coefficient @xmath72 follows eq .
[ eq_smalltau2 ] ( horizontal line ) when @xmath73 . ] fig .
[ fig_alltausmalltau]a shows an example where the behaviors of eqs .
, , can be seen in exact numerical calculations .
the small oscillations at large @xmath12 on top of eqs . , are interference effects , discussed later .
we now consider small @xmath12 , _ i.e. _ , almost instantaneous quenches .
the main observation is that the residual energy has the dependence @xmath74 the quadratic behavior is very robust , and is present for all values of @xmath1 , @xmath8 , @xmath13 we have checked ( fig . [ fig_alltausmalltau]a , b ) . in fig .
[ fig_alltausmalltau]a , we can see the quadratic behavior ( straight line in log - log plot ) for a @xmath75 case , which is the parameter region we analyze .
[ fig_alltausmalltau]b shows quadratic behaviors for cases outside this parameter regime . to explain this behavior , we start with @xmath35 bosons . writing the wavefunction as @xmath76
, one can consider equations of motion for @xmath77 , _
e.g. _ , @xmath78 , and solve for small orders in time : @xmath79 since @xmath80 is entirely imaginary , the linear terms in @xmath81 are imaginary .
linear-@xmath81 terms will thus cancel out from observable quantities like @xmath82 and @xmath83 , which appear in the expression for the final energy : @xmath84 .
the energy at @xmath12 thus has a constant and a @xmath28 term , but no @xmath85 term . for @xmath35
, we can express the coefficient @xmath72 in eq .
analytically in terms of the initial @xmath86 , and also write analytic expressions for @xmath87 , as solutions of cubic polynomials . unfortunately , these expressions are too cumbersome to be useful . for @xmath88 ,
one can calculate perturbatively , yielding @xmath89 . for @xmath90
, the same argument holds for a leading @xmath28 correction . for very large @xmath8 and small enough @xmath12 , one can use the approximation that only the most imbalanced configurations are excited , and restrict to the subspace @xmath23 , @xmath91 , @xmath92 .
the calculation is then similar to the @xmath35 case .
one obtains @xmath93 at leading order in @xmath94 .
this also contains the @xmath35 result .
[ fig_alltausmalltau]a , c show that this expression works well for @xmath95 .
while this compact result is valid only for @xmath96 , the level crossing structure plays no role in this analysis , so in principle it can be applied also to other parameter regimes ; however the expressions for @xmath72 are too complicated to be useful in such cases . fig .
[ fig_alltausmalltau]a also displays the entire instantaneous - to - adiabatic crossover , which includes our slow - quench and fast - quench results but also a range of @xmath12 ( @xmath970.02 to @xmath9720 ) for which we do not have simple descriptions .
( color online . ) ( a ) quantum interference at large @xmath13 .
( b ) interferences resulting from using @xmath23 as initial state . ]
our treatment of multi - crossing configurations ( fig .
[ fig_crossings_excessenergy ] ) utilized unit excitation probabilities at the higher crossings , which is valid at small @xmath13 .
however , if there is substantial splitting at each crossing , the configurations offer rich possibilities for quantum interference of different paths .
we have already seen interference signatures in the mild oscillatory behavior in fig .
[ fig_alltausmalltau]a for ( @xmath1,@xmath13 ) = ( 10,0.1 ) in the near - adiabatic regime . in fig .
[ fig_oscil]a we show more pronounced interference effects at large @xmath13 .
interference between paths also becomes prominent when the initial state is not the ground state but has weights in more than one eigenstate , _
e.g. _ , if one starts with the state @xmath23 ( fig .
[ fig_oscil]b ) .
this is not exactly the ground state at @xmath22 as long as @xmath8 is finite .
it could however easily be an experimentally relevant initial state .
( color online . )
fate of self - trapping .
( a - d ) number imbalance dynamics during and after quench , for various quench times .
note different scale on top panel a. ( e ) long - time average of relative number imbalance after quench is over at @xmath98 .
averaging is performed between @xmath98 and @xmath99 .
( f ) quadratic deviation of @xmath100 from instantaneous case , for small @xmath12 . ]
we now consider the effect of finite quench times on the quantum self - trapping phenomenon .
@xcite self - trapping involves dynamics starting with a biased state , which is the case for an instantanous @xmath19 quench from large negative @xmath22 .
this is the type of dynamics analyzed in ref .. the relative number imbalance @xmath7 can oscillate around a nonzero value of @xmath101 if @xmath9 and the starting @xmath101 are large enough . since the state @xmath102 after a finite-@xmath12 quench is closer to the unbiased ( @xmath103 ) ground state at @xmath11 ,
self - trapping is clearly weakened at larger @xmath12 .
we can characterize self - trapping through the long - time average of @xmath7 at times @xmath104 .
a nonzero @xmath100 indicates self - trapping .
the definition is somewhat subtle because quantum self - trapping involves nonzero @xmath100 up to some very large time but not really infinite times .
( this very large timescale is due to the tiny splitting energies of the higher level crossings in fig.[fig_crossings_excessenergy]a .
@xcite ) the long - time average therefore depends on the upper time limit @xmath105 up to which @xmath10 is averaged , and will always be zero for large enough @xmath105 .
this is in contrast to the classical self - trapping observed in the two - site discrete nonlinear schrdinger equation , where the ` self - trapped ' states are truly stationary states
. @xcite in fig .
[ fig_selftrapping]a - d we demonstrate the weakening of the self - trapping effect at larger @xmath12 , through the behavior of @xmath10 .
the logarithmic scale highlights the fact that there are several imporant time scales in the dynamics .
the intermediate-@xmath12 cases have more complicated wavefunctions at the end of the quench , as analyzed in section [ sec_largetau ] .
hence the dynamics is most interesting for intermediate @xmath12 .
this can be seen for example through a larger number of peaks in the fourier transform of the intermediate-@xmath12 time evolution data . in comparison ,
the @xmath106 and @xmath107 cases show fewer high - frequency components in the dynamics .
[ fig_selftrapping]e shows the dependence of @xmath108 on @xmath12 .
the averaging displayed in fig .
[ fig_selftrapping]e , f was performed up to @xmath109 ; the @xmath10 behaviors in fig .
[ fig_selftrapping]a - d suggests this is a reasonable value for capturing the self - trapping phenomenon .
the behavior of @xmath110 is similar to the @xmath12-dependence of the residual energy @xmath29 ( fig .
[ fig_energyevolution]e ) ; howerver the ambiguity in the averaging procedure makes it difficult to perform a systematic analysis similar to what we have presented for @xmath29 in previous sections .
the similarity of fig .
[ fig_energyevolution]e and fig . [ fig_selftrapping]e is not surprising because both the @xmath111 dynamics and the final energy are determined by the state at the end of the quench , _
i.e. _ , the weight of excited states in @xmath102 . at small @xmath12
there is also a quadratic deviation of @xmath100 from the instantaneous case ( fig .
[ fig_selftrapping]f ) , as there is for @xmath29 .
this work analyzes finite - rate quenches most relevant to the self - trapping phenomenon , namely , from @xmath22 to @xmath11 with parameters @xmath75 .
while quenches of the tilt @xmath5 are natural for the dimer , it is relatively new in the non - equilibrium literature because bias quenches do not appear naturally in the many - site case .
we provide concise analytical results for the residual energy in parametrically different regimes of small , intermediate , and large @xmath12 ( eqs
. [ eq_smalltau2 ] , [ eq_intermediate1 ] , [ eq_largetau2 ] ) .
the small @xmath12 result is perturbative in @xmath12 , while eqs .
[ eq_intermediate1 ] , [ eq_largetau2 ] are obtained from the landau - zener formula and are thus non - perturbative , despite the linear dependence at intermediate @xmath12 .
the importance of bose - hubbard dimer dynamics reaches far beyond the cold - atom context , as the model appears in diverse areas of physics . for example
, it is equivalent to a large - spin hamiltonian : @xmath112 .
this describes single - molecule magnets .
however for molecular magnet experiments the anisotropy ( @xmath113 ) term usually has negative coefficient , and effects of host lattice vibrations or nuclear spins can not always be neglected .
the residual energy is in principle experimentally measurable in cold - atom realizations through time - of - flight measurements that provide energy information from the cloud expansion rate ( _ e.g. _ , ref . ) . for molecular magnet experiments energy
is difficult to measure , but magnetization dynamics ( analogous to our @xmath10 dynamics ) is commonly reported .
our work raises a number of open issues .
there are a number of parameter regimes other than ours which might be of interest .
are other values of initial and final @xmath5 , or of @xmath114 . refs .
have considered sweeps of @xmath5 from negative to positive infinity , for parameter regions and initial states quite different from ours .
clearly , we are only seeing the beginning stages of an emerging unified dynamical picture .
another intriguing issue is the connection to the mean - field description via the discrete gross - pitaevskii equation .
ref . has worked out some landau - zener issues for the mean - field dimer , for @xmath115 and @xmath116 sweeps .
numerical explorations for @xmath117 quenches have shown us behaviors similar to what we have presented for the full quantum case .
finally , the present results may need to be adapted for specific realizations , once such experiments are designed . for recent work ,
e.g. _ , k. sengupta and d. sen , phys .
a * 80 * , 032304 ( 2009 ) .
s. miyashita , h. de raedt , and b. barbara , phys .
b * 79 * , 104422 ( 2009 ) .
f. pollmann , s. mukerjee , a. g. green , j. e. moore , arxiv:0907.3206 .
d. patan _ et .
b * 80 * , 024302 ( 2009 ) .
u. divakaran and a. dutta , phys .
b * 79 * , 224408 ( 2009 ) .
s. mondal , k. sengupta , and d. sen , phys .
b * 79 * , 045128 ( 2009 ) .
l. cincio , j. dziarmaga , j. meisner , and m. m. rams , phys .
b * 79 * , 094421 ( 2009 ) .
s. mondal , d. sen , and k. sengupta , phys .
b * 78 * , 045101 ( 2008 ) .
r. schtzhold , j. low temp . phys .
* 153 * , 228 ( 2008 ) .
r. barankov and a. polkovnikov , phys .
* 101*,076801 ( 2008 ) .
d. sen , k. sengupta , and s. mondal , phys .
* 101 * , 016806 ( 2008 ) .
k. sengupta , d. sen , and s. mondal , phys . rev . lett .
* 100 * , 077204 ( 2008 ) . v. mukherjee , u. divakaran , a. dutta , and d. sen , phys .
b * 76 * , 174303 ( 2007 ) .
m. uhlmann , r. schtzhold , and u. r. fischer , phys .
* 99 * , 120407 ( 2007 ) .
a. das , k. sengupta , d. sen , and b. k. chakrabarti , phys .
b * 74 * , 144423 ( 2006 ) .
r. schtzhold , m. uhlmann , y. xu , and u. r. fischer , phys .
* 97 * , 200601 ( 2006 ) .
w. h. zurek , u. dorner , and p. zoller , phys .
lett . * 95 * , 105701 ( 2005 ) .
j. dziarmaga , phys .
lett . * 95 * , 245701 ( 2005 ) .
t. caneva , r. fazio , and g. e. santoro , phys .
b * 76 * , 144427 ( 2007 ) .
f. pellegrini , s. montangero , g. e. santoro , and r. fazio , phys .
b * 77 * , 140404(r ) ( 2008 ) .
t. caneva , r. fazio , and g. e. santoro , phys .
b * 78 * , 104426 ( 2008 ) .
a. n. salgueiro _ et .
j. d * 44 * , 537 ( 2007 ) .
g. j. milburn , j. corney , e. m. wright , and d. f. walls , phys .
a * 55 * , 4318 ( 1997 ) .
a. p. tonel , j. links , and a. foerster , j. phys .
a * 38 * , 1235 ( 2005 ) .
s. raghavan , a. smerzi , s. fantoni , and s. r. shenoy , phys .
a * 59 * , 620 ( 1999 ) .
s. raghavan , a. smerzi , and v. m. kenkre , phys .
a * 60 * , r1787 ( 1999 ) .
g. kalosakas , a. r. bishop , and v. m. kenkre , phys .
a * 68 * , 023602 ( 2003 ) .
g. kalosakas , a. r. bishop , and v. m. kenkre , j. phys .
b * 36 * , 3233 ( 2003 ) .
w. wernsdorfer and r. sessoli , science * 284 * , 133 ( 1999 ) .
w. wernsdorfer , r. sessoli , a. caneschi , d. gatteschi , and a. cornia , europhys .
lett . * 50 * , 552 ( 2000 ) .
j. johansson _ et .
_ , phys .
b * 80 * , 012507 ( 2009 ) .
a. polkovnikov , s. sachdev , and s. m. girvin , phys .
a * 66 * , 053607 ( 2002 ) .
m. jskelinen and p. meystre , phys .
a * 71 * , 043603 ( 2005 ) .
a. k. tuchman , c. orzel , a. polkovnikov , and m. a. kasevich , phys .
a * 74 * , 051601(r ) ( 2006 ) .
j. c. eilbeck , p. s. lomdahl , and a. c. scott , physica d * 16 * , 318 ( 1985 ) . v. m. kenkre and d. k. campbell , phys .
b * 34 * , r4959 ( 1986 ) .
a. smerzi , s. fantoni , s. giovanazzi , and s. r. shenoy , phys .
* 79 * , 4950 ( 1997 ) . | for a bose - hubbard dimer , we study quenches of the site energy imbalance , taking a highly asymmetric hamiltonian to a fully symmetric one . the ramp is carried out over a finite time that interpolates between the instantaneous and adiabatic limits .
we provide results for the excess energy of the final state compared to the ground state energy of the final hamiltonian , as a function of the quench rate .
this excess energy serves as the analog of the defect density that is considered in the kibble - zurek picture of ramps across phase transitions .
we also examine the fate of quantum ` self - trapping ' when the ramp is not instantaneous . |
the masses of astronomical bodies are usually determined directly , using motions of other bodies around or within the body under study .
in some cases total mass estimates found by different methods , differ by a large fraction .
it is customary to call the hypothetical matter , responsible for such mass discrepancy , dark matter ( dm ) .
the timeline of the study of dark matter is shown in table [ tab1 ] .
actually there are two dark matter problems the local dark matter close to the plane of our galaxy , and the global dark matter surrounding galaxies and clusters of galaxies .
however , this difference was understood only later , thus we show in the table the whole story .
ll year & description + 1915 & first estimates of local dm : @xcite , @xcite , @xcite + 1932 & galactic model and local dm : @xcite + 1933 & dm in coma cluster : @xcite + 1952 & galactic models and local dm : @xcite + 1957&large @xmath0 on the periphery of m31 : @xcite , @xcite + 1959 & mass of the local group : @xcite + 1961 & cluster stability conference : @xcite + 1965 & discovery of cmb : @xcite + 1972 & cluster x - ray data on mass of hot gas : @xcite + 1972 & local and global dm different , global dm non - stellar : @xcite + 1974 & parameters of dm coronas / halos : @xcite , @xcite + 1975 & dm contradicts classical cosmological paradigm : @xcite + 1977 & @xmath0 of galactic bulges low : @xcite + 1978 & extended flat rotation curves : @xcite , @xcite + 1984 & cold dm accepted : @xcite + 1989 & absence of large amounts of local dm accepted : @xcite + 2012 & cdm particle annihilation detected ?
: @xcite , @xcite + [ tab1 ]
the first indication for the possible presence of dark matter came from the dynamical study of our galaxy .
@xcite was probably first to estimate the dynamical density of matter in the galaxy in the vicinity of sun .
pik analyzed vertical motions of stars near the plane of the galaxy and calculated the dynamical density .
he also estimated the density due to all stars near the galactic plane using the luminosity function of stars .
dutch astronomer jacobus @xcite made a similar analysis .
pik and kapteyn found that the spatial density of known stars is sufficient to explain the vertical motions .
british astronomer james @xcite reanalysed vertical motions of stars near the plane of the galaxy , and found that some dark matter probably exists near the sun .
a new model of the galaxy was calculated by jan hendrik @xcite , who also determined the dynamical density of matter near the sun .
oort accepted as most probable value 0.092 solar masses per cubic parsec .
he found the density due to visible stars 0.038 solar masses per cubic parsec .
this difference is often considered as an indication for the presence of dark matter .
oort estimated the total expected mass of faint stars , which is very near to the value found from vertical motions of stars .
@xcite and his students in tartu observatory showed that the amount of dm in the galactic disk is small ; in contrast @xcite , @xcite , @xcite and some other astronomers found evidence that up to a half of matter in the solar vicinity may be dark .
more accurate recent data showed that the amount of local dark matter is small @xcite .
if there is some local dark matter , it must be dissipative to release extra kinetic energy during the contraction of matter to a flat disk .
this population probably consists of very faint stars or jupiter - like objects .
@xcite measured redshifts of galaxies in the coma cluster and found that the velocities of individual galaxies with respect to the cluster mean velocity are much larger than those expected from the estimated total mass of the cluster , calculated from masses of individual galaxies .
the only way to hold the cluster from rapid expansion is to assume that the cluster contains huge quantities of some invisible dark matter . according to his estimate
the amount of dark matter in this cluster exceeds the total mass of cluster galaxies at least tenfold , probably even more . at this time astronomers
were interested in the structure and evolution of stars , and zwicky s work seemed to be remote and uninteresting .
slowly more dynamical data on clusters of galaxies were collected , and the discrepancy between the cluster galaxy measured velocities and expected velocities for a stable cluster could not be ignored . in 1961 during the international astronomical union ( iau ) general assembly a special meeting to discuss the stability of clusters of galaxies was organised by @xcite .
however , opinions of astronomers were different and no definite conclusions were achieved . @xcite paid attention to the fact that most galaxies have positive redshifts as a result of the expansion of the universe ; only the andromeda galaxy m31 has a negative redshift of about 120 km / s , directed toward our galaxy .
this fact can be explained if both galaxies form a physical system . from the approaching velocity ,
the mutual distance , and the time since passing the perigalacticon ( taken equal to the present age of the universe ) , the authors calculated the total mass of the double system , about 5 times the sum of the conventional masses of the galaxy and m31 .
the authors suggested that the extra mass is probably in the form of hot gas of temperature about @xmath1 k. a similar problem exists in double elliptical galaxies .
the mean mass - to - luminosity ratio of double elliptical galaxies is @xmath266 @xcite , much higher than estimated masses of individual elliptical galaxies .
a certain discrepancy was detected also between masses of individual galaxies and masses of groups of galaxies @xcite .
@xcite and @xcite discovered that rotation curves of spiral galaxies are flat on the periphery of galaxies .
this is contrary to expectations since the luminosity of galaxies falls rapidly on the periphery and a keplerian decrease of the rotation curve is expected . if rotation velocities are identified with circular velocities , these observations suggested very high values of mass - to - luminosity ratios @xmath3 on the periphery of galaxies .
detailed models of galaxies using available data on all basic stellar populations ( core , disk , bulge , halo , flat population of young stars and interstellar gas ) suggested that it is impossible to reproduce the rotation curves of galaxies using independent data on @xmath0-ratios of galactic populations @xcite . in order to bring rotation data into agreement with data on known populations
it is needed to assume the presence of a new population with very large @xmath0-value , mass and radius @xcite . to find main parameters of the new population
corona the motion of satellite galaxies was studied by @xcite .
this analysis showed that the mass and effective ( harmonic ) radius of the corona exceeds the mass and radius of known populations almost tenfold .
this analysis was confirmed by @xcite ; both independent data suggested that the previously unknown population dominates the mass budget of the universe : the mean density of matter is about 0.2 of the critical cosmological density .
ostriker et al .
used the term ` halo ' to denote the massive population .
the dark matter problem was discussed in a regional conference on january 1975 in tallinn .
the main subject of the discussion was the nature of the dark matter .
two basic models were suggested for dark matter : faint stars or hot gas .
it was found that both models have serious difficulties .
the problem was discussed again at the third european astronomical meeting in tbilisi on june 1975 .
this meeting was the first well documented international discussion between supporters and opponents of the dark matter concept .
@xcite concluded that systems of galaxies are stable with conventional masses .
their most serious argument was : big bang nucleosynthesis suggests a low - density universe with the density parameter @xmath40.05 ; the smoothness of the hubble flow also favours a low - density universe .
soon new observational data arrived which supported the presence of massive halos / coronae of galaxies .
both optical @xcite , and radio @xcite data confirmed flat rotation curves of galaxies at large galactocentric distances .
@xcite found that velocity dispersions and mass - to - light ratios for elliptical galaxies and bulges of galaxies are considerably lower that expected before .
this observation demonstrates clearly that conventional stellar populations can not be identified with dark halos / coronae .
new independent observations confirmed large masses of clusters of galaxies .
already early x - ray observations of clusters of galaxies showed that the mass of the hot x - ray emitting gas is not sufficient to hold clusters together .
these data also allowed to estimate total masses of clusters , and confirmed earlier measurements on the basis of velocity dispersions of galaxies in clusters @xcite .
additional estimate of masses of clusters of galaxies came from gravitational lensing , also supporting large masses of clusters of galaxies .
by the end of the 1970 s most objections against the dark matter hypothesis had been rejected .
however , there remained three problems : * it was not clear how to explain the big bang nucleosynthesis constraint on the low density of matter , and the smoothness of the hubble flow the main argument in favour of the classical cosmological paradigm .
* if the massive halo ( corona ) is neither stellar nor gaseous , of what stuff is it made of ? * and a more general question : what is the role of dark matter in the evolution of the universe ?
answers to these problems came from completely new areas of research observations of the cosmic microwave background ( cmb ) radiation , and large - scale distribution of galaxies . according to the current understanding ,
the universe began with a big bang and was initially very hot .
it expanded rapidly and cooled , and at a certain epoch was cool enough for atoms to recombine .
the effective temperature of this radiation drops as the universe expands .
the cosmic microwave background radiation , a remnant of the initial hot universe , was detected by the american radio astronomers @xcite .
as emphasised by @xcite and @xcite structures in the universe were created by the growth of small inhomogeneities of the density . during the initial hot phase of the evolution of the universe the matter and radiation were coupled and density inhomogeneities could not grow . as the universe expanded , the gas cooled , and at recombination the gas became neutral . from this time on , density fluctuations in the gas had a chance to grow by gravitational instability .
the density fluctuations are of the same order as temperature fluctuations .
thus astronomers started to search for temperature fluctuations of the cmb radiation .
none were found , only lower upper limits for the amplitude of cmb fluctuations were obtained . on the other hand , theoretical calculations show that at the epoch of recombination the density ( and temperature ) fluctuations must have an amplitude of the order of @xmath5 .
otherwise structure can not form , since the gravitational instability works very slowly in an expanding universe .
this controversy can be solved if non - baryonic elementary particles , such as massive neutrinos , form dark matter particles .
there were several reasons to search for non - baryonic particles as a dark matter candidate .
first of all , no baryonic matter candidate fit the observational data .
second , the total amount of matter is of the order of @xmath6 in units of the critical cosmological density , while the nucleosynthesis constraints suggest that the amount of baryonic matter can not be higher than about 0.04 of the critical density .
if dark matter is non - baryonic , then this helps to explain the paradox of small temperature fluctuations of the cosmic microwave background radiation .
density perturbations of non - baryonic dark matter already start growing during the radiation - dominated era , whereas the growth of baryonic matter is damped by radiation .
the only known non - baryonic particle was neutrino , thus it is natural that first neutrinos were considered as dark matter particle candidates .
the power spectrum of neutrino - dominated dark matter is cut at small scales due to rapid motion of particles .
figure [ fig : model ] shows results of numerical simulation , based on such model . in fig .
[ fig : wedges ] the actual distribution of galaxies , groups and clusters of galaxies is shown .
the comparison shows that the neutrino - dominated model has no fine structure in the distribution of galaxies and systems of galaxies .
to avoid these difficulties dissipationless particles heavier than neutrinos were suggested by @xcite , @xcite , and @xcite . here
particles like axions , gravitinos or photinos play the role of dark matter .
these particles were called cold since free streaming of particles is unimportant and particles behave as a cold non - dissipative gas .
numerical models based on cold dark matter ( cdm ) represent the fine structure of the universe well @xcite .
the properties of the cold dark matter model were analysed in detail by @xcite .
searches for elementary particles which could serve as candidates for dark matter particles have been carried out in particle acceleration centres , so far with no definite results . thus indirect evidence for
the presence of dark matter has been explored .
one of the recently analysed datasets of interest to investigate possible effects of dark matter comes from the fermi gamma - ray space telescope , launched on june 11 , 2008 .
its large area telescope ( lat ) can detect gamma rays in an energy interval from about 20 mev to 300 gev .
recently @xcite claimed that there is strong evidence of a monochromatic gamma - ray line from the galaxy centre with an energy @xmath7 gev present in the fermi large area telescope data .
soon it was detected that actually there is a double line spectrum with peaks at energies at 111 and 129 gev in the galactic center .
the double peak - like excess can be interpreted as a signal of dark matter direct two - body annihilations into two channels with monochromatic final - state photons .
+ @xcite and @xcite analysed the gamma - ray spectrum of the galaxy center , as well as of nearby clusters of galaxies .
these authors found from stacked gamma - ray spectra of clusters a similar double - peak signal .
figure [ fig : fermi ] shows the observed 110 and 130 gev excess in fermi lat data .
the signal from galaxy clusters is boosted due to galaxy cluster subhalos .
since the signal from galaxy centre and from nearby galaxy clusters shows exactly the same double peak structure , the signal must come from the same physics .
authors conclude : _ `` the presence of double peak is a generic prediction of dark matter annihilation pattern in gauge theories , corresponding to @xmath8 and @xmath9 final states .
thus the two seemingly unrelated gamma - ray spectra , from the galactic centre and from the galaxy clusters , favour the particle physics origin of the excess over any astrophysics origin''_. if these claims are true , this could be a strong evidence that dm is of particle physics origin , representing a breakthrough both in cosmology and in particle physics .
the presence of large amounts of matter of unknown origin has given rise to speculations on the validity of the newton s law of gravity at large distances .
one such attempt is modified newtonian dynamics ( mond ) , suggested by @xcite .
mond and other similar models are able to explain a number of observational data without assuming the presence of dark matter .
however , there exist several arguments which make these models unrealistic .
the strongest argument in favour of the presence of non - baryonic dark matter comes from the cmb data . in the absence of large amounts of non - baryonic matter during the radiation dominated era of the evolution of the universe it would be impossible to get for the relative amplitude of density fluctuations a value of the order of @xmath5 , needed to form all observed structures .
the other strong argument in favour of the presence of some matter in addition to ordinary baryonic matter comes also from cmb data .
the wavenumber of the first acoustic peak in the cmb spectrum is a very accurate indicator of the total matter / energy density of the universe .
experiments show that with great accuracy the total density is equal to the critical cosmological density . on the other hand ,
both direct determinations as well as the nucleosynthesis constraints show that the density of baryonic matter is only about 4 % of the critical density .
in other words , there must exist some other forms of matter / energy than ordinary matter .
the other forms are dark matter and dark energy .
dark energy causes the acceleration of the universe , it was detected by comparison of nearby and distant supernovae by @xcite and @xcite .
there exist direct observations of the distribution of mass , visible galaxies and the hot x - ray gas , which can not be explained in the mond framework .
one of such examples is the `` bullet '' cluster 1e 0657 - 558 @xcite .
this is a pair of galaxy clusters , where the smaller cluster ( bullet ) has passed the primary cluster almost tangentially to the line of sight .
weak gravitational lensing observations show that the distribution of matter is identical with the distribution of galaxies .
the hot x - ray gas has been separated by ram pressure - stripping during the passage .
this separation is only possible if the mass is in the collisionless component , i.e. in the non - baryonic dark matter halo , not in the baryonic x - ray gas . with the discovery of dark energy the buildup of the modern cosmological paradigm has reached a mature stage .
however , the story of dark matter is not over yet we still do not know of what non - baryonic particles the dark matter is made of , and the nature of dark energy is completely unknown .
the main conclusions of the study of dark matter can be formulated as follows . 1 .
the discovery of dark matter was the result of combined study of galaxies , their populations , and systems of galaxies .
dark matter story is a typical paradigm shift .
evidence for dark matter has been collected independently in many centres .
3 . there are two dark matter problems dark matter in the galactic disk , and dark matter around galaxies and clusters .
dark matter in the galactic disk is baryonic ( faint stars or jupiters ) .
the amount is small .
dark matter around galaxies is non - baryonic cold dark matter .
it constitutes about 0.25 of the critical cosmological density .
dark matter is needed to start early enough gravitational clustering to form structure .
this solves the big - bang nucleosynthesis controversy .
essential information on the nature of dark matter comes from the structure of the cosmic web .
the nature of dm particles is still unknown . | i give a review of the development of the concept of dark matter . the dark matter story passed through several stages from a minor observational puzzle to a major challenge for theory of elementary particles .
modern data suggest that dark matter is the dominant matter component in the universe , and that it consists of some unknown non - baryonic particles .
dark matter is the dominant matter component in the universe , thus properties of dark matter particles determine the structure of the cosmic web . |
given a meromorphic linear differential system on the riemann sphere , @xmath0 it is important to determine whether a singularity of @xmath1 is a _ regular _ singular point for the system ( [ sd ] ) . unlike with scalar linear differential equations , for which there is a purely algebraic condition on the orders of the poles of the coefficients due to l. fuchs @xcite , a system ( [ sd ] ) can display arbitrarily high pole orders at a regular singularity .
consider the differential system , expanded in the neighbourhood of the singular point ( assumed for simplicity to be @xmath2 ) as follows , where we put @xmath3 , @xmath4 the integer @xmath5 is traditionally known as the _ poincar rank _
@xmath6 of the system .
finding the type of singularity involves knowing the minimum value @xmath7 , sometimes known as the _ true poincar rank _ , of this rank under gauge transformations @xmath8}=p^{-1}ap - p^{-1}\theta p\textrm { with } p=\sum_{k\gsl k_0}p_kz^k\in \g(k)\textrm { where } k=\c((z)).\label{gauge}\ ] ] several lines of research have been opened to tackle this problem .
the most classical tries to iteratively construct a suitable gauge transformation @xmath9 , usually coefficient by coefficient in the series expansion . featured methods rely on the linear algebra over @xmath10 involved by equation ( [ gauge ] ) , like moser and continuators ( @xcite ) , whose methods are widely used nowadays in computer algebra , or other researchers such as @xcite , while @xcite use lie group theoretic tools .
the nature of a singularity of @xmath1 can also considered from point of view of meromorphic connections @xcite , and especially , as a question of stability of certain _ lattices _ under the differential operator induced by the connection @xcite .
we focus here specifically on the approach of _ saturating lattices _ used by grard and levelt @xcite : the true poincar rank @xmath11 is the minimum integer @xmath12 such that the sequence of @xmath12-saturated lattices ( recalled in section [ gls ] ) eventually stabilises .
recent work has shown close relations between the geometric framework of the bruhat - tits building of @xmath13 , for some discrete valued field such as @xmath14 , and tropical convexity @xcite .
we show here that grard and levelt s approach can be formulated and efficiently computed in this framework , as a projection on the tropical linear space @xmath15 attached to the valuated matroid @xmath5 corresponding to a given _ membrane _ in the bruhat - tits building .
results of keel and tevelev @xcite show that , when lattices are in a same membrane of the bruhat - tits building , they are _ homothetic _ if and only if they are projected on the same point of a tropical linear space by an explicit nearest point projection map ( @xcite , see also @xcite ) . we need to be able to tell whether lattices are actually equal .
we show that the projection map respects the valuation , and projects only _ equal _ lattices on a same point ( theorem [ kt ] ) .
we then construct the _ grard - levelt membrane _ that contains all the @xmath12-saturated lattices ( proposition [ glm ] ) , and give a tropical version of grard - levelt s lattice stabilisation criterion ( corollary [ p ] ) .
third , we show that the same formula amazingly allows for the computation of a more subtle invariant , namely the _
katz rank _ of the connection ( theorem [ k ] ) , without having to either to compute a single gauge transformation nor perform the usually required ramification of the variable .
the projection map given in @xcite has unfortunately a high complexity .
we give in section [ proj ] an efficient algorithm to compute this projection map onto the linear tropical space attached to a valuated matroid .
we generalise the algorithmic approach to tropical projection developed by ardila @xcite for ordinary matroids , to the case of _ valuated matroids _ defined by dress and collaborators
@xcite . the algorithm presented here , which computes the nearest @xmath16-projection on the tight - span of a valuated matroid , has a wider applicability than the differential computations explained in the previous parts , especially in phylogenetics @xcite .
i have benefited from very interesting and stimulating discussions with federico ardila , josephine yu , michael joswig , annette werner , felipe rincn and stphane gaubert , held at the tropical geometry workshop at the ciem in castro urdiales ( spain ) in december 2011 .
a _ meromorphic connection _ is a map @xmath17 which is @xmath10-linear and satisfies the leibniz rule @xmath18 the _ matrix @xmath19 _ is given by @xmath20 for a basis @xmath21 .
a basis change @xmath22 _ gauge - transforms _ the matrix of @xmath23 by @xmath24}=p^{-1}\omega p - p^{-1}dp.\ ] ] contracting with @xmath25 yields a differential operator @xmath26 , and system ( [ sd ] ) is the expression of @xmath27 in the basis @xmath21 . a _ lattice _
@xmath28 in @xmath29 is a free sub-@xmath30-module of rank @xmath31 , that is a module of the form @xmath32 the _ poincar rank of @xmath23 on the lattice @xmath28 _ is defined as the integer @xmath33 the _ true poincar rank _
@xmath34 characterises the nature of the singularity of ( [ sd ] ) , in the sense that @xmath2 is regular if and only if @xmath35 . for any vector @xmath36 and any derivation @xmath37 , define for @xmath38
the family @xmath39 the module @xmath40 induced over @xmath30 by @xmath41 only depends on the valuation @xmath42 of the derivation @xmath43 .
we can therefore restrict ourselves to the particular derivations @xmath44 for @xmath45 . in this case
, we put @xmath46 , and note @xmath47 and @xmath48 the corresponding objects . for @xmath49 , grard and levelt define the lattices @xmath50 note that @xmath51 the true poincar rank @xmath11 of @xmath23 is @xmath52 this means that @xmath53 if and only if @xmath54 for some ( equivalently , any ) lattice @xmath28 in @xmath29 , that is , that the poincar rank on @xmath55 is at most @xmath12 . stated otherwise , finding the true poincar rank is finding the largest lattice whose poincar rank is bounded by its index in the following sequence @xmath56 let us extend this notation to multi - indices .
let @xmath57 , and let @xmath58 be an integer multi - index of length @xmath59 and _ weight _ @xmath60 .
let us define also the partial multi - indices @xmath61 and @xmath62 let by convention @xmath63 and @xmath64 for the empty sequence @xmath65 .
let finally @xmath66 be the @xmath30-module spanned by the sequence @xmath67 [ chder ] for any @xmath58 , one has @xmath68.\ ] ] the proof goes by induction on the length of the multi - index @xmath69 .
let @xmath70 .
the claim obviously holds for a multi - index of length 0 , with @xmath71 , so assume that there exists @xmath72 $ ] such that @xmath73 for @xmath74 .
let @xmath75 . then by definition , we have @xmath76 indeed , @xmath77 commutes with @xmath78 since it has by assumption constant coefficients .
the result follows , since we have then @xmath79 .
let @xmath28 be a lattice in @xmath29 .
for any @xmath80 and @xmath81 , the @xmath30-module @xmath66 is spanned over @xmath30 by the family @xmath82 according to lemma [ chder ] , the family @xmath83 is related to @xmath84 by the matrix @xmath85 where @xmath86 and @xmath1 is an upper triangular integer matrix with diagonal entries equal to @xmath87 , therefore @xmath88 . the families @xmath89 and @xmath90 are related by the matrix @xmath91 since @xmath1 is upper triangular , and the partial sums @xmath92 are non - negative , the matrix @xmath93 is in @xmath94 , and therefore both families span the same @xmath30-module .
let @xmath95 be lines in @xmath29 such that @xmath96 , and consider the subset of @xmath97 defined by @xmath98=\{\ell_1+\cdots+\ell_m \tq \ell_i\textrm { is a lattice in } d_i\}.\ ] ] following keel and tevelev , who call in @xcite the set induced by @xmath99 $ ] modulo homothety a membrane , we call this the _ affine membrane spanned by @xmath100_. for a choice @xmath101 of non - zero vectors in the lines @xmath102 , any lattice in the membrane defined by @xmath95 can be represented ( non uniquely ) by an integer valued point as follows : a lattice point @xmath103 corresponds to the lattice @xmath104 membranes spanned by @xmath105 lines in the bruhat - tits building have a faithful representation as tropical linear spaces in @xmath105-dimensional space .
let @xmath106 be the tropical semialgebra , where the operations are @xmath107 an affine membrane @xmath100 and a basis @xmath21 of @xmath29 determine a _ valuated matroid _
@xmath108^n } { \tt}{\omega}{v(\det_{(e ) } m_{\omega})}\ ] ] where @xmath109 is the subfamily of vectors of @xmath100 indexed by @xmath110 . to a valuated matroid @xmath5 of rank @xmath31 over @xmath111 $ ]
there is a tropical linear space @xmath112 attached as follows @xmath113}\choose{n+1}},\:\min_{1\lsl i\lsl n+1}p(\tau\minus{\tau_i})+x_{\tau_i}\textrm { is attained twice}\right\}.\label{lp}\ ] ] depending on the authors , @xmath15 is said to be a _ tropical convex cone _ ( @xcite ) or a _ convex polytope _
( @xcite ) in @xmath114 .
both definitions mean that @xmath115 according to @xcite , for @xmath116 , the formula @xmath117 defines the _ nearest point projection map _ @xmath118 .
there are at least two other known ways to characterise or compute @xmath119 .
[ [ blue - rule . ] ] blue rule .
+ + + + + + + + + + adapting @xcite , the authors of @xcite show that @xmath120 with @xmath121}\choose{n-1}}}\max_{j\neq \sg}(p(\sg{\cup \{i\}})-p(\sg{\cup \{j\}})+x_j).\ ] ] [ [ red - rule . ] ] red rule .
+ + + + + + + + + similarly , starting with @xmath122 , for every @xmath123}\choose{n+1}}$ ] such that @xmath124 is only attained once , say at @xmath125 , compute @xmath126 where @xmath127 is the second smallest number in that collection , and put @xmath128 .
then @xmath129 .
[ kt ] the nearest point projection map @xmath118 induces a bijection @xmath130 between @xmath99 $ ] and the lattice points in @xmath15 @xmath131 in particular , @xmath132 and @xmath133 span the same lattice @xmath28 if and only if @xmath134 . let @xmath135 , and let @xmath136 . according to @xcite ,
th . 4.17 ( see also @xcite th .
18 ) , there exists @xmath137 such that @xmath138 for all @xmath139 . by definition , @xmath140 .
accordingly , we have @xmath141 , and thus @xmath142 . by formula ( [ projdef ] ) , we get @xmath143 and thus @xmath144 by construction , if @xmath145 , then @xmath146 , for @xmath139 , hence @xmath147 . since in particular @xmath148 holds , we get @xmath149 .
conversely , we have @xmath150 . therefore , if @xmath151 then @xmath152 .
the converse follows directly from ( [ c ] ) .
[ glm ] fix a basis @xmath21 of @xmath28 , and @xmath153 .
let @xmath154 $ ] be the membrane spanned by the vectors @xmath155 . then @xmath156 $ ] for all @xmath157 and @xmath158 . for the considered basis @xmath21 ,
the lattice @xmath159 satisfies @xmath160 reordering terms as @xmath161 , formula ( [ rep ] ) and lemma [ rgo ] imply that @xmath162 can be represented in the membrane @xmath163 $ ] by the lattice point @xmath164 since by definition , @xmath165 holds for any @xmath166 , the module @xmath162 can also be represented as an element of the membrane @xmath154 $ ] by @xmath167 the lattices @xmath168 for @xmath169 can therefore all be seen as elements of the same membrane @xmath170 $ ] .
@xmath171 $ ] is called the _ grard - levelt membrane _ attached to @xmath28 . for any basis @xmath21 , the lattice @xmath168 is represented by the lattice point @xmath172 if @xmath173 for an basis @xmath21 of @xmath28 , then @xmath174 is described in @xmath21 by the @xmath175 matrix @xmath176 the tropical projection @xmath177 onto the tropical linear space @xmath178 attached to the grard - levelt membrane @xmath174 maps a point @xmath179 to a _
unique _ representative .
checking if @xmath53 requires to know if the lattice points @xmath180 and @xmath181 represent the same lattice , that is @xmath182 [ p ] for any @xmath28 , we have @xmath183 .
the tropical setting is compatible with the ramification of the variable .
this implies the following result .
[ k ] let @xmath184 be the tropical nearest point projection map of the grard - levelt membrane @xmath174 of any lattice @xmath28 onto its attached tropical linear space @xmath162 .
then the katz rank @xmath185 of the connection @xmath23 satisfies @xmath186 the katz rank is the minimum poincar rank of the connection @xmath187 induced on the pure algebraic extension @xmath188/(t^n - z)$ ] of @xmath189 with @xmath190 ( see _ e.g. _ @xcite ) . if we put @xmath191 for the class of @xmath192 , then @xmath193 .
thus if @xmath194 satisfies @xmath195 the system satisfied by @xmath196 is @xmath197 put @xmath198 .
the sequence @xmath199 defined by relation ( [ iter ] ) of iterated @xmath200-derivatives of @xmath201 satisfies @xmath202 let @xmath203 be the valuated matroid defined by @xmath204 , for any @xmath31-subset @xmath110 of indices of the columns of @xmath205 with respect to the @xmath191-adic valuation @xmath206 .
by construction we have @xmath207 the lattice @xmath208 has tropical representation in @xmath209 as the projection of the point @xmath210 with respect to the matroid @xmath211 . by corollary [ p ]
, we have @xmath212 . on the other hand
, @xmath213 holds .
therefore , we get @xmath214 this formula holds for _ any _ extension @xmath215 of degree divisible by the denominator @xmath216 of @xmath185 , hence the result also holds in the limit , yielding the claimed result .
let @xmath217 with @xmath218 * @xmath219 * @xmath220 . @xmath221
the blue and red rules from @xcite recalled in section [ ltrop ] have unfortunately a high computational complexity , since it involves loops over cardinality @xmath222 sets . in our case , it is especially impractical since for the grard - levelt membrane , we have @xmath223 . in this section
, we present an efficient algorithm , inspired by ardila s work on ordinary matroids @xcite , to compute the projection of a point @xmath224 onto the tropical linear space @xmath15 attached to a valuated matroid @xmath5 .
let us recall the setup of valuated matroids , and fix the notations that we will use .
for the results listed in this section , we refer to @xcite , although their definition , following @xcite , comes with the opposite sign . let @xmath225 be a finite set , and a map @xmath226 .
let @xmath227 .
the pair @xmath228 is a _ valuated matroid _
if @xmath229 and for @xmath230 and @xmath231 there exists @xmath232 such that @xmath233 a subset @xmath234 is called a basis of @xmath5 . in particular
, @xmath235 is the set of bases of an ordinary matroid @xmath9 on @xmath225 , that we call the _ matroid underlying @xmath5_. any vector of the form @xmath236 for some basis @xmath237 and @xmath238 is a circuit of @xmath5 . if @xmath239 is a circuit of @xmath5 , its _ support _
@xmath240 is a circuit of the matroid @xmath9 .
more precisely , it is the fundamental circuit of @xmath237 and @xmath241 , that is the unique circuit of @xmath9 included in @xmath242 .
similarly , any vector of the form @xmath243 for some basis @xmath237 and @xmath244 is thus a cocircuit of @xmath5 .
some important features of circuits and cocircuits of @xmath5 are in fact encoded in the underlying matroid @xmath9 . for any circuit @xmath245 of @xmath9 ,
the set of circuits of @xmath5 that have @xmath245 as support is of the form @xmath246 conversely , for any circuit @xmath239 of @xmath5 , @xmath247 for @xmath137 is a circuit of @xmath5 .
the same result applies to cocircuits .
recall the following result .
let @xmath245 be a circuit of @xmath9 . by definition , for any @xmath250 , the set @xmath251 is contained in some basis @xmath237
. therefore @xmath252 holds .
but there is a unique circuit satisfying this condition . since the cocircuits are the circuits of the dual matroid , the same result holds .
in what follows , we will speak by abuse of notation of the fundamental ( co-)circuit of @xmath237 and @xmath241 for a valuated matroid @xmath5 .
this is harmless as long as the results that we state are invariant up to the addition of a constant .
if we need to specify a representative , we will often use the only one with non - negative coordinates and with minimum coordinate equal to 0 , or with some fixed value at some element of @xmath225 . by the definition of a circuit of @xmath5 , circuits of @xmath259 have
coordinates @xmath262 hence , @xmath263 .
similarly , we have @xmath264 hence , @xmath265 . by the projectivity property of circuits and cocircuits ,
the result is established . since the sets of bases for @xmath5 and @xmath259 coincide , these are indeed the only ( co)circuits of @xmath259 . a valuated matroid @xmath266 of rank @xmath31 over a finite set @xmath267 $ ] induces a tropical linear space @xmath15 defined by ( [ lp ] ) .
this subspace of @xmath114 corresponds ( up to sign ) to what dress and terhalle call the _ tight span _ of a valuated matroid . in this section
, we present an efficient algorithmic method to compute the tropical projection from @xmath268 onto @xmath15 , that generalises results obtained by ardila for ordinary matroids in @xcite .
@xmath270 : assume that @xmath237 is a minimal basis containing @xmath179 .
let @xmath271 be the fundamental cocircuit of @xmath237 and @xmath179 . by definition
, we have @xmath272 that is , @xmath179 is minimal in the cocircuit of @xmath237 and @xmath179 . @xmath273
: suppose that @xmath179 in the unique minimum for @xmath5 on a circuit @xmath245 .
assume that @xmath274 is a cocircuit of @xmath5 where @xmath179 is minimal .
by assumption , we have @xmath275 accordingly , @xmath276 has a unique minimum at @xmath179 . by orthogonality of circuits and cocircuits ( @xcite ,
th . 3.11 , p. 204 ) , the set of indices that minimise @xmath276 can not have cardinality one .
therefore , the contradiction is established .
let us finally prove @xmath277 : consider a minimum basis @xmath237 . if @xmath278 , let @xmath279 be the circuit generated by @xmath237 and @xmath179 . by assumption ,
the minimum in @xmath245 is attained at @xmath280 .
the support of @xmath245 is equal to the fundamental circuit of @xmath237 and @xmath179 for the ordinary matroid underlying @xmath5 .
therefore , @xmath281 is a basis of @xmath5 and @xmath282 putting @xmath283 we get @xmath284 .
since we assumed that @xmath237 was minimal , we get @xmath285 .
a. @xmath286.[ari ] b. every element of @xmath225 belongs at least to one @xmath287-minimal basis of @xmath5.[arii ] c. every circuit of @xmath5 contains at least 2 @xmath287-minimal elements.[ariii ] d. every element of @xmath225 is @xmath287-minimal in at least one cocircuit of @xmath5.[ariv ] note that the previous characterisation does not apply when @xmath287 has an infinite coordinate , since @xmath259 is then no longer a valuated matroid .
however , @xmath290 happens only when @xmath179 does not belong to any basis .
the computation of @xmath119 can be performed independently for every element of the vector @xmath287 . for a given @xmath269
, there is a ( unique ) normalisation of a circuit @xmath245 of @xmath5 containing @xmath179 such that @xmath291 .
if @xmath269 violates any one of the three conditions of proposition [ r ] for the matroid @xmath259 , then @xmath179 satisfies them for the modified vector @xmath292 with @xmath293 where all the circuits are normalised so that @xmath291 .
moreover , the conditions of proposition [ r ] are not satisfied at @xmath179 for @xmath294 with @xmath295 . by assumption
, @xmath179 is the unique @xmath287-minimum over some circuit @xmath296 containing @xmath179 .
the support of such a circuit @xmath245 can be defined as @xmath297 the fundamental circuit of @xmath179 and a basis @xmath298 .
the @xmath287-value at @xmath299 of the circuit @xmath245 is of the form @xmath300 so we may choose as representative of any circuit @xmath245 containing @xmath179 the only one such that @xmath291 , namely the one defined by @xmath301 .
then by assumption @xmath302 and for any circuit @xmath303 , we have @xmath304 therefore @xmath305 so @xmath179 can not be the unique @xmath306-minimum of any circuit containing @xmath179 . on the other hand , there exists a circuit @xmath245 containing @xmath179 such that @xmath307 . putting @xmath308 for any @xmath295 , then @xmath179 will be the @xmath309-unique minimum over the circuit @xmath245 .
thus @xmath306 is the smallest vector that corrects the value of @xmath287 at @xmath179 . if @xmath312 holds , then @xmath313 is @xmath287-minimal in the fundamental cocircuit @xmath314 .
therefore all conditions of proposition [ r ] apply to @xmath313 .
otherwise , let @xmath315 be the fundamental circuit of @xmath237 and @xmath313 , normalised so that @xmath316 .
we have to prove that @xmath317 by construction , @xmath318 holds . actually , it is sufficient to prove that @xmath319 for any circuit @xmath245 containing @xmath313 .
say that @xmath320 . for any circuit @xmath245 containing @xmath313
, there exists @xmath321 such that @xmath322 is a basis . since @xmath323 , the circuit @xmath324 containing @xmath313
is well defined , and @xmath325 holds . by definition , we have @xmath326 accordingly
, we get @xmath327 . 1 .
compute a minimal basis @xmath237 of @xmath259 .
this can be performed by the greedy algorithm described in @xcite .
2 . for @xmath328 ,
compute @xmath329 . to do this
, it suffices to compute the fundamental circuit of @xmath237 and @xmath313 for the underlying matroid @xmath9 .
3 . find the minimum element of @xmath330 outside @xmath313 ; note that there are at most @xmath31 non infinite elements to consider .
z99 f. ardila , subdominant matroid ultrametrics , _ annals of combinatorics _ , 8 , 2004 ,
. 379389 .
d. g. babbitt , v. s. varadarajan , formal reduction theory of meromorphic differential equations : a group theoretic view , _ pacif . j. math .
_ , * 109 * , 1983 , pp .
. w. balser , w. b. jurkat , d. a. lutz , a general theory of invariants for meromorphic differential equations , part i : formal invariants , _ funkcialaj ekvacioj _ , vol .
22 , 1979 , pp .
g. cohen , s. gaubert , j.p .
quadrat , duality and separation theorems in idempotent semimodules , _ linear alg . and appl .
379 , 2004 , pp . 395422 .
e. corel , on fuchs relation for linear differential systems , _ compositio math .
_ , * 140 * , 2004 , pp . 13671398 .
p. deligne , _ quations diffrentielles points singuliers rguliers _ , lect .
notes in math .
163 , springer - verlag , 1970 .
a. w. m. dress , w. terhalle , the tree of life and other affine buildings , _ documenta mathematica _ , extra volume icm 1998 , part iii , pp .
a. w. m. dress , w. wenzel , valuated matroid : a new look at the greedy algorithm , _ appl . math .
, 3 , 1990 , pp .
l. i. fuchs , zur theorie der linearen differentialgleichungen mit vernderlichen coeffizienten , _
_ * 66 * , 1866 , pp . 121160 .
s. gaubert , r. katz , minimal half - spaces and external representation of tropical polyhedra , _ j. alg .
combinatorics _ , vol .
33 , n@xmath3313 , 2011 , pp . 325348 .
r. grard et a. h. m. levelt , invariants mesurant lirrgularit en un point singulier dun systme dquations diffrentielles linaires , _ ann .
fourier _ , grenoble ,
* 23 * ( 1 ) , 1973 , pp . 157195 .
w. jurkat , _ meromorphe differentialgleichungen _
notes in math .
637 , springer verlag , 1978 .
n. katz , nilpotent connections and the monodromy theorem .
applications of a result of turrittin , _ publ .
ihes _ , * 39 * , 1970 , pp .
a. hilali , a. wazner , formes super - irrductibles des systmes diffrentiels linaires , _ num
_ * 50 * ( 4 ) , 1987 , pp . 429449 . m. joswig , b. sturmfels , j. yu , affine buildings and tropical geometry , _ alb . j. math . _ , 4 ,
pp 187211 .
s. keel , j. tevelev , geometry of chow quotiens of grassmannians , _ duke math .
j. _ * 134 * , no . 2 , 2006 , pp .
y. manin , moduli fuchsiani , _ ann .
sc . norm .
* 19 * ( 1 ) serie iii , 1965 , pp .
j. moser , the order of a singularity in fuchs theory , _ math .
z. _ * 72 * , 1960 , pp .
k. murota , a. tamura , on circuit valuation of matroids , _ adv .
_ , 26 , n@xmath332 , 2001 , pp . 192225 .
a. werner , a tropical view on bruhat - tits buildings and their compactifications , _ cent .
eur . j. math .
_ 9 , n@xmath3312 , 2011 , pp . 390402 . j. yu , d. s. yuster , representing tropical linear spaces by circuits , _ proceedings of fpsac _ , 2007 . | we present an unexpected application of tropical convexity to the determination of invariants for linear systems of differential equations .
we show that the classical grard - levelt lattice saturation procedure can be geometrically understood in terms of a projection on the tropical linear space attached to a subset of the local affine bruhat - tits building , that we call the _ grard - levelt membrane_. this provides a way to compute the true poincar rank , but also the katz rank of a meromorphic connection without having to perform gauge transforms nor ramifications of the variable .
we finally present an efficient algorithm to compute this tropical projection map , generalising ardila s method for the case of the bergman fan to the case of the tight - span of a valuated matroid . |
and inter - dimer hopping parameter @xmath0 .
( b ) band - structure of the model described in panel ( a ) with @xmath1 and the fatness of the bands calculated according to eq .
( [ equnfolding ] ) scaled by a factor of 0.2 for clarity .
the doted lines are the dimer bz .
( c ) same as panel ( b ) but now for ( from left to right ) @xmath2 , @xmath3 , @xmath4 , @xmath5.,scaledwidth=50.0% ] in order to recover the original band structure from a super cell calculation one needs to unfold the bands back to the first brillouin zone of the original cell .
several methods are available , which either look at the momentum dependent hamiltonian,@xcite or at the eigenfunctions.@xcite in case of some additional distortion , i.e. the super cell is not a perfect repetition of the original primitive - cell unfolding can become a rather non - trivial task .
a possible method to define unfolded bands is to assign a certain weight to each eigenstate so that the eigenstates with a weight of 1 represent the original band structure while the eigenstates with a weight of zero identify the folded bands . in this section
we first demonstrate how this can be accomplished with a rather simple procedure , after the example we present a more thorough derivation .
consider a one - dimensional tight binding model with one s orbital per site and a nearest neighbor hopping integral @xmath6 .
the solution to this problem is the wave function that is symmetric at the zone center and antisymmetric at the zone boundary : @xmath7 where @xmath8 . as a super cell
we take a doubling of the unit cell .
this could model an infinite chain of dimers with intra - dimer hopping @xmath6 and inter - dimer hopping parameter @xmath0 .
assuming equal on - site energies and constant inter - atomic distances of size @xmath9 , the matrix representing the hamiltonian is given by : @xmath10 the hamiltonian can be understood by looking at fig .
( [ figdimerexample ] ) .
there are two sites within the unit - cell , labeled @xmath11 and @xmath12 .
( the unit cell , indicated by the black box has length @xmath13 . )
starting from the left site in the unit - cell at the atom labeled @xmath11 one can hop to the right with the strength @xmath6 or to the left with strength @xmath0 .
both processes represent a hopping from site @xmath11 to site @xmath12 , and thus enter as off - diagonal elements between @xmath14 and @xmath15 . in the hamiltonian
the hopping is multiplied by a momentum and directional dependent phase , whenever the hopping connects two species in a different unit - cell . for a hopping crossing the super cell boundary
the additional phase is @xmath16 for a hopping to the right , or @xmath17 for a hopping to the left . with @xmath18 , the size of the super cell . for @xmath19 the eigenvalues and the eigen - vectors
are : @xmath20 the main band has the energy of @xmath21 ( remember that @xmath8 ) at the zone center and @xmath22 at the zone boundary .
the wave - function of the unfolded band within the super cell consists of the sum of @xmath14 and @xmath15 at the zone center and the difference at the zone boundary ( @xmath23 ) .
the second solution , i.e. the folded replica , shows the opposite behavior in energy and its wave - function at the zone center ( boundary ) is made by the difference ( sum ) of @xmath14 and @xmath15 .
both bands can be seen in panel ( b ) of fig .
( [ figdimerexample ] ) . in a true multi - component system one often uses a so - called fat band representation in order to separate various orbital contributions into the band structure at given energy and @xmath24-point .
there the weights ( i.e. @xmath24 dependent band thickness ) are determined by the squares of the eigen - vector components : @xmath25 with @xmath26 the band index , @xmath27 and @xmath28 some orbital or site index , @xmath29 the crystal momentum ( @xmath30 ) dependent eigen - vector component , and @xmath31 a sum over equivalent sites within the unit - cell .
it is clear that weights calculated using the eigen - vectors and weights as defined above results in a constant weight throughout the whole brillouin zone for both bands .
a simple change in the definition of the weighting - function allows one to include information about the relative phase in the wave - functions between different sites . if one defines the weight as : @xmath32 one basically changed the sum of the norm - squared of the eigenstate pre - factors into the norm of the fourier transform of the eigenstate pre - factors . defining the positions of the atoms such that @xmath33 and @xmath34 and applying eq .
( [ equnfolding ] ) to the wave - functions as found in the dimer model immediately shows that the band belonging to the original , unfolded system , with the wave - function @xmath35 gets the weight @xmath36 . the folded band , with the wave - function @xmath37 obtains the weight , @xmath38 . super cell . in both panels colored
_ fat bands _
are used to show the @xmath39 and @xmath40 orbital character within the first brillouin zone of the primitive unit cell .
unfolding is done with the use of eq .
( [ equnfolding ] ) , traced over the @xmath39 and @xmath40 orbital character .
the zero of energy is at the fermi energy.,scaledwidth=50.0% ] as a proof of principle , we show in fig .
( [ figdftexample ] ) the band structure of li metal calculated with tb - lmto density functional code@xcite using a @xmath41 super cell .
the intensity is calculated according to eq.([equnfolding ] ) in both super cell and primitive cell calculations .
li is a @xmath42 metal with a body centered crystal structure and a lattice constant of 3.51 .
the primitive - cell contains one li atom , leading to 4 bands per @xmath30-point of different orbital character . at @xmath43 , for example ,
the li @xmath40 derived band is at @xmath44 - 4ev .
the three fold degenerate @xmath39 bands are at about 14 ev above the fermi energy . in other regions
these bands cross and intermix .
the variable @xmath27 labels the orbital character : @xmath45 .
the index @xmath31 sums over the @xmath46 equivalent sites within the super cell . the band - index @xmath26 labels the @xmath47 different eigenstates found at each @xmath30 point within the super cell .
to highlight the complete band structure , we use the total intensity , traced over the orbital index @xmath27 : @xmath48 .
there is a complication with the definition for unfolding presented so - far . in principle one
is free to choose the phase of each basis function separately . multiplying the phase of the basis function in the dimer labeled as @xmath11 in fig .
( [ figdimerexample ] ) should not change any physical results .
however it does change the unfolded weight as calculated with eq .
( [ equnfolding ] ) .
the solution is simple , one should keep track of the phases of the basis functions and introduce a complex - conjugated of this phase in the definition of eq .
( [ equnfolding ] ) . below
we will derive the equation for the weight of an unfolded band - structure more rigorously .
the here sketched phase - problem will then come out naturally as additional form - factors in the sum of eq .
( [ equnfolding ] ) . the weight of a band can be calculated in similar fashion as one would calculate the momentum ( inter- , and intra - brillouin - zone ) depended photo - emission spectra .
neglecting polarization effects , assuming a high photon energy such that the final states can be approximated by plane waves , and neglecting the fact that the action of the electric field of the light on the electrons is given by a momentum operator one can write the spectral weight ( @xmath49 ) of a one particle eigenstate @xmath50 as : @xmath51 whereby @xmath24 labels the momentum , @xmath30 the corresponding reduced or crystal momentum , and @xmath26 the band - index .
( a momentum dependent mixture of @xmath40 , @xmath52 , @xmath53 , or @xmath54 orbital character in the case of li as shown in fig .
( [ figdftexample ] ) . ) for a super cell calculation of an undistorted structure it is straight - forward to show that the spectral weight function as defined in eq .
( [ equnfoldingfourier ] ) will be zero for bands that do not belong to the unfolded bands of the primitive unit cell .
take the primitive cell to be defined by the matrix @xmath11 , consisting of three vectors @xmath55 , @xmath56 , and @xmath57 , which span the parallel piped defining the primitive unit cell . using bloch s theorem for a periodic system we know : @xmath58 with ,
@xmath59 here @xmath60 is an integer vector ( @xmath61 ) and @xmath62 is a periodic function with the period of the primitive unit cell , @xmath63 is periodic up - to a phase : @xmath64 for a calculation done in the primitive unit cell one can use the block diagonal in @xmath30 characteristic of the hamiltonian .
solving @xmath65 for each @xmath30 vector separately yields the band - structure . within a super - cell calculation
one does not obtain the momentum of a wave - function directly . if one takes a super - cell without perturbing the hamiltonian the eigenstates are non - the - less given by bloch s theorem as given in eq .
( [ eqbloch ] ) .
the fourier transform , of such a function , which defines the spectral weight as defined in eq .
( [ equnfoldingfourier ] ) is given as : @xmath66 with @xmath30 the crystal momentum of the wave - function in the primitive unit cell .
the reduced momentum @xmath30 , is an eigenstate property according to bloch s theorem , but undetermined from the super - cell calculation .
the reduced momentum index within the super - cell has been suppressed as it is up to a super - cell reciprocal lattice vector implicitly labeled by @xmath30 .
@xmath67 labels an integral over all space and @xmath68 an integral over the original primitive unit cell , and the sum over all possible independent shifts by an integer lattice vectors ( @xmath69 ) of the primitive unit cell .
the first term of the unfolded spectral weight in the primitive unit cell ( @xmath70 ) contains a sum over all unit - cells .
this term leads to the selection rule : @xmath71 with @xmath60 an integer vector and @xmath72 a matrix representing the reciprocal lattice : @xmath73 .
this term is responsible for the fact that only those bands that belong to the principle unit cell will gain weight , i.e. it unfolds the folded band - structure .
the second term of the unfolded spectral weight is a material and brillouin zone dependent form factor . in photo - emission experiments it is
this term ( together with the here neglected polarization dependence ) that is responsible for the brillouin zone dependent spectral weight .
the unfolding of bands as described in eq .
( [ equnfoldingfourier ] ) can be simplified further by approximating the brillouin zone dependent form factor .
we will first briefly describe model calculations and then extend it to a formalism that can readily , without much effort , be implemented in most _ ab initio _ density functional programs
. for model calculations one often expresses the band - structure in terms of some form of tight - binding representation .
if we take the variable @xmath27 as the index for the local orbital and or spin character of the tight - binding basis set ( in the case of li @xmath27 is @xmath40 , @xmath52 , @xmath53 , or @xmath54 ) then one can define the spectral weight matrix for a given one - particle eigenstate ( @xmath63 ) as : @xmath74 whereby the operator @xmath75 annihilates an electron at momentum @xmath24 and of orbital ( spin ) character @xmath27 . the momentum dependent annihilation operator can be written as a sum of position dependent operators as : @xmath76 which describes the relation between the fourier transform of a one - particle wave - function and the expectation value of the momentum dependent occupation number operator .
if the orbital character of a certain band is of no importance one can write the total unfolded spectral weight of a certain band as the trace over the spectral weight matrix : @xmath77 in the case of an _ ab initio _ density - functional theory calculation one could naturally first create a tight binding model and then use eq .
( [ equnfoldingmodel ] ) in order to calculate the unfolded spectral weight .
such a scenario has recently been proposed by wei ku _ et al .
_ @xcite and successfully used on several systems.@xcite the unfolded band - structures found in this way are expected to be rather similar to the unfolded band - structure one obtains by defining the weight of a band via a fourier transform ( see eq .
( [ equnfoldingfourier ] ) ) . for large super cells with several impurity atoms , defining a tight - binding hamiltonian might become troublesome .
an alternative method that allows one to directly obtain the unfolded band - structure from an _ ab initio _ calculation therefore is useful .
one can simplify the brillouin zone dependent form factor of the unfolded spectral weight as it appears in eq .
( [ unfoldingblochwave ] ) by looking at the one - particle wave - function character within a certain ( muffin - tin ) sphere defined around each atom .
these projections are implemented in most _ ab initio _ dft codes , where these projections are used to calculate the partial character of a given eigenstate needed to plot partial density of states , or bands of a certain orbital character .
the projected wave - function @xmath78 can be written as : @xmath79 with @xmath26 a band index of the one - particle eigenstates , @xmath27 an orbital and spin index , the sum over all sites @xmath31 that contain a muffin tin sphere onto which one projects , @xmath80 the position of the sphere labeled with the index @xmath31 , @xmath81 the functions projected onto ( with orbital character @xmath27 and centered at position @xmath80 ) , and @xmath82 the projected pre - factor . similarly as done for model calculations it makes sense to define an orbital ( @xmath27 ) specific unfolded weight , starting from : @xmath83 unfolding such a projected one - particle wave - function with the use of eq .
( [ equnfoldingfourier ] ) yields : @xmath84 the integral : @xmath85 defines a brillouin zone dependent form factor , and although this is a physical quantity one can simplify the unfolding of bands substantially by neglecting this term .
if the orbital basis set is defined such that the main outer part of @xmath86 is real and positive for all @xmath27 then a powerful approximation is to set the integral in eq .
( [ eqformfactor ] ) equal to 1 . in this case
the unfolded weight becomes : @xmath87 in the case that the super cell is undistorted , i.e. a pure repetition of the system one can understand that this definition of the spectral weight indeed unfolds the band - structure with the argumentation based on bloch s theorem .
this can be done by following the argumentation as given in the beginning of this section from eq .
( [ eqbloch ] ) to eq .
( [ unfoldingblochwave ] ) .
when the super cell is not a perfect repetition of primitive basis cells then the weight of the folded bands starts to deviate from zero .
the weight in the folded bands is related to the mixing of the folded and original bands in the band - structure and therefore a measure of the distortion .
one should be slightly careful how the form - factor as presented in eq .
( [ eqformfactor ] ) is neglected .
the phase of this integral depends on the phase of the basis function @xmath86 , which can be chosen freely . in many calculations the projections
are done onto atomic like wave - functions .
these functions are real , which fixes most of the phase , but still can either be positive or negative .
the radial part of atomic like wave - functions has nodes .
the important outer part of the radial wave - function is either positive or negative , depending on the number of nodes .
the number of nodes for an atomic like wave - function is given as @xmath88 the phase of the basis function therefore is @xmath89 .
naturally different implementations might use different phases for the basis functions .
( a complex phase of @xmath90 can for example be a useful alternative choice . ) using our phase definition , including the factor @xmath89 for the basis functions , the brillouin zone dependent form factor will be approximated as : @xmath91 whereby @xmath92 is the principle quantum number of the orbital with index @xmath27 and position @xmath80 and @xmath93 is the angular momentum of the orbital with index @xmath27 and position @xmath80 .
the unfolded weight then becomes : @xmath94 the partial character @xmath82 is the projected weight of the eigenstate with band index @xmath26 , orbital and spin character @xmath27 , at site @xmath31 in the super cell for the crystal momentum @xmath30 .
the unfolded weight of this band is calculated by taking a fourier transform summed over ( pseudo ) equivalent sites ( @xmath31 ) within the super - cell .
this can either be the same atom at positions in the super cell slightly deviating from a perfect repetition of the primitive unit cell , or different atoms in the super cell in the case of substitution .
the norm - squared of this fourier transform defines the unfolded weight of a certain band and orbital character .
substituted atoms for a cluster size of 4 ( left top ) to 128 ( right bottom ) @xmath95 total atoms and an average substitution of 5% ( @xmath96).,scaledwidth=50.0% ] there is a crucial difference between a calculation for an ordered structure and a system with randomly substituted atoms . in a system with ordered impurities gaps
will open in the band - structure , whereas for a random system this may not be the case and gaps can become smeared out structures , as will be shown by several examples later . in order to simulate the situation of an infinite crystal with a random impurity distribution one would like to create very large super cells . in order to do these calculations with the use of tractable
cluster sizes it is important to average over different random distributions of impurities with equal impurity density and over different impurity densities within the cluster . lets define the chance to have a substituted atom at site @xmath31 to be equal to @xmath97 . for a cluster of @xmath95 total sites the probability to have a number of @xmath98 substituted
atoms within the cluster is given by the binomial distribution : @xmath99 in order to illustrate the impurity distribution further we show in fig .
( [ figdistributionfunction ] ) a plot of this distribution for a doping of 5% ( @xmath96 ) and a cluster size of 4 , 8 , 16 , 32 , 64 , and 128 @xmath95 total atoms .
one should realize that even for a reasonable cluster size there still is a finite chance to have no substituted atoms within the cluster . for each total number of substituted atoms in the cluster
there are still many different ways to arrange these impurities .
it is important to take this into account , since the scattering of nearest neighbor impurities is in general very different from the scattering of isolated impurities .
in general one should determine the band structure in a super cell method with as large a cell as possible in real space and average over a varying number of impurities and a varying distribution of these impurities .
each of these calculations should be weighted with the probability to find this particular kind of configuration , including both the chance to have this specific density within the cluster and this specific order within the cluster . in practice
it is not needed to calculate all possible impurity configurations ( including variation of density and distribution ) within the cluster .
a random subset will already give a reasonable approximation to the real solution , provided that the cell is large enough and a large number of random configurations are taken .
we define a configuration to be a specific realization of a super cell with a random number of substituted atoms at random positions .
the average density of substituted sites is given by @xmath97 .
here we exemplify several aspects of random impurity calculations , using a simple model system .
the model we use is well studied which allows us to compare the super cell calculations to theoretical approximations used elsewhere.@xcite we use a single band model with constant nearest neighbor hopping ( @xmath6 ) and a onsite impurity potential @xmath100 at a fraction @xmath97 of all atoms randomly placed within the solid .
the hamiltonian is given as : @xmath101 with the sum over all sites @xmath31 and nearest neighbor sites @xmath102 .
@xmath6 is the hopping - strength , resulting in a band - width of @xmath103 .
@xmath104 is the dimension of the system .
@xmath105 is a potential with the value @xmath100 at a fraction @xmath97 of randomly placed sites and zero at other sites .
@xmath97 is the density of substituted atoms .
as the hamiltonian only includes one - particle interactions it is natural to fourier transform it to momentum ( @xmath24 ) space : @xmath106 with @xmath107 and @xmath108 the position of atom @xmath31 . within this representation
the hopping part becomes diagonal in @xmath24 .
the random potential allows for scattering between different @xmath24-vectors . in the limit of large cluster sizes or for a potential averaged over many different cluster configurations
one finds that @xmath109 and that @xmath110 for @xmath111 scales as @xmath112 .
the averaged hamiltonian ( @xmath113 ) , is thus defined as @xmath114 .
one might be tempted to state that for infinitely large clusters or for a cluster of finite size , averaged over a large number of configurations the hamiltonian becomes diagonal in momentum @xmath24 and thus momentum is always a good quantum number .
care has to be taken as there is indeed only an infinitesimal small coupling to different momenta for infinitely large clusters , but there are an infinite number of different momenta possible to which one can scatter thus leading to a finite perturbation .
the one particle green function is defined as : @xmath115 the spectral function whose poles define the band - structure , is given as @xmath116 $ ] .
the off diagonal elements of @xmath117 in @xmath24 , i.e. those elements for which @xmath111 scale as a function of cluster size , in the same way as the off diagonal elements of the hamiltonian , namely as @xmath112 . for large clusters or for an average over many different random configurations
the green function becomes diagonal in momentum @xmath118 .
calculating the green function , using the average hamiltonian ( which is diagonal in momentum ) is however different from using the full hamiltonian , even if one averages the green function afterward . in the following whenever we write @xmath119
we will assume that this is the green function averaged over several configurations ( @xmath118 ) .
the random onsite energy impurity model has been extensively studied with the use of series expansions.@xcite the @xmath120-th moment ( @xmath121 ) of the momentum integrated spectral function ( density of states ) is given as : @xmath122,\ ] ] with @xmath123 the total number of sites in the cluster .
the hamiltonian is given as @xmath124 with @xmath125 and @xmath126 .
analytical expressions can immediately be written down , if one neglects contributions to the moment which arise from products of @xmath127 and @xmath128 .
the moments of @xmath127 are given as : @xmath129\right)^n \delta\mathbf{k},\ ] ] which are the moments of the original , unperturbed system .
the non - zero lowest order moments are @xmath130 , @xmath131 , and @xmath132 .
the moments of @xmath128 are : @xmath133 contributions due to products of @xmath127 and @xmath128 are important for the moments with @xmath134 . the momentum @xmath24 dependent green function can be expressed in terms of a series in @xmath128 with the use of the dyson equations . for a hamiltonian only containing one electron terms
one can write : @xmath135 for the random configuration averaged green function , @xmath136 one finds to first order in the onsite impurity potential @xmath100 : @xmath137 it is often useful to define the self energy for the averaged green function as : @xmath138 if one compares the definition of the self energy with eq .
( [ eqdyson ] ) one finds a striking similarity .
multiplying eq .
( [ eqselfenergy ] ) with @xmath139 and @xmath119 , one reproduces the form of eq .
( [ eqdyson ] ) , but now for the configuration averaged , single momentum dependent green function , instead of the original green matrix . using eq .
( [ eqgseries ] ) one finds that to first order in @xmath100 the self energy is given as @xmath140 . to linear order in the perturbing potential
one can thus replace the random impurity system by a periodic system , which has an averaged potential on each site .
this approximation is known as the virtual crystal approximation .
it can correctly describe the averaged band - structure and total density of states to first order in @xmath100 .
it can not describe changes to the one - particle wave - functions important for the band - character , the spatially varying charge density,@xcite the broadening of bands , or the possible appearance of impurity bound states , which arise in quadratic and higher order of the perturbing potential .
tight binding model as presented in eq .
( [ eqmodel ] ) , for various cluster sizes .
the left column shows calculations for one random impurity distribution with @xmath141 .
the right column shows calculations for an average of 1000 different random configurations , which only on average have an impurity density of 25% , and vary both in impurity density as well as impurity distribution within a cluster . in blue ( red )
the character of the host ( substituted ) atoms is shown .
the potential of the substituted atoms is at @xmath142 , with @xmath143 the band - width ( the potential of the host atoms is at 0 ) .
the fermi - surface plots assume a filling of on average 1 electron per atom.,scaledwidth=50.0% ] in order to fully describe the effects of random impurities it is necessary to go beyond a series expansion in the perturbing potential .
for such cases doing exact diagonalization of large clusters is a powerful method , especially with the current computational resources . within this section
we compare several calculations done for different cluster sizes .
we furthermore show that the cluster sizes can be considerably reduced if one averages over several random configurations , i.e. both varying the impurity density as well as the impurity distribution within the cluster . in the left most column of fig .
( [ figclustersize ] ) we show calculations for the band - structure of a single random configuration , with fixed concentration .
the different rows show the calculation for a different total cluster size .
the calculations are done for a 25% ( @xmath141 ) substitution and @xmath144 .
the smallest cluster possible which allows for a substitution of 25% , namely with a total of @xmath145 sites and one impurity atom , shows the opening of large gaps in the band - structure .
this is to be expected as for a @xmath146 cluster with @xmath147 substitution one has a perfectly ordered super - structure , which is very different from a random system .
the larger the cluster becomes the smoother the band - structure becomes .
the gaps become smaller and smaller and more and more distributed to random places in momentum space .
one however needs very large clusters to find convergence if one does not average over different random configurations . even for a single calculation of a @xmath148 super cell
there is still some gaping visible in the band - structure ( left column bottom row of fig .
( [ figclustersize ] ) ) .
this becomes much better if one averages over different random configurations ( i.e. averaging over both a varying number of substituted atoms within the cluster and averaging over the possible different ways to substituted these atoms ) .
the forth column shows the same band - structure calculations , but now averaged over 1000 random configurations .
as one can see these look much more smooth .
the gaps disappear when an average over different configurations is taken . even for a @xmath149 super cell calculation
the averaged band - structure looks quite reasonable , whereas the calculation of a single configuration is far from converged .
what happens is that each different random configuration has gaps at different energies and @xmath24 vectors , leading to an averaged broadening of the bands and not to a gaping of the bands .
a similar effect can be seen for the density of states .
the calculations for a single configuration are rather noisy ; the average over 1000 configurations however shows a nice continues behavior .
averaging is crucial for the calculation of the fermi - surface .
the fermi - surface is calculated by assuming a filling of 1 electron per atom in the unit cell after averaging all calculations done for different configurations .
a single configuration shows only some points on the fermi - surface , whereas an average over several configurations shows a very nice , though broadened fermi - surface .
note that in the fermi - surface plots for small super cell calculations one can see the periodicity of the super cell very clearly as features with the periodicity of the super cell brillouin zone .
the use of small super cell calculations can lead to spurious shadow bands
. model calculation as defined in eq .
( [ eqmodel ] ) , for a @xmath150 cluster , with @xmath151 and @xmath152 to @xmath153 .
the fermi - surface plots assume a filling of on average 1 electron per atom.,scaledwidth=50.0% ] next we compare the direct method of calculating the band - structure , fermi - surface and density of states as presented here with other methods like the coherent potential approximation and its non - local version.@xcite in fig .
( [ figpotentialdependence2d ] ) we present calculations as a function of the impurity potential for a @xmath154 50% ( @xmath155 ) substituted system .
the model and parameter range used are the same as presented in fig .
( 5 ) of the paper by batt and rowlands @xcite , which allows for a direct comparison between the different approximations . for a potential of @xmath142 one finds a clear fermi - surface at a filling of one electron per site .
the fermi - surface is rather broad .
these findings are basically equivalent to the results found with the coherent potential approximation and for the non - local coherent potential approximation calculated using a @xmath146 cluster .
increasing the impurity potential further will eventually lead to two separate bands separated by a gap . in principle
one could state that the system becomes insulating at half filling .
however , due to the large broadening of the bands , one still finds for a very large potential range a finite intensity at the fermi - energy .
it is in this range that one finds differences between the coherent potential approximation and their cluster extensions @xcite . for @xmath156
one finds a new fermi - surface topology .
around @xmath43 one finds an electron pocket which is related to the host band . around @xmath157 one finds a hole pocket derived from the impurity band .
it is note - worthy that the coherent potential approximation is not able to reproduce this effect .
@xcite the non - local coherent potential approximation , for large cluster sizes should naturally exactly reproduce the current calculations , under the condition that the cluster size taken into account is large enough .
for a cluster size of @xmath146 one can see that some features of the gaping are indeed correctly reproduced .
the exact potential at which the fermi - surface topology change takes place , as well as some details of the fermi - surface are however different . for accurate calculations it is important to take relatively large cluster sizes , within the non - local coherent potential approximation .
the effect of the bath in the non - local coherent potential approximation would be relatively minor for such large clusters sizes .
it is therefore relatively straight forward to extend the non - local coherent potential approximation to the current cluster calculations . for large
cluster sizes both methods are equivalent .
a powerful concept for the understanding of random impurity calculations is the self energy @xmath158 .
the self energy is a measure of how - much the band - structure of the system with impurities deviates from the system without substituted atoms .
the full green function is given in terms of the original green function ( @xmath139 ) and the self energy as : @xmath159 the real part of @xmath160 defines an energy shift of @xmath119 , with respect to @xmath139 .
the imaginary part of @xmath160 is related to an energy broadening of the green - function .
the self - energy is casual , meaning that in our representation the imaginary part is strictly negative , and goes to zero as @xmath161 for large absolute values of @xmath162 .
the real - part is related to the imaginary part by kramers - kronig .
note that for a system of impurities there are no specific relations for the self - energy at the fermi energy . as seen in fig .
( [ figclustersize ] ) and ( [ figpotentialdependence2d ] ) there is a substantial broadening of the fermi surface ( non - zero imaginary part of the self energy ) as well as a possible change of fermi surface topology ( energy and momentum dependent real and imaginary part of the self energy ) .
it is generally excepted that a system with random impurities has a broadened band - structure compared to a pure system .
the imaginary part of the self energy must be integrable and goes to zero for large and small energies . by virtue of the kramers - kronig relations
this means that one then also must have a downward energy - shift for energies lower then the main broadening and an upward energy shift for energies higher then the energy at which the broadening takes place .
the broadening of bands will thus go hand in hand with shifts in the band - structure .
this will become more clear with the use of the example given in fig .
( [ figpotentialdependence3d ] ) . in fig .
( [ figpotentialdependence3d ] ) we show the band - structure , density of states , fermi - surface ( for a filling of one electron per atom ) and real and imaginary part of the self energy at @xmath43 as well as their momentum dependence for a @xmath163 cluster as a function of the impurity potential at @xmath164 .
the real part of the self energy at @xmath43 ( sixth column ) has a constant part , independent of the energy , equal to @xmath165 , which one would expect in linear order in @xmath100 . in the forth column one
can see the imaginary part of the self , energy , which peaks around @xmath100 .
this leads by kramers - kronig to a real - part of the self - energy , which is related to a shift of the band - structure .
the shift of the band - structure due to the real - part of the self energy can be seen very clearly in the fermi - surface plots . for a small potential of
the order of @xmath166 or smaller the band - structure , density of states and fermi - surface are , aside from some broadening , not changed much compared to the original host electronic structure .
the fermi - surface of the host is shown in the third column , by the small green pockets around @xmath157 .
for @xmath166 the fermi - surface of the system with impurities overlaps with the original fermi - surface . for a larger impurity potential the
band - broadening becomes more substantial , and a corresponding shift of the band - structure can be seen . for the fermi - surface this results in apparently larger pockets around @xmath157 .
the electron count does not change between the different calculations shown in the different rows in fig .
( [ figpotentialdependence3d ] ) , the fermi - energy is defined such that there are 1000 electrons in the cluster containing @xmath167 atoms .
it appears that luttinger s theorem is violated . calculating the number of electrons in the system by looking at the fermi - surface as defined as the maximum intensity of @xmath168 at the fermi energy gives an incorrect number of electrons .
one can understand these findings by realizing that the calculations are not done for a periodic system , but averaged over many different randomly ordered periodic systems , within a super cell .
the green function is obtained by unfolded these super cell calculations back to the original brillouin zone . the conditions for luttinger
s theorem to be applicable are not fulfilled .
the change in fermi - surface can be understood in a different way . for large negative impurity potentials
the substituted atoms will build an impurity band , at a lower energy then the original host band .
this impurity band , will be doubly occupied , i.e. full . this doubly occupation of the impurity band removes @xmath97 electrons from the host band , leading to an apparent change in electron occupation . in order to test this simple picture we added the fermi - surface one would expect for an unperturbed system with @xmath169 holes , i.e. on average only 0.7 electrons per unit cell .
these are the large green fermi - surface contours shown in the third column of fig .
( [ figpotentialdependence3d ] ) .
indeed it seems that for large potentials the apparent hole doping can be understood in terms of doubly occupied bound states around the substituted atoms , which do not contribute to the luthinger s count . there is no clear border between the two regimes where for small potentials the virtual crystal approximation
is reasonably fulfilled to a regime for large potentials where the system can be understood in terms of an impurity band .
there is always some deviation from the virtual crystal approximation , however only to quadratic order in the impurity potential .
there is an important difference between model calculations and self - consistent first principle studies . within a model the potential of the host and impurity
are fixed , irrespectively of their electron count . within a self - consistent calculation
the potential depends on the local electron count .
this later effect will lead to screening , and different impurity potentials when two impurities are nearest neighbors compared to impurities infinitely far separated from each other .
the model presented before will go astray if one places a large patch of impurity atoms next to a large patch of host atoms . in this case there will be a charge donation of the host into the impurity , independent of the distance between them , thus even if the impurity atom is very far away from the host . in reality
this will cost coulomb energy , which scales with the distance and should prevent too much of a charge flow .
it is thus important to do self consistent calculations .
density functional theory captures these effects reasonably well .
the method , language and general features described in the previous sections are directly transferable to density - functional super cell calculations .
$ ] , @xmath172$],@xmath173 $ ] ) .
middle ; a super cell of ( @xmath174$],@xmath175$],@xmath176 $ ] ) .
right ; a super cell of ( @xmath174$],@xmath177$],@xmath178 $ ] ) .
bottom row shows the super cell and the total self - consistent lda energy with respect to the most stable arrangement .
middle row shows the folded and un - folded lda band - structure of pure li within the super cell .
the bands with weight zero ( should disappear in the unfolding scheme ) are plotted in orange , the unfolded bands ( with weight 1 ) in black .
top row shows the unfolded bands for the li@xmath170mg@xmath171 ordered alloy within the 3 different super cell configurations.,scaledwidth=50.0% ] mean - field approximated model calculations including several thousands of sites are possible and finite size effects therefore can be largely eliminated .
density functional theory calculations are more involved and although modern computers are very powerful it is important to try to optimize cluster sizes and shapes
. naturally mean - field theories like most implementations of density functional theory are still numerically easier to solve then true many - body calculations , from which some knowledge about optimal cluster sizes can be obtained.@xcite in fig .
( [ figdftclusterform ] ) we show three calculations for the lda band - structure of li@xmath179 ( middle panels ) and li@xmath180mg@xmath181 ( top panels ) , calculated with the use of lmto.@xcite the calculations all correspond to different ordered structures of the same chemical composition .
the difference between the three rows is the shape of the super cell . in the middle row
one can see in orange that the folded band - structure depends on the choice of the super cell used .
the unfolded bands are independent of this choice ( when no impurities are introduced ) . in the top row
one can see the unfolded bands when one li atom is replaced by mg .
if one compares the bands of li@xmath180mg@xmath181 with the folded bands of li@xmath182 then a striking feature appears .
looking for example to the lowest @xmath183 derived band at -4 to 0 ev going from @xmath43 to @xmath123 in the unfolded brillouin zone .
whenever the li@xmath179 folded bands cross the li@xmath179 main bands ( see middle rows ) a gaping of the bands at that place in @xmath184 space appears when one li atom is replaced by mg .
this is strongly related to the difference between the calculation of an ordered structure and the calculation of a randomly substituted material . for ordered superstructures gaps
appear when the folded bands cross the original bands .
the dependence of this folding on the shape of the super cell opens the possibility to not only average over different configurations , i.e. different impurity densities and different orderings of the impurities , but also to average over different cluster shapes .
the beauty of _ ab initio _ calculations over model calculations is that one for each calculation assuming a certain impurity configuration and super cell shape , not only obtains the electronic properties , but also a total energy .
this total energy can be used as a weighting factor when the different cluster sizes are averaged . on the bottom of fig .
( [ figdftclusterform ] ) we show the total energy of the 3 different super cell calculations per site , with respect to the energy of the lowest energy state .
the material studied will be grown or annealed at a certain temperature , where one can assume the system to be in thermal equilibrium .
then one could rapidly cool the system .
the distribution of the different clusters can be assumed to be governed by the bolzman statistics one would have at the annealing temperature .
annealing the limg crystal at 300k would give a distribution of 0.29 , 0.33 and 0.38 for the three different super cells as shown in fig .
( [ figdftclusterform ] ) .
if the sample would be in thermal equilibrium at 30 k the distribution function would be 0.05 , 0.23 and 0.72 . naturally the before sketched numbers are just exemplary .
a real calculation needs much larger cluster sizes and the average over more then three different configurations or super cell shapes in order to give realistic results .
mg@xmath171 calculated using a super cell of size @xmath41 ( @xmath185 ) for panel ( a ) and ( b ) or a super cell of size @xmath186 ( @xmath187 ) for panel ( c ) and ( d ) .
the calculations are averaged over different configurations , whereby in panel ( a ) and ( c ) the number of mg atoms is fixed at 4 and 32 and only the random positions varies between the different configurations , whereas in panel ( b ) and ( d ) the number of mg atoms in the super cell ( @xmath98 ) as well as their random positions varies .
the intensities are scaled according to probabilities given by the distribution function in eq.([eqdistributionfunction ] ) with @xmath185 ( @xmath187 ) and @xmath141.,scaledwidth=50.0% ] the model calculations as shown in the previous section show averages over configurations where both the possible number of substituted atoms as well as the random placement of these atoms is varied .
one would expect that for large enough cluster sizes the variation of the number of impurities becomes less important . in fig .
( [ figdftsizerho ] ) we show in panel ( a ) a calculation done for several configurations where for each configuration the number of substituted atoms is chosen to be 4 , which for a super cell of 16 sites results in a density of @xmath141 . in panel
( b ) we show a calculation for the same super cell , but now we also vary the total number of substituted atoms .
the calculation in panel ( b ) includes averaging over all configurations according to their distribution function as given in eq .
( [ eqdistributionfunction ] ) with @xmath185 and @xmath141 .
the average density is @xmath141 in both cases .
one can see that including an average over different number of substituted atoms for small cluster sizes improves the calculation .
one can compare these results to panel ( c ) and ( d ) of the same fig .
( [ figdftsizerho ] ) . here
the cluster size is increased from @xmath41 to @xmath186 . again in panel ( c ) the total number of substituted sites is kept fixed at @xmath188 for each configuration which varies the random positions of these substituted sites . in panel ( d ) also the number of substituted sites is varied between the different configurations . for larger clusters ( at these impurity levels ) it becomes less important to have the exact random distribution as given in eq .
( [ eqdistributionfunction ] ) .
one furthermore observes that for a @xmath186 cluster of a bcc structure containing @xmath187 sites the averaged band - structure already looks very reasonable . to illustrate the effect of disorder on the band structure as obtained from density functional theory , we perform dft calculation of li metal substituted with hydrogen using tb - lmto code and super cell approximation .
the size of the super cell is @xmath186 ( 128 sites due to bcc structure ) .
a random number generator is used to generate 100 configurations for any given h concentration .
the unfolding procedure is used to present the calculated band structures in the original bz .
all 100 band structures are plotted in the same plot .
( [ figdftexamplesec ] ) shows such plots for 6.25 , 12.5 , 25 and 50% of hydrogen .
the impurity effect is twofold , first , the host band structure is broadened in both energy and momentum , and , second , there is an upward shift in energy with respect to the fermi energy .
the shifts can be understood in terms of on - site energy differences between li and h states .
since the ionization energy of h is about 8.2ev higher than that of li , the h @xmath183-states are expected to have higher binding energy than that of li .
this means that the hydrogen impurity is an acceptor .
the shifts , however , are compensated somewhat by the changes in the band widths . at about 50 percent ,
the band structure of lih alloy turns over to that of pure h which is also broadened and shows the shifts in the fermi energy which is now due to li@xmath189 impurities as can be seen in the right bottom panel of fig .
( [ figdftexamplesec ] ) . h@xmath190 . on the left calculations
are shown for an ordered li h structure .
on the right calculations are shown for a disordered system .
calculations are done for a 128 site super cell with random impurity distributions , unfolded to the original brillouin zone.,scaledwidth=50.0% ] the specific case of li@xmath191h@xmath191 is rather interesting since it allows us to study the effect of order on the electronic structure .
the electronic structure of the ordered alloy is shown in the left columns of fig .
( [ figdftexamplethird ] ) .
it is calculated with the conventional bcc unit cell that corresponds to two interpenetrating simple cubic lattices .
the material is an insulator with the gap of about 1ev which separates the occupied h 1s band with the conduction band formed by the states of mainly li character .
the random alloy is , however , metallic ( see right columns of fig .
( [ figdftexamplethird ] ) ) .
we note the very large smearing effect of the band structure especially in the region dominated by lithium @xmath192 based states above the fermi energy .
there is , of course , no background nor smearing of the bands in the ordered case .
in conclusion we have discussed a band structure based procedure which can be successfully used to describe the electronic structure of disordered alloys and randomly substituted systems .
we have used a simple tight binding based model to describe the physics of the differences between minimum size super cell calculations and large super cells with random distributions of impurity atoms in the cell . in order to visualize the changes we present a simple unfolding procedure which is demonstrated to work very well indeed .
we have shown that with small impurity potentials the band structure is qualitatively described by a rigid shift as in a virtual crystal approximation .
however even here the bands become broadened in both energy and momentum space .
this broadening becomes very large for larger impurity potentials .
we also show that the fermi surfaces are strongly smeared out for medium size impurity potentials and that the peak position of the broadened structure as a function of momentum would yield a fermi surface determined by the crossing at an energy corresponding to the electron count i.e. the `` fermi energy '' which does not agree with the luttenger theorem size of the fermi surface . in order to quantify the effects of the impurity scattering effects we present the real and imaginary parts of the self energy which clearly demonstrates the various effects impurity scattering and randomness has on the electronic structure .
the smearing effects at the fermi energy can cause a semiconducting density of states in the ordered case turn into a metallic density of states .
this effect is quite spectacular in the comparison of a density functional based _ ab initio _ calculation of the electronic structure of ordered and disordered lih .
the results we show for mg li and li h alloys using density functional methods demonstrate that _ ab initio _ methods can be used which opens the door to the study of a large number of real material systems .
we would like to thank o. k. andersen and m. berciu for stimulating discussion .
we acknowledge financial support from the candian funding agencies nserc , cfi , and the crc program .
this work was done within the max planck/ ubc funded center for quantum materials at ubc and was enabled in part by the use of computing resources provided by westgrid and compute / calcul canada .
h. m. james , and a. s. ginzbarg , j. phys .
chem . * 57 * , 840 ( 1953 ) .
m. f. ling , and d. j. miller , phys .
b * 38 * , 6113 ( 1988 ) .
b. bouhafs , f. benkabou , m. ferhat , b. khelifa , j. p. dufour , and h. aourag , infrared phys . tech .
* 36 * , 967 ( 1995 ) .
b. a. orner , and j. kolodzey , j. appl . phys . *
81 * , 10 ( 1997 ) . c. persson , r. ahuja , and b. johansson , phys.rev .
b * 64 * , 033201 ( 2001 ) .
a. maier , and m. jarrell , phys .
lett . * 89 * , 077001 ( 2002 ) .
b. zorman , s. krishnan , d. vasileska , j. xu , and m. van schilfgaarde , j. comp .
* 3 * , 351 ( 2004 ) .
y. kakehaski , m. atiqur r. patoary , and t. tamashiro , phys .
b * 81 * , 245133 ( 2010 ) .
a. alam , t. saha - dasgupta , and a. mookerjee , phys .
b * 81 * , 054201 ( 2010 )
. a. alam , and a. mookerjee , phys .
b * 81 * , 184205 ( 2010 ) . h. wadati , i. elfimov , and g. a. sawatzky , phys .
lett . * 105 * , 157004 ( 2010 ) .
w. ku , t. berlijn , and c .- c .
lee , phys .
lett . * 104 * , 216401 ( 2010 ) .
t. berlijn , d. volja , and w. ku , phys .
* 106 * , 077005 ( 2011 ) .
s. konbu , k. nakamura , h. ikeda , and r. arita , arxiv 1108.0595 ( 2011 ) .
k. nakamura , r. arita , and h. ikeda , phys .
b * 83 * , 144512 ( 2011 ) .
a. marmodoro , and j. b. staunton , j. phys . :
series * 286 * , 012033 ( 2011 ) .
j. s. faulkner , prog .
science * 27 * , 1 ( 1982 ) .
j. r. klauder , ann .
physics * 14 * , 43 ( 1961 ) .
m. jarrell , and h. r. krishnamurthy , phys .
b * 63 * , 125102 ( 2001 ) . v. janis , phys . rev .
b * 64 * , 115115 ( 2001 )
. m. s. laad , and l. craco , j. phys . : condens .
matter * 17 * , 4765 ( 2005 ) .
g. m. batt , and d. a. rowlands , j. phys . : condens .
matter * 18 * , 11031 ( 2006 ) .
m. potthoff , and m. balzer , phys .
b , * 75 * , 125112 ( 2007 ) .
d. kdderitzsch , h. ebert , d. a. rowlands , and a. ernst , new j. physics * 9 * , 81 ( 2007 ) .
d. a. rowlands , rep .
phys * 72 * , 086501 ( 2009 ) .
a. marmodoro , and j. b. staunton , j. phys .
: conf . ser . *
286 * , 012033 ( 2011 ) . | we describe procedures to obtain the electronic structure of disordered systems using either tight binding like models or quite directly from _ ab inito _ density functional band structure calculations .
the band structure is calculated using super cells much larger than those containing a single minority component atom .
we average over a large number of different super cell calculations containing different randomly positioned minority component atoms in the super cell as well as a varying total number of minority component atoms , weighted by the statistical probability .
we develop an efficient and simple algorithm for unfolding of these bands , based on bloch s theorem .
the unfolded band - structure obtained in this way exhibits momentum and energy broadened structures replacing the gaps observed in often used single super cell calculations . using the super cell averaged
band - structure one can introduce a self - energy , resulting from the scattering of randomly positioned alloy components .
the self - energy is causal , and shows strong energy and some momentum dependence .
the self - energy shows rather non - trivial behavior and is in general non - zero at the fermi - energy , resulting in an ill or undefined fermi surface .
the real - part of the self - energy at the fermi - energy relates to an apparent violation of luttinger s theorem .
there is no simple relation between the apparent fermi - surface volume and the electron count .
examples introducing these effects both for model and real binary alloy systems are presented .
# 1*_#1 _ * the material specific properties of many systems depend crucially upon chemical substitution or doping , i.e. the intentionally introduction of impurities in an otherwise pure crystal structure .
substitutions can be chosen to introduce or change , the number of charge carriers , local magnetic moments , specific optical properties , or vortex pinning centers , just to mention a few commonly known applications .
these substituted atoms also introduce a destruction of translational symmetry and therefore scattering into the band structure as an important deviation from the pure system .
the effects of disorder resulting from random substitutions alter material properties in specific ways and the vast field of materials applications is strongly dependent on random substitutions.@xcite although we have powerful _ ab initio _ band structure methods to treat pure transitionally symmetric systems at least for systems in which electron correlation effects are less important these methods are generally not applicable directly to randomly substituted systems or disordered alloys .
there are several useful approximation schemes available to treat randomly substituted systems or alloys , which can treat the disorder to a high level of accuracy.@xcite it is desirable to create a method able to describe disordered systems with the same level of understanding and approximations as one has developed for periodic systems .
density functional theory ( dft)@xcite , within the local density approximation ( lda ) is an _ ab - initio _ method proved to be useful for a large class of materials .
as long as the materials under consideration are not too correlated , the one electron wave functions in a kohn and sham@xcite implementation can account for the material specific band - structure .
first principle methods used for the description of crystals use the periodicity of the system in order to make calculations tractable .
two distinct problems arise when dft is used to study the electronic structure of real materials : 1 ) how to account for disorder breaking the translational symmetry and 2 ) how to present the results in a meaningful way .
often , _ ab - initio _ calculations including disorder are done assuming periodic boundary conditions within a so - call super cell approximation , i.e. using a cell which is a multiple of the primitive unit cell .
it is also typical to choose the smallest cell size possible , containing one impurity atom , in order to reduce computational costs .
of course , the smallest cell with only one impurity or dopant , combined with periodic boundary conditions , can not account for a random impurity distribution in real materials .
such calculations basically result in the band structure of a fictitious new perfectly ordered material .
although in some cases such as the extreme dilute limit where the impurity impurity interference effects can be neglected this may be a reasonable approximation to the real problem it is important to develop theoretical methods based on the same approaches and approximations used for pure systems to check for the importance of randomness . for a detailed discussion of development in _ ab - initio _
theory of alloys we refer to a recent review paper by ruban and abrikosov.@xcite we too use the super cell approximation but allow for randomness in the impurity distribution across the super cell .
we then propose a simple procedure to present the unfolded calculated band structures in a way as seen by angle resolved photo emission spectroscopy ( arpes ) and demonstrate the effect of impurity scattering on the band structure and the self energy .
we first present the method used to unfold the band - structure in the first brillouin zone of the super cell to the first brillouin zone of the original cell . which is basically based on bloch s theorem and symmetrization of the wave - functions .
the method can be readily applied to all first - principle codes able to calculate partial densities of states and local symmetry or orbital projected band structures .
we then demonstrate the basic concepts of the random impurity calculations using a simple tight binding like model and on site impurity potentials .
in the last section before the conclusion we describe the use of these concepts in the dft calculation of a real alloy system composed of li substituted with either h or mg . |
the barrel part of the cms pixel detector consists of about 800 detector modules that are mounted in three layers on a structure of thin - walled aluminum pipes and carbon fibre crosspieces @xcite .
while the majority of the modules ( 672 ) are full modules as seen in fig.1 on the right , the edges of the six half - shells are equipped with 16 half - modules each ( 96 in total , see fig.1 on the left ) .
a half - module has 8 readout chips and is half the size of a full module . as seen on fig.1 a module
is composed of the following components : * 250 @xmath1 m thick silicon nitride basestrips * 8 to 16 readout chips ( rocs ) with 52x80 pixels of size 150x100 @xmath1m@xmath2 * sensor made from 285 @xmath1 m thick dofz - silicon .
* high density interconnect ( hdi ) , a flexible low mass 3 layer pcb ( flex print ) * token bit manager chip ( tbm ) controling the readout of the rocs * signal & power cable a completed full - module has the dimensions 66.6x26.0 mm@xmath2 , weight 2.2 g plus up to 1.3 g for cables , and consumes 2 w of power .
more details can be found in @xcite .
the assembly procedure of a cms barrel module divides in four parts ( see fig 2 ) : 1 .
assembly of the hdi 2 .
processing of the roc- and sensor- wafers 3 .
bump bonding of the raw module 4 .
final assembly of the module the flex print is delivered pretested and equipped with all passive electrical components by the vendor . afterwards the power and signal - cable and the tbm - chip are attached and connected with aluminum wirebonds at psi .
the assembled hdi is then tested for the functionality of the tbm and any shortcuts or broken traces in the power line or token passage .
up to 16 hdis can be assembled per working day . in the early stage of the production the bondability of the hdis after several month of storage at room conditions was not satisfactory .
it was then improved by a mechanical cleaning step with a soft rubber .
unfortunedly this method was likely to create shortcuts between adjacend bondpads and caused a considerable amount of extra work in testing and debugging of assemled hdis and fabricated modules ( 23 % of hdis not useable , 22 % of modules needed repairs ) .
the storage of the flexes in a dry nitrogen atmosphere introduced directly after the first problems were discovered eliminated the need for cleaning the bondpads and led to an excelent yield of the assembled hdis .
the fraction of failing hdis was reduced to less than @xmath32% .
the different parts are addressed in the following subsections .
the sensor and roc wafers have to undergo certain processing and manipulation steps before they can be used for the module production : first of all an acceptance test of sensors and rocs is conducted on the wafer .
after 72 roc - wafers tested the overall yield is about 70% .
the acceptance - tests on sensor wafers reflect just the results of the vendor . then follows a composition of photolithographic steps , ubm sputtering and indium evaporation and a final liftoff step to prepare the wafers for the subsequent bump bonding process @xcite the roc wafers
are then thinned down to 170 @xmath1 m and roc and sensor - wafers are diced and picked based on the results of the acceptance tests .
finally the raw devices are tested again . in the case of the rocs
this is done as part of the bump bonding process .
only chips with less than 1% statistically distributed dead or noisy pixels and without pixel masking defects are used in the following assembly steps .
a sensor is discarded when either the dark current is above 2 @xmath1a at 150 v and room temperature or the ratio of the dark currents measured at 100 v and 150 v is greater than two .
the sensors are then cleaned and undergo the reflow process where the deposited indium is melted up to prepare the indium bumps needed for the later bump bonding .
after another optical inspection the sensors are stored unter a dry nitrogen atmosphere untill they are used .
initial problems during the liftoff - process leading to very work intensive procedures could be solved by modifying the fotolithographic masks .
now the openings in the fotoresist are distributed over the whole surface of the wafer .
this reduced both the duration and complexity of the liftoff procedures and reduced the use of the ultrasonic bath that caused partial removing of the deposited indium bumps on several sensors to an absolute minimum .
another problem with corrosion like effects on the aluminum bondpads of the roc chips was solved by consequently renewing the liftoff solvend after each liftoff step , especially before the use of the ultrasonic bath .
the sandwich of sensor and the 8 to 16 roc chips needed to build a module is called raw module. building it involves the following steps : first the set of rocs is cleaned and arranged with the help of a special jig that ensures their defined positions on the gelpacks which are then placed on the bump bonding machine .
sensor and rocs are then bump bonded using the automatic bump bonding machine developed at psi @xcite ( see fig .
the individual roc chips are tested during the bump - bonding process to ensure that only electrically working chips are bump - bonded to the sensor . to form a mechanical and electrical connection between the indium bumps of sensor and rocs
the module is reflowed in the reflow oven .
afterwards the completed raw module is tested again .
the so called raw test gives a very fast feedback on the bump bonding quality and was upgraded several times to enhance its abilities .
it finally measures the roc power consumption , does a threshold scan and measures the level of the analog output .
in addition to that it does an iv curve measurement on the sensor and makes a bump yield test .
based on the result of the test the modules are released for further assembly or undergo a reworking step to remove defective chips or improve the bump yield if possible and usefull .
raw modules drawing a high current ( @xmath415 @xmath1a @ 200v ) are not repaired and are not used further on . the ability to rework single chips of a bare module with a special reworking jig ( see fig .
4 ) has constantly improved during the module production .
the rework quality is either excellent ( less than 12 bump defects per chip ) or very poor ( @xmath01000 defects per roc ) .
the overall chip replacement yield is about 85% now ( 60 of the 71 replaced chips have been good ) and even full modules ( 16 chips ) were reworked with almost perfect results ( see fig .
when the raw module has passed the last test , the construction of the module is finalised on the psi module assembly line ( see fig .
6 ) in the following steps : first the base strips are glued to the roc side of the raw module . after the glue has cured and a small amount of glue has been applied between the edge of the sensor and the rocs to reinforce the connection near the wirebonding area the pre - assembled hdi is glued on the sensor side of the raw module . after the electrical connection between the flex and the rocs
is formed with wire bonds the module assembly is complete .
each glueing step is done on a separate jig that ensures the exact placement of the parts and keeps them in place with vacuum until the glue has cured . instead of using a glue disperser to apply the desired small quantities of glue ,
the adhesive is applied with a stamp that is lowered into a glue bath and matches the form of the glueing area .
this method is a very easy but well defined way to apply the glue exactly where it is needed . in order to apply the very small amount of glue in between the rocs and the sensor ,
the stamp is replaced with kapton@xmath5 @xcite flaps that bring the glue in the opening between the sensor and the rocs .
the assembly procedure has proven itself to be very reliable and fault proof during the production so far and has not been changed even after more than 200 modules build .
the number of modules that after assembly are worse than before is only about 8 , which is less than 4% of the modules produced . at the end
the completed module undergoes the final testing .
this includes thermocycling between room temperature and -20 @xmath6c . the fully automatic test setup ( see fig .
7 ) can handle up to 4 modules in parallel and up to 2 testloads per working day .
the modules are then graded according to their test results .
the major criterias are the following : grade a modules are excellent with less than 1 permill defectice or unbonded pixels per chip and sensor dark currents of less than 2 @xmath1a at 17 @xmath6c and 150 v. grade b modules are useable modules with less than 4 percent defectice or unbonded pixels per chip and sensor dark currents up to than 10 @xmath1a at 17 @xmath6c and 150 v. all other modules are graded as c if they can not be repaired by the module doctor , which is the case for sensor defects ( high dark current ) or chip defects . if only a bond has loosened or a short on the hdi has developed or even the tbm ceases to function the module will be repaired and tested again .
the module doctor was introduced to understand and possibly repair the hdi - induced module defects and has proven to be very usefull in the first phase of the production , where many modules needed to be repaired . due to the improved quality of the assembled hdis
the module doctor-station was used only very frequent during the production of the last 100 modules .
the module assembly procedure is capable of a constant rate of 4 full and 2 half modules per working day .
in parallel to that 16 hdis can be assembled together with the processing of rocs and sensors for the bump bonding .
the manpower required to maintain this rate is about 5 ftes including the final testing of the modules .
more details about the module assembly can be found in @xcite .
up to end of november 2007 , 286 modules were bump bonded , but 30 modules were in different stages of production and testing at this time .
out of the 256 completed moduls 210 ( 82% ) modules are graded a(177 ) or b(33 ) ( 59 after repair or rework ) .
half of the 46 modules with grade c failed in the state of a raw module . about half of the modules were lost due to problems and errors in the startup of the production .
the others are rejected due to sensor problems . the cms collaboration .
cms tracker .
technical design report lhcc 98 - 6 , cern , geneva , switzerland , 1998 .
cms barrel pixel detector overview .
kaestli these proceedings .
kaestli . design and performance of the cms pixel detector readout chip .
nim a 565(2006 ) 188 - 194 t.rohe et al.fluence dependence of charge collection of irradiated pixel sensors .
arxiv : physics/0411214 v3 4 jan 2005 kapton@xmath5 is a registrated trademark of dupont for its polyimide film ch .
brnnimann et al .
development of an indium bump bond process for silicon pixel detectors at psi .
these proceedings .
assembly of the cms pixel barrel modules .
nim a 565(2006 ) 62 - 66 | for the barrel part of the cms pixel tracker about 800 silicon pixel detector modules are required .
the modules are bump bonded , assembled and tested at the paul scherrer institute .
this article describes the experience acquired during the assembly of the first @xmath0200 modules .
cms , pixel detector , module assembly |
the spread of an epidemic among a community of individuals has been described by several types of models , either deterministic or stochastic @xcite . among the latter
we find models in which the space structure is explicitly taken into account @xcite .
these models are defined on a lattice which represents the space where the community lives , and in which each site is occupied by just one individual . in the susceptible - infected - recovered ( sir ) stochastic lattice model @xcite ,
each individual can be either susceptible ( s ) , infected ( i ) or recovered ( r ) .
a susceptible individual becomes infected ( s@xmath4i ) through an autocatalytic reaction , and an infected individual recovers ( i@xmath4r ) spontaneously .
this model describes an epidemic in which the immunization is permanent ; in other words , once an individual recovers , it becomes immune forever .
the main features of the sir model are as follows .
when the rate of infection of a susceptible individual by an infected individual is small compared to the immunization rate , there is no spreading of the disease . increasing the infection rate , one reaches a critical value above which the infection spreads over the whole lattice .
the transition from one regime to the other is regarded a continuous phase transition whose critical behavior places the model into the dynamic percolation universality class , which corresponds to the standard percolation class with the addition of dynamical growth exponents @xcite . as shown below , the sir model can be described by just one parameter , either the reduced infection rate @xmath5 or the reduced immunization rate @xmath6 .
the phase transition occurs at a critical value @xmath7 , which has been estimated as @xmath8 on a square lattice by means of time - dependent numerical simulations @xcite .
a more accurate result @xmath9 was later determined by extensive numerical simulation through the use of a technique borrowed from standard percolation theory @xcite .
a close relationship exists between the sir model and dynamic ( isotropic ) percolation @xcite . starting from a single infected individual in a lattice full of susceptibles , a cluster composed of infected and recovered individuals grows , the infected individuals staying at the border of the cluster and the recovered individuals inside it .
eventually the cluster becomes composed of recovered individuals only due to the spontaneous immunization . for @xmath10 ( the spreading regime ) , an infinite cluster of recovered individuals
percolates the whole lattice . for @xmath11 ( the non - spreading regime ) , only finite clusters are present .
it has been shown @xcite that the cluster probability distribution obeys the same scaling laws as apply to standard percolation models .
here we report numerical simulations and finite - size scaling analysis to obtain the critical behavior of the stochastic sir model on a square lattice . to this end
we determine the order parameter @xmath2 ( defined below ) , the mean number of recovered sites @xmath12 , and the mean value of the squared number of recovered sites @xmath13 .
these quantities are determined as a function of @xmath14 and the linear system size @xmath15 .
we show that the ratio @xmath16 between the second moment and the square of the first moment is not independent of the system size @xmath15 , at criticality , as occurs , e. g. , in the contact model @xcite . instead
, we show here that the quantity that is independent of the system size at criticality , and therefore universal , is the product @xmath0 between this ratio and the order parameter , consistent with the dynamical percolation universality class .
this paper is organized as follows . the stochastic sir model on a regular lattice
is defined in section [ 2 ] together with numerical simulations . in section
[ 3 ] we explain how the model is related to percolation around criticality . the finite - size analysis is introduced in section [ 4 ] .
a conclusion is drawn in the last section .
the stochastic sir model is defined on a regular lattice of @xmath17 sites as follows . at each time
step a site is chosen at random and the time is incremented by an amount equal to @xmath18 . if the chosen site is in state s then it becomes i with probability @xmath5 multiplied by the fraction of nearest - neighbor sites in state i. if the chosen site is in state i then it becomes r with probability @xmath6 .
if it is in state r it remains in this state .
the number of individuals of type s , i and r are denoted by @xmath19 , @xmath20 and @xmath21 .
the total number of individuals equals the total number of sites of the lattice , @xmath22 .
the quantities that we have measured in the monte carlo simulation are the following : the mean number of recovered individuals @xmath23 the mean value of the square of the number of recovered individuals @xmath24 and the order parameter @xmath2 , defined below .
we also considered the ratio @xmath1 between the second moment @xmath13 and the square of the first moment @xmath12 of the probability distribution of recovered individuals , that is , @xmath25 the simulations were performed on a square lattice of @xmath26 sites and periodic boundary conditions .
we begin with an infected individual placed at the center of the lattice full of susceptible individuals . to speed up the simulations we keep a list of the i sites . at each step of the simulation
we choose randomly an i site among the list of the @xmath20 i sites .
( if we were interested in the time , which is not the case here , we would increment it by an amount equal to @xmath27 . ) with probability @xmath14 the chosen i site becomes an r site . with the complementary probability @xmath28 we choose one of its four nearest - neighbor sites ; if the nearest - neighbor site is an s site , it becomes an i site , otherwise it remains unchanged .
these rules are equivalent to the definition of the model given in the beginning of this section . here
we are interested only in the stationary states , which are characterized by the absence of infected sites . starting from a single infected site ,
the number of infected sites may increase but eventually decreases and vanishes . without infected sites there is no activity and the dynamics stops .
the stationary state is then an absorbing state constituted of s and r sites only . for each value of the parameter @xmath14 and linear size
@xmath15 we performed a set of independent runs , ranging from @xmath29 to @xmath30 , and measured the quantities @xmath12 , @xmath13 and @xmath2 related to the final clusters of r sites . in figure [ dens_recxc ]
we show the density of recovered individuals @xmath31 as a function of @xmath14 for several values of the system size @xmath15 . in figure [ cumxc ]
we show the ratio @xmath16 as a function of @xmath14 for several values of the system size @xmath15 . to find the order parameter @xmath2 , we checked whether the growing cluster of infected individuals reached the border of the lattice , and the fraction of runs for which this happens is defined as @xmath2 . in the thermodynamic limit
, it becomes the probability that the central site belongs to the infinite cluster in accordance with the definition of order parameter in standard percolation theory . in figure [ pxc ]
we show @xmath2 versus @xmath14 for several values of the linear size @xmath15 .
we note that in percolation one often uses the fraction of sites belonging to the largest cluster in a lattice fully occupied with clusters as the order parameter .
however , this method can not be used here because by definition there is only one cluster in each epidemic sample .
note also that in our definition of @xmath2 , one can think of the system as having open boundary conditions and we are finding if the cluster hits that boundary ; this is related of the idea of midpoint percolation considered recently @xcite .
an equivalent interpretation is that we have an infinite lattice , and we are seeing if the epidemic starting from the origin crosses the @xmath32 boundary .
here we summarize some results of the percolation theory @xcite that will be useful for showing the relation to the sir model . in standard percolation theory the probability that a site belongs to a cluster of size @xmath33 is @xmath34 where @xmath35 is the mean number of clusters of size @xmath33 per lattice site .
( for site percolation , one should also technically multiply this by @xmath36 , the probability that the site is occupied , but we will suppress this factor . ) from this quantity one obtains the mean epidemic size @xmath12 , @xmath37 and the mean - square epidemic cluster size @xmath13 , @xmath38 the order parameter @xmath2 is the probability that a site belongs to the infinite percolating cluster . around the critical point @xmath39 for an infinite system ,
these quantities behave as @xmath40 @xmath41 and @xmath42 where @xmath43 and @xmath44 and @xmath45 are critical exponents associated to the order parameter and to the mean cluster size , respectively .
the quantity @xmath36 is the parameter associated with percolation problem and @xmath46 is its critical value . in site ( bond )
percolation the parameter @xmath36 is the probability that a site ( bond ) is occupied .
it has been argued that the clusters of recovered individuals generated by the dynamics of the stochastic sir model follow the statistics of the cluster size distribution of the standard percolation theory summarized here @xcite . the mean cluster size in percolation theory is then identified as the mean number of recovered individuals .
this leads us to identify the quantities @xmath2 , @xmath12 and @xmath13 defined in the previous section for the sir model with the quantities @xmath2 , @xmath12 and @xmath13 defined in this section . as a consequence
, their critical behavior is given equations ( [ 8 ] ) , ( [ 9 ] ) and ( [ 10 ] ) with @xmath47 an implication of the critical behavior ( [ 9 ] ) and ( [ 10 ] ) of @xmath12 and @xmath13 is that the ratio @xmath16 should behave as @xmath48
we start from the assumption that a standard finite - size scaling analysis can be performed to obtain the critical behavior of this model .
we assume that the phase transition in the sir model is characterized by a correlation length @xmath49 which diverges in the limit where the system is infinite as @xmath50 here we use @xmath51 rather than the usual @xmath52 of standard percolation because we consider the sir cluster growth as a dynamical process .
the linear size of the system @xmath15 scales as @xmath49 and a finite - system quantity @xmath53 will behave according to the finite - size scaling as @xmath54 where @xmath55 is a universal function .
the exponent @xmath56 describes the behavior of @xmath57 in the limit of the infinite system , that is , @xmath58 using the finite - size scaling , we may write the following relations for the quantities @xmath2 , @xmath12 , @xmath13 and @xmath1 , @xmath59 @xmath60 @xmath61 @xmath62 and at the critical point @xmath63 , we have @xmath64 @xmath65 @xmath66 the quantities @xmath2 and @xmath12 are plotted as a function of @xmath15 in figures [ pbordaxl_rq ] , and [ nzxl_rq ] .
each curve was obtained by performing a number of runs of the order of @xmath29 . from the log - log plots we may estimate the critical exponents . from the slope of a straight line fitted to the data points of figure [ pbordaxl_rq ] we get the value @xmath67 and from figure [ nzxl_rq ]
we get the exponent @xmath68 .
these results should be compared with the exact results @xmath69 and @xmath70 coming from the exact values @xmath71 , @xmath72 and @xmath73 of percolation in two dimensions .
if we multiply equations ( [ pfinite ] ) and ( [ ufinite ] ) for the order parameter @xmath2 and the ratio @xmath1 we get @xmath74 at the critical point @xmath63 the product @xmath0 is then a quantity independent of @xmath15 and may be used to located the critical point .
figure [ upxp ] shows a plot of the quantity @xmath0 versus @xmath14 for several values of the system size @xmath15 .
we see that the curves indeed cross each other , for sufficiently large values of @xmath15 , at a point identified as the critical point . from the plot
we get the value @xmath75 in agreement with the result @xmath9 @xcite , and the value @xmath3 for @xmath0 at the critical point . using the critical points and the critical exponents we have done a data collapse for the quantities @xmath2 , @xmath12 and @xmath0 , shown in figures [ pb_reescalado ] , [ nz_reescalado ] and [ up_colapso ] , respectively .
these plots confirm that the critical behavior of the stochastic sir model obeys the finite - size scaling defined above .
in figure [ upcriticalplot ] we plot @xmath0 at the critical point @xmath76 for systems of sizes @xmath77 , @xmath78 , @xmath79 , @xmath80 , and @xmath81 .
here we did at least @xmath30 samples for each size .
as seen in that figure , we find a very good fit assuming the finite - size corrections are proportional to @xmath82 .
the data extrapolate to a value of @xmath83 for an infinite system .
we also ran similar simulations of standard site and bond percolation at their critical points @xmath84 and @xmath85 respectively , using an epidemic growth algorithm to generate the percolation clusters , and also show those results in figure [ upcriticalplot ] . for bond percolation
, we characterized the cluster by the number of sites that are visited or wetted . " for site percolation , we consider two definitions of the cluster mass : the first is the number of occupied sites of the cluster , which conforms to the standard definition in percolation , while for the second we used both the number of occupied sites and the number of vacant sites surrounding the clusters ( the so - called perimeter sites ) to characterize the cluster size .
the latter definition corresponds to the site - percolation limit of the sir model , in which an i site simultaneously infects all its s neighbors with probability @xmath36 , and then recovers , so that the r sites correspond to both occupied and vacant sites of the percolation cluster .
these systems show similar correction - to - scaling @xmath86 , but with different coefficients
positive for bond and regular site percolation , negative for the sir model and the occupied+vacant form of site percolation .
the extrapolation of all three systems to @xmath87 is a common value @xmath3 , showing that the sir model is equivalent to percolation not just for the critical exponents but for this amplitude ratio as well .
note that a correction - to - scaling behavior of @xmath88 is often seen in percolation problems when there are boundaries or lattice effects present @xcite , but the precise source of the corrections to scaling is not clear here .
also , it is not clear that the exponent is exactly @xmath89 ; the data in figure [ upcriticalplot ] is well fit for exponents in the range of @xmath89 to @xmath90 ( ignoring higher - order corrections ) , depending upon the curve .
while the universality of amplitude ratios in percolation has been studied for many combinations of quantities ( e.g. , @xcite ) , it seems that the quantity @xmath0 has not been examined previously for percolation , either away from , or exactly at , the critical point ( finite - size scaling ) . for scaling away from @xmath46
, @xmath0 corresponds to the critical amplitude @xmath91 or @xmath92 , which has been studied for the ising model @xcite but evidently not for percolation . for finite - size scaling at the critical point in percolation , studies have been carried out on other universal amplitude ratios @xcite , but not for @xmath0 .
the value of @xmath0 is very close to @xmath93 .
this can be understood easily for standard percolation as follows : at the critical point , the largest cluster @xmath94 will be of the order of the size of the system and much bigger than the other clusters , implying @xmath95 , @xmath96 , and ( by definition ) @xmath97 , thus yielding @xmath98 . by universality
, this also applies to the sir model .
as can be seen in figure [ up_colapso ] , for @xmath99 ( which corresponds to @xmath100 for regular percolation ) , @xmath0 goes to the value exactly 1 , as would be expected by these arguments .
we have used numerical simulations to investigate the critical behavior of the stochastic sir model on finite square lattices .
we have determined the order parameter @xmath2 , the mean number of recovered individuals @xmath12 and the mean squared number of recovered individuals @xmath13 .
these quantities obey the same scaling laws used in percolation theory .
the cluster size distribution of percolation theory is identified with the cluster distribution of recovered individuals generated by the sir dynamics . by studying lattices of different sizes
we obtain the critical behavior by means of a finite - size scaling borrowed from the stardard percolation theory .
the value of the critical exponents are in agreement with those of the isotropic percolation as one would expect .
we have shown that the ratio @xmath16 is not universal at the critical point , as is the case of the models belonging to the directed percolation universality class , and diverges as @xmath101 .
instead , we have shown that the quantity that is universal is the product @xmath0 .
the value of the critical point @xmath102 found from the fact that @xmath0 is independent of the system size is found to be in excellent agreement with previous calculations . the value of @xmath0 at the critical point
, @xmath3 is shown to be consistent with measurements of standard site and bond percolation , thus confirming that the amplitude ratios of the two models are the same , and showing a deeper level of agreement between the sir and percolation models than just having common critical exponents .
we wish to acknowledge the financial support of the brazilian agency cnpq , and rmz acknowledges partial support from the u. s. national science foundation grants no . dms-0553487 . | the critical behavior of the stochastic susceptible - infected - recovered model on a square lattice is obtained by numerical simulations and finite - size scaling .
the order parameter as well as the distribution in the number of recovered individuals is determined as a function of the infection rate for several values of the system size .
the analysis around criticality is obtained by exploring the close relationship between the present model and standard percolation theory .
the quantity @xmath0 , equal to the ratio @xmath1 between the second moment and the squared first moment of the size distribution multiplied by the order parameter @xmath2 , is shown to have , for a square system , a universal value @xmath3 that is the same as for site and bond percolation , confirming further that the sir model is also in the percolation class . |
we acknowledge j. a. w. van houwelingen for his work on testing the spiral phase plates .
this work is part of the research program of the ` stichting voor fundamenteel onderzoek der materie ( fom ) ' and is supported by the eu program atesit . | the singular nature of a non - integer spiral phase plate allows easy manipulation of spatial degrees of freedom of photon states . using two such devices ,
we have observed very high dimensional ( @xmath0 ) spatial entanglement of twin photons generated by spontaneous parametric down - conversion . in the early 1990s , it was realized that the quantum state of a photon in a laguerre - gaussian laser mode lg@xmath1 is an eigenstate of the orbital angular momentum ( oam ) operator , with eigenvalue @xmath2 ( @xmath3 integer ) @xcite . the mode functions lg@xmath1
( @xmath4 , @xmath5 ) can be used as a basis of an infinite - dimensional hilbert space , and thus to describe , within the paraxial approximation , a photon state represented by an arbitrary spatial wave function . according to theory ,
the twin photons created in spontaneous parametric down - conversion ( spdc ) are entangled in the full , infinite - dimensional ( spatial ) hilbert space ; in fact , they are maximally entangled @xcite .
several years ago , mair _ et al . _
@xcite demonstrated oam entanglement in a 3d subspace of that hilbert space , in an experiment where they used integer - oam analyzers .
it has been recently proposed that spatial entanglement can be demonstrated in an _
infinite_-dimensional subspace @xcite .
this can be achieved by using _
fractional_-oam analyzers , in view of the fact that fractional oam states are coherent superpositions of an infinite number of lg@xmath1 states with integer oam .
here we report an experimental implementation of this idea , employing two half - integer oam analyzers ; we confirm the predicted entanglement of fractional oam and deduce a dimensionality of @xmath6 per photon , the entangled space having a dimensionality @xmath7 . , where @xmath8 is the step height , @xmath9 and @xmath10 are the refractive indices of the spp and the surrounding medium , respectively , and @xmath11 is the wavelength of the incident light.,height=113 ] the key element in our approach is the use of a spiral phase plate ( spp ) @xcite , shown in fig . [
fig : spp ] , as part of our analyzer .
this plate imprints an azimuth - dependent phase retardation on an incident field .
the difference in phase retardation @xmath12 at azimuthal angles @xmath13 and @xmath14 is equal to @xmath15 , whereby the plate adds an average oam of @xmath16 per photon . by adjusting the spp parameters ( see caption of fig . [
fig : spp ] ) , the spp index @xmath17 can be tuned to any desired value .
if @xmath17 is _ non - integer _ , the output field is _ not _ cylindrically symmetric ( cf .
fig .
[ fig : quantumsetup](b ) ) and carries a mixed screw - edge dislocation in the near field , i.e. a singularity with a transverse component ; this dislocation is associated with the radial edge of the spp @xcite .
thus , for all non - integer @xmath17 , the spp is characterized by two parameters , namely the step index @xmath17 itself and the orientation of the radial edge of the spp , @xmath18 .
the latter is varied by rotating the spp around the beam axis .
our approach involves the use of two fractional - oam analyzers ( one per arm ) in an spdc setup .
such an analyzer consists of a non - integer spp , a single - mode fiber and a detector ; the orientations of the radial edges of the spps are varied ( see fig . [
fig : quantumsetup](a ) below ) .
because the spps have a non - integer step index @xmath17 , the fractional oam states provide a natural basis to describe the two - photon state @xcite .
each element of this fractional basis can be written as an @xmath17 and @xmath18 dependent superposition of an infinite number of lg@xmath1 states , extending over both @xmath3 and @xmath19 . because this superposition changes
when we reorient the spp , we explore an _
infinite_-dimensional subspace of the full , ( spatial ) hilbert space by systematically varying @xmath18 ( at constant step index @xmath17 ) .
we stress that the contribution of lg@xmath1 modes with @xmath20 to this superposition is significant ; this is caused by the fact that the dislocation imprinted by the spp on the incident field has a mixed screw - edge character @xcite .
the most convenient implementation of this scheme is by choosing the non - integer part @xmath21 of the step index @xmath17 to be half - integer , i.e. @xmath22 ; the states with @xmath23 and @xmath24 are then orthogonal ( independent of the integer part of @xmath17 ) @xcite .
when we place an analyzer containing an spp with index @xmath17 in the signal path of a spdc setup , and one containing an spp with index @xmath25 in the idler path , the coincidence probability as a function of the orientations of the two spps is predicted to be @xcite @xmath26 viz . a parabolic coincidence fringe with a visibility of unity .
the constant @xmath27 depends on the overlap of the radial part of the two - photon state as emanating from the spdc crystal , with the radial part of the fiber mode @xcite .
this parabolic coincidence fringe , which is a function of only the _ relative _ orientation of the two spps , reveals the entanglement of the twin photons .
the key to success in the experiment lies in the analyzers or , more specifically , in the two spps .
our spps have been manufactured using a polymer molding technique starting from a micro - machined mold , yielding identical plates with @xmath28 at the ( degenerate spdc ) wavelength of 813 nm .
the step of each spp is not abrupt but in fact a linear ramp with an azimuthal width of @xmath29 .
furthermore , the plates show a central ` anomaly ' with a diameter of about @xmath30 m @xcite .
the rest of the plate has an rms deviation of @xmath31 nm from the ideal shape @xcite .
the high - dimensional nature of the @xmath32 spp is illustrated in fig .
[ fig : modal](a ) , which shows a modal decomposition for mode order @xmath33 if the input is a fundamental gaussian lg@xmath34 ( i.e. the fiber mode ) .
the fidelity of this truncated decomposition is only @xmath35 ( fig .
[ fig : modal](b ) ) .
when truncating at mode order 20 ( corresponding to 231 lg modes ) , the fidelity becomes @xmath36 .
extrapolation shows that a fidelity of 98% requires about 125000 lg modes .
we have checked that these results remain essentially unchanged when using the actual topography of our spp plates ; in particular , the finite azimuthal width of the step has no significant effect .
the spdc setup is shown in fig .
[ fig : quantumsetup](a ) .
the pump laser is a @xmath37 mw krypton - ion laser ( spectra physics 2060 ) operating at a wavelength of @xmath38 nm .
its output is weakly focused on a nonlinear crystal ( bbo ) , with beam waist @xmath39 mm .
the crystal has a thickness @xmath40 mm and has been cut for nearly collinear type - ii spdc . before inserting a fractional - oam analyzer in each arm , we test the setup by observing polarization entanglement ; we achieve 95% and 92% fringe visibility in the crystal ( h , v ) and @xmath41 basis , respectively @xcite .
we fix the polarizers in the crystal basis and insert a fractional - oam analyzer with spp index @xmath42 in the signal path and an analyzer with spp index @xmath43 in the idler path . the spdc cone - crossings in the far field
are found to be much larger than the far - field intensity pattern of the analyzer mode , as illustrated in fig .
[ fig : quantumsetup](b ) .
the spdc intensity varies about @xmath44 over the spp diffraction profile . the spps have to be aligned with great precision . in particular , both the rotation axis and the center of each spp have to be aligned , with a precision of order @xmath45 m , with the core of the corresponding single - mode fiber
only when this amount of care is invested in aligning the optical elements , can one rotate both spps to make use of the @xmath46 and @xmath47 degrees of freedom without losing any signal .
our experimental results are shown in fig .
[ fig : quantumresults ] : four coincidence fringes ( obtained for idler spp - settings @xmath48 ) as a function of the signal spp - setting @xmath46 .
these fringes are shown together with the expected coincidence fringes as predicted by eq .
( [ eq : coincidenceprobability ] ) .
the data show that the coincidence - count rate as a function of the orientation of the signal spp has a parabola - like shape .
additionally , when the idler spp is re - oriented , we see that the fringe shifts accordingly . consequently , we find , according to the quantum - mechanical prediction , that the parabolic coincidence fringe depends only on the _ relative _ orientation of the spps . of the spp in the signal arm .
the parabolic fringe shifts concordantly with the orientation @xmath47 of the spp in the idler arm.,width=244 ] analytical calculations in which one assumes a completely separable ( as opposed to entangled ) two - photon state while oam is conserved per pair , result in fringes of _ zero _ visibility . in the context of a quantum - mechanical description of the spdc process
, the experimental data show thus entanglement .
however , based on the present set of data , we can not prove quantum non - locality using the two - dimensional clauser - horne - shimony - holt version of the bell inequality @xcite , in contrast to what we predicted earlier @xcite .
the dimensionality of the hilbert space in which each photon of the pair lives , must be @xmath49 as can be concluded from the parabolic shape of the fringes in fig .
[ fig : quantumresults ] ; after all , for @xmath50 , as in polarization entanglement , a sinusoidal fringe would appear . regretfully , the fringes do not directly betray the precise dimensionality ; nevertheless , we can make an estimate based on the following considerations .
the dimensionality @xmath51 of the subspace probed in the experiment is either limited by the number of entangled modes emitted by the bbo crystal , expressed by the schmidt number @xmath52 @xcite , or by the ( spatial ) modal bandwidth of the analyzers , @xmath53 . to estimate @xmath52
, we follow law and eberly s approximate analytical theory @xcite , @xmath54 our experimental values for the pump - beam waist @xmath55 , crystal length @xmath56 and the pump wavenumber @xmath57 result in a lower limit to the schmidt number , @xmath58 .
a comparison between the approximate and exact theories for the value of @xmath52 shows that , for high values of @xmath52 , as in the present case , the approximate theory yields a severe underestimation .
to make sure that our analyzers are not limiting the dimensionality of the probed subspace to @xmath59 , we need to establish that their modal bandwidth @xmath60 .
this can be accomplished by examining the conversion efficiency of the two analyzers , positioned as an optical train in a classical - optics experiment : we use one analyzer to convert the fundamental gaussian mode from a laser beam to an extended superposition of lg modes , while the other is used to convert this superposition back to the fundamental gaussian mode .
the conversion efficiency , which we found to be 98% , basically tells us how well the modal spectra of the two analyzers overlap .
combined with knowledge of the modal spectrum of a single analyzer ( cf .
fig .
[ fig : modal ] ) , this value yields the modal bandwidth @xmath53 of the analyzers .
the experimental value of 98% @xcite , which is the fidelity of one analyzer mode as measured by the other analyzer , corresponds to a modal bandwidth @xmath61 ( which corresponds , in fact , to the diffraction limit of the full optical setup ) .
consequently , the analyzers are not a bottleneck in this experiment and we can therefore conclude that the dimensionality of the probed hilbert space is @xmath62 . in conclusion , we have reported an experiment on spdc twin - photons demonstrating , within the quantum - mechanical description of spdc , the entanglement of twin photons with respect to their fractional oam . as this experiment involves a unitary transformation of basis from the natural lg states with integer oam to a new basis with fractional oam , the data explicitly shows that a very large part of the spatial hilbert space is entangled , the dimensionality of which we estimate as @xmath6 per photon .
the entangled space that we have probed in this experiment only represents one particular cross section of the full hilbert space of ( spatial ) two - photon states .
presently , we are designing analyzers with alternative spatial geometries and step heights , and , additionally , analyzers that operate on both amplitude and phase of the photon state .
experiments with such analyzers will possibly allow us to sample a large variety of cross sections of the full hilbert space . |
expert advice has become a well - established paradigm of machine learning in the last decade , in particular for prediction .
it is very appealing from a theoretical point of view , as performance guarantees usually hold in the worst case , without any ( statistical ) assumption on the data .
such assumptions are generally required for other statistical learning methods , often however not resulting in stronger guarantees . using expert advice in the standard way
seems a rather bad idea in some cases where the decisions of the learner or master algorithm influence the behavior of the environment or adversary .
one example is the repeated prisoner s dilemma when the opponent plays
tit for tat " ( see section [ sec : active ] ) .
this was noted and resolved by @xcite , who introduced a strategic expert algorithm " for so - called reactive environments .
their algorithm works with a finite class of experts and attains asymptotically optimal behavior
. no convergence speed is asserted , and the analysis is quite different from that of standard experts algorithms . in this paper , we show how the more general task with a countably infinite expert class can be accomplished , building on standard experts algorithms , and simultaneously also bounding the convergence rate ( @xmath0 , which can be actually improved to @xmath1 ) . to this aim , we will combine techniques from @xcite and obtain a master algorithm which performs well on loss functions that may _ increase in time_. then this is applied to ( possibly ) reactive problems by yielding the control to the selected expert for an increasing period of time steps . using a universal expert class defined by the countable set of all programs on some fixed universal turing machine , we obtain an algorithm which is in a sense asymptotically optimal with respect to _ any _ computable strategy .
an easy additional construction guarantees that our algorithm is computable , in contrast to other universal approaches which are non - computable @xcite . to our knowledge
, we also propose the first algorithm for non - stochastic bandit problems with countably many arms .
the paper is structured as follows .
section [ sec : algorithm ] introduces the problem setup , the notation , and the algorithm . in sections [ sec :
master ] , we give the ( worst - case ) analysis of the master algorithm . the implications to active experts problems and a universal master algorithms are given in section [ sec : active ] .
we discuss our results in section [ sec : discussion ] .
we are acting in an online decision problem .
we " is here an abbreviation for the master algorithm which is to be designed .
an online decision problem " is to be understood in a very general sense , it is just a sequence of decisions each of which results in some loss
. this could be e.g. a prediction task , a repeated game , etc . in each round , that is at each time step @xmath2 , we have access to the recommendations of countably infinitely many experts " or strategies .
( for simplicity , we restrict our notation to a countably infinite expert class , all results also hold for finite classes . ) we do not specify what exactly a
recommendation " is we just follow the advice of one expert . _ before _ we reveal our move , the adversary has to assign losses @xmath3 to _ all _ experts @xmath4 .
there is an upper bound @xmath5 on the maximum loss the adversary may use .
this quantity may depend on @xmath2 and is not controlled by the adversary . after the move , only the loss of the selected expert @xmath4 is revealed .
our goal is to perform nearly as well as the best available strategy ( expert ) in terms of cumulative loss , after any number @xmath6 of time steps which is not known in advance .
the difference between our loss and the loss of some expert is also termed _ regret_. we consider the general case of an _ adaptive _ adversary , which may assign losses depending on our past decisions . if there are only finitely many experts or strategies , then it is common to give no prior preferences to any of them .
formally , this is realized by defining uniform _ prior weights _
@xmath7 for each expert @xmath4 .
this is not possible for countably infinite expert classes , as there is no uniform distribution on the natural numbers . in this case , we need some non - uniform prior @xmath8 and require @xmath9 for all experts @xmath4 and @xmath10 .
we also define the complexity of expert @xmath4 as @xmath11 .
this quantity is important since in the full observation game ( i.e. after our decision we get to know the losses of _ all _ experts ) , the regret can usually be bounded by some function of the best expert s complexity . our algorithm follow or explore " ( @xmath12 , specified in fig .
[ fig : foe ] ) builds on mcmahan and blum s online geometric optimization algorithm @xcite .
it is a bandit version of a follow the perturbed leader " experts algorithm .
this approach to online prediction and playing repeated games has been pioneered by hannan @xcite . for the full observation game and uniform prior
, @xcite gave a very elegant analysis which is clearly different from the standard analysis of exponential weighting schemes .
it has one advantage over other aggregating algorithms such as exponential weighting schemes : the analysis is not complicated if the learning rate is dynamic rather than fixed in advance .
a dynamic learning rate is necessary if there is no target time @xmath6 known in advance . for non - uniform
prior , an analysis was given in @xcite .
the following issues are important for @xmath12 s design . .
since we are playing the bandit game ( as opposed to the full information game ) , we need to explore sufficiently @xcite . at each time step @xmath2 , we decide randomly according to some exploration rate @xmath13 whether to explore or not .
if so , we would like to choose an expert according to the prior distribution .
there is a caveat : in order to make the analysis go through , we have to assure that we are working with _ unbiased _ estimates of the losses .
this is achieved by dividing the observed loss by the probability of choosing the expert . but this quantity could become arbitrarily large if we admit arbitrarily small weights .
we address this problem by _ finitizing _ the expert pool at each time @xmath2 . for each expert @xmath4
, we define an _ entering time _
@xmath14 , that is , expert @xmath4 is active only for @xmath15 .
we denote the set of active experts at time @xmath2 by @xmath16 . for exploration , the prior
is then replaced by the finitized prior distribution @xmath17 , [ eq : unonu ] ( u_t = i)=w^i_t^i_j w^j_t^j .
consequently , the maximum unbiasedly estimated instantaneous loss is ( note that the exploration probability also scales with the exploration rate @xmath18 ) [ eq : maxest ] b_t=. it is convenient for the analysis to assign estimated loss of @xmath19 to all currently inactive experts .
observe finally that in this way , our master algorithm @xmath12 always deals with a finite expert class and is thus computable .
( @xmath20 , specified in fig .
[ fig : fpl ] ) is invoked if @xmath12 does not explore .
just following the leader " ( the best expert so far ) may not be a good strategy @xcite .
instead we subtract an exponentially distributed perturbation @xmath21 from the current score ( the complexity penalized past loss ) of the experts .
an important detail of the @xmath20 subroutine is the _ learning rate _ @xmath22 , which should be adaptive if the total number of steps @xmath6 is not known in advance .
please see e.g.@xcite for more details .
also the variant of @xmath20 we use ( specified in fig . [ fig : fpl ] ) works on the finitized expert pool .
note that each time randomness is used , it is assumed to be _ independent _ of the past randomness .
performance is evaluated in terms of true or estimated cumulative loss , this is specified in the notation .
e.g. for the true loss of @xmath20 up to and including time @xmath6 we write @xmath23 , while the estimated loss of @xmath12 and not including time @xmath6 is @xmath24 .
the following analysis uses mcmahan and blum s trick @xcite in order to prove bounds against adaptive adversary . with a different argument
, it is possible to circumvent lemma [ lemma : behbeh ] , thus achieving better bounds @xcite .
this will be briefly discussed in the last section .
let @xmath25 be some sequence of upper bounds on the instantaneous losses , @xmath13 be a sequence of exploration rates , and @xmath22 be a _ decreasing _ sequence of learning rates .
the analysis proceeds according to the following diagram ( where @xmath26 is an informal abbreviation for the loss and always refers to cumulative loss , but sometimes additionally to instantaneous loss ) .
[ eq : diagram ] ^ ^ each @xmath27 " means that we bound the quantity on the left by the quantity on the right plus some additive term .
the first and the last expressions are the losses of the @xmath12 algorithm and the best expert , respectively . the intermediate quantities belong to different algorithms , namely @xmath12 , @xmath20 , and a third one called @xmath28 for infeasible " fpl @xcite .
@xmath28 , as specified in fig .
[ fig : ifpl ] , is the same as @xmath20 except that it has access to an oracle providing the current estimated loss vector @xmath29 ( hence infeasible ) . then it assigns scores of @xmath30 instead of @xmath31 .
the randomization of @xmath12 and @xmath20 gives rise to two filtrations of @xmath32-algebras .
by @xmath33 we denote the @xmath32-algebra generated by the @xmath12 s randomness up to time @xmath2 , meaning _ only _ the random variables @xmath34 .
then @xmath35 is a filtration ( @xmath36 is the trivial @xmath32-algebra ) .
we may also write @xmath37 .
similarly , @xmath38 is the @xmath32-algebra generated by the @xmath12 s _ and _
@xmath20 s randomness up to time @xmath2 ( i.e. @xmath39 )
. then clearly @xmath40 for each @xmath2 .
the reader should think of the expectations in ( [ eq : diagram ] ) as of both ordinary and _ conditional _ expectations .
conditional expectations are mostly with respect to @xmath12 s past randomness @xmath41 .
these conditional expectations of some random variable @xmath42 are abbreviated by _
t[x]:=. then @xmath43 $ ] is an @xmath41-measurable random variable , meaning that its value is determined for fixed past randomness @xmath41 .
note in particular that the estimated loss vectors @xmath44 are random vectors which depend on @xmath12 s randomness @xmath33 up to time @xmath2 . in this way , @xmath12 s ( and @xmath20 s and @xmath28 s ) actions depend on @xmath12 s past randomness . note , however , that they do not depend on @xmath20 s randomness @xmath45 .
we now start with proving the diagram ( [ eq : diagram ] ) . in order to understand the analysis , it is important to consider each intermediate algorithm as a stand - alone procedure which is actually executed ( with an oracle if necessary ) on the specified inputs ( e.g. on the estimated losses ) and has the asserted performance guarantees ( e.g. again on the estimated losses ) .
[ lemma : foefoe ] @xmath46 $ ] for each @xmath47 and @xmath48 , with probability at least @xmath49 we have _ t=1^t_t_t + .
the sequence of random variables @xmath50 $ ] is a martingale with respect to the filtration @xmath38 ( not @xmath33 ! ) . in order to see this ,
observe @xmath51= \expect\big(\expect[\loss\foe_t|\cala_{t-1}]\big|\calb_{t-1}\big)$ ] and @xmath52=\loss_t\foe$ ] for @xmath53 , which implies & = & _ t=1^t ( - ) + & = & _ t=1^t-1 ( _ t- ) = x_t-1 .
its differences are bounded : @xmath54 .
hence , it follows from azuma s inequality ( see e.g.@xcite ) that the probability that @xmath55 exceeds some @xmath56 is bounded by @xmath57 .
requesting @xmath58 and solving for @xmath59 gives the assertion .
[ lemma : foefpl ] @xmath60 $ ] @xmath61 holds @xmath62 .
this follows immediately from the specification of @xmath12 .
clearly , a corresponding assertion for the ordinary expectations holds by just taking expectations on both sides .
this is the case for all subsequent lemmas , except for lemma [ lemma : behbeh ] .
the next lemma relating @xmath63 and @xmath64 is technical but intuitively clear .
it states that in expectation , the real loss suffered by @xmath20 is the same as the estimated loss .
this is simply because the loss estimate is unbiased . a combination of this and the previous lemma was shown in @xcite .
note that @xmath65 is the loss @xmath66 estimated by @xmath12 , but for the expert @xmath67 chosen by @xmath20 .
[ lemma : fplfpl ] @xmath68 $ ] for each @xmath69 , we have @xmath70 .
let @xmath71 $ ] be the probability distribution over actions @xmath4 which @xmath20 uses at time @xmath2 , depending on the past randomness @xmath41 .
let @xmath17 be the finitized prior distribution ( [ eq : unonu ] ) at time @xmath2 . then
_ t[_t ] ( 1-_t)0 + _ t _ i=1^f_t^i [ ( 1-u_t^i)0+u_t^i _ t^i|_r_t=1i_t = i ] _ i=1^f_t^i_t^i = _ t[_t ] , where @xmath72 is the estimated loss under the condition that @xmath12 decided to explore ( @xmath73 ) and chose action @xmath74 .
the following lemma relates the losses of @xmath20 and @xmath28 .
it is proven in @xcite and @xcite .
we give the full proof , since it is the only step in the analysis where we have to be careful with the upper loss bound @xmath75 .
let @xmath19 be the upper bound on the estimated loss ( [ eq : maxest ] ) .
( we remark that also for _ weighted averaging forecasters _ , losses which grow sufficiently slowly do not cause any problem in the analysis . in this way , it is straightforward to modify the algorithm by auer et al .
@xcite for reactive tasks with a finite expert class . )
[ lemma : fplifpl ] @xmath76 $ ] for all @xmath69 , @xmath77 holds . if @xmath78 , @xmath79 and thus @xmath80 holds .
this happens with probability @xmath81 .
otherwise we have [ eq : efpl ] _ t_t= _
i=1^_i_t = i _
t^i d(x ) , where @xmath82 denotes the ( exponential ) distribution of the perturbations , i.e. @xmath83 and density @xmath84 .
the idea is now that if action @xmath4 was selected by @xmath20 , it is because of the exponentially distributed perturbation with high probability also selected by @xmath28 .
formally , we write @xmath85 for @xmath86 , abbreviate @xmath87 , and denote by @xmath88 the integration leaving out the @xmath4th action .
then , using @xmath89 @xmath90 if @xmath91 in the first equation , and @xmath92 in the last line , we get [ eq:1 ] _ i_t = i_t^i d(x ) & = _ x_i _ t^i
d(x_i ) d(x_i)= _
t^i^-(_ji\{_t(_i-_j)+x_j})^+ d(x_i ) + & _ t^i^_t b_t^-(_ji\{_t(_i-_j)+x_j}+_t b_t)^+ d(x_i ) + & ^_t b_t
_ t^i^-(_ji\{_t(_i+_t^i-_j-_t^j)+x_j})^+ d(x_i ) + & = ^_t b_t_i_t = i _ t^i d(x ) . summing over @xmath4 and using the analog of ( [ eq : efpl ] ) for @xmath28 , we see that if @xmath73 , then @xmath93 holds .
thus @xmath94 .
the assertion now follows by taking expectations w.r.t .
@xmath95 .
the next lemma relates the losses of @xmath28 and the best action in hindsight . for an oblivious adversary ( which means that the adversary s decisions do not depend on our past actions ) ,
the proof is quite simple @xcite .
an additional step is necessary for an adaptive adversary @xcite .
[ lemma : ifplbeh ] @xmath96 $ ] assume that @xmath97 and @xmath14 depends monotonically on @xmath98 , i.e.@xmath99 if and only if @xmath100 .
assume decreasing learning rate @xmath101 .
for all @xmath102 and all @xmath103 , _
t=1^t_t _ t^i+ .
this is a modification of the corresponding proofs in @xcite and @xcite .
we may fix the randomization @xmath104 and suppress it in the notation .
then we only need to show [ eq : showtau ] _
i1 \{^i+ } , where the expectation is with respect to @xmath28 s randomness @xmath105 .
assume first that the adversary is oblivious .
we define an algorithm @xmath106 as a variant of @xmath28 which samples only one perturbation vector @xmath107 in the beginning and uses this in each time step , i.e. @xmath108 .
since the adversary is oblivious , @xmath106 is equivalent to @xmath28 in terms of expected performance .
this is all we need to show ( [ eq : showtau ] ) .
let @xmath109 and @xmath110 , then @xmath111 .
recall @xmath16 .
we argue by induction that for all @xmath102 , [ eq : ifplbehtau ] _ t=1^t _ t^a_t^i+ _
t\{}. this clearly holds for @xmath112 . for the induction step , we have to show [ eq : showtau2 ] _ t^i+ _ t\{}+ _
t+1^a & ^i^a_t+1 + _ t+1\{}+ _ t+1^i^a_t+1 + & = _ t+1_1:t+1^i+
_ t+1\{}. the inequality is obvious if @xmath113 .
otherwise , let @xmath114 .
then _ t^i+ _ t\ { } ^j+ & = & _ t=1^t_t^j _ t=1^tb_t + & = & _ t=1^t_t^i_t+1^a ^i^a_t+1 + _ t+1\ { } shows ( [ eq : showtau2 ] ) . rearranging terms in ( [ eq : ifplbehtau ] ) , we see _ t=1^t _ t^a _ t^i + _
t^i\ { } + _ t=1^t ( q - k)^i_t^a(- ) the assertion ( [ eq : showtau ] ) still for oblivious adversary and @xmath108 then follows by taking expectations and using [ eq : showtau3 ] _ t^i _ t\{^i+ - } & & _ i1\{^i+ } + [ eq : showtau4 ] _
t=1^t ( q - k)^i_t^a(- ) & & _
t\ { } . the second inequality of ( [ eq : showtau3 ] ) holds because @xmath14 depends monotonically on @xmath98 , and @xmath115 , and maximality of @xmath116 for @xmath117 . the second inequality of ( [ eq : showtau4 ] )
can be proven by a simple application of the union bound , see e.g. ( * ? ? ?
* lem.1 ) . sampling the perturbations
@xmath21 independently is equivalent under expectation to sampling @xmath107 only once .
so assume that @xmath21 are sampled independently , i.e. that @xmath28 is played against an oblivious adversary : ( [ eq : showtau ] ) remains valid .
in the last step , we argue that then ( [ eq : showtau ] ) also holds for an _ adaptive _ adversary .
this is true because the future actions of @xmath28 do not depend on its past actions , and therefore the adversary can not gain from deciding after having seen @xmath28 s decisions .
this argument can be made formal , as shown in ( * ? ? ?
* lemma 12 ) .
( note the subtlety that the future actions of @xmath12 would depend on its past actions . ) finally , we give a relation between the estimated and true losses ( adapted from @xcite ) .
[ lemma : behbeh ] @xmath118 $ ] for each @xmath102 , @xmath48 , and @xmath103 , we have ( i ) & & ^i ^i+ + _ t=1^^i-1b_t + ( ii ) & & ^i ^i+ + + _ t=1^^i-1b_t . for @xmath15 , @xmath119 is a martingale , since = -^i = x_t-1 + -_t^i = x_t-1 .
it is clear that @xmath120 .
moreover , @xmath121 for @xmath15 , i.e. we have bounded differences . by azuma s inequality , the actual value @xmath122 does not exceed @xmath123 with probability @xmath49 .
this proves @xmath124 .
to arrive at @xmath125 , take expectations and observe that @xmath124 fails with probability at most @xmath126 , in which case @xmath127 holds .
we now combine the above results and derive an upper bound on the expected regret of @xmath12 against an adaptive adversary .
[ th : general ] let @xmath97 , @xmath14 depend monotonically on @xmath98 , and the learning rate @xmath101 be decreasing .
let @xmath128 be some possibly adaptive assignment of ( true ) loss vectors satisfying @xmath129 .
then for all experts @xmath4 , we have with probability at least @xmath130 ^i++ _ t=1^^i-1b_t + _ t=1^t_t_tb_t^2 + _
t=1^t_t b_t + ( + ) .
consequently , in expectation , we have ^i++ _ t=1^^i-1b_t + _ t=1^t_t_tb_t^2 + _
t=1^t_t b_t + + . this follows by summing up all excess terms in the above lemmas . recall that we only need to take expectations on both sides of the assertions of lemmas [ lemma : foefpl][lemma : ifplbeh ] in order to obtain the second bound on the expectation ( and we do nt need lemma [ lemma : foefoe ] there ) .
[ cor : arbitrary ] assume the conditions of theorem [ th : general ] and choose @xmath131 and @xmath132 . then ( i ) & b_t1,^i=(w^i)^-8&^i+o(()^11 + k^i t^ ) , + ( ii ) & b_t1,^i=(w^i)^-8&^i+o(()^11 + k^i t^ ) 1-t^-2 , + ( iii ) & b_t = t^,^i=(w^i)^-16&^i+o(()^22 + k^i t^ ) , + ( iv ) & b_t = t^,^i=(w^i)^-16&^i+o(()^22 + k^i t^)1-t^-2 , for all @xmath4 and @xmath102 ( recall @xmath11 ) . moreover , in both cases ( bounded and growing @xmath75 ) @xmath12 is asymptotically optimal w.r.t.each expert , i.e. for all @xmath4 , _ t 0 the asymptotic optimality is sometimes termed _
hannan - consistency _ , in particular if the limit equals zero .
we only show the upper bound .
assertions @xmath124-@xmath133 follow from the previous theorem : set @xmath134 , abbreviate @xmath135 , and observe that for @xmath136 and @xmath137 , we have w^_t=\{w^i : t(w^i)^-}&\{w^i : t^-w^i}t^- + _ t=1^^i-1b_t ( ^i-1)b_^i-1 & ( note @xmath138 )
. then @xmath124 and @xmath125 follow from @xmath139 , @xmath140 , and @xmath141 and @xmath133 follow from @xmath142 , @xmath143 .
the asymptotic optimality finally follows from the borel - cantelli lemma , since according to @xmath125 and @xmath133 , for an appropriate @xmath144 . as mentioned in the first paragraph of this section , it is possible to avoid lemma [ lemma : behbeh ] , thus arriving at better bounds .
e.g. in @xmath124 , choosing @xmath145 , @xmath132 , and @xmath131 , a regret bound of @xmath146 can be shown .
of course , also a corresponding high probability bound like @xmath125 holds .
likewise , for a similar statement as @xmath141 , we may set @xmath147 , @xmath148 , @xmath132 , and @xmath131 , arriving at a regret bound of @xmath149 generally , in this way any regret bound @xmath150 is possible , at the cost of increasing @xmath151 where @xmath152 .
_ regret _ can become a quite subtle notion if we start considering reactive environments , i.e. care for future consequences of a decision .
an extreme case is the
heaven - hell " example : we have two experts , one always playing 0 ( saying a prayer " ) , the other one always playing 1 ( cursing " )
. if we always follow the first expert , we stay in heaven and get no loss in each step .
as soon as we curse " only once , we get into hell and receive maximum loss in all subsequent time steps . clearly , _ any _ algorithm without prior knowledge must fail " in this situation .
one way to get around this problem is taking into account the _ actual _ ( realization of the ) game we are playing .
for instance , after cursing " once , also the praying expert goes to hell together with us and subsequently has maximum loss .
hence , were are interested in a regret defined as @xmath153 as in the previous section .
so what is missing ?
this becomes clear in the following example .
consider the repeated
prisoner s dilemma " against the tit - for - tat strategy @xcite .
if we use two strategies as experts , namely always cooperate " and always defect " , then it is clear that always cooperating will have the best long - term reward .
however standard expert advice or bandit master algorithm will not discover this , since it compares only the losses in one step , which are always lower for the defecting expert .
to put it differently , minimizing short - term regret is not at all a good idea here .
e.g. always defecting has no regret , while for always cooperating the regret grows _ linearly_. but this is only the case for short - term regret , i.e. if we restrict to time intervals of length one .
we therefore give the control to a selected expert for _ periods of increasing length_. precisely , we introduce a new time scale @xmath154 ( the _ basic _ time scale ) at which we have single games with losses @xmath155 .
the master s time scale @xmath2 does not coincide with @xmath154 .
instead , at each @xmath2 , the master gives control to the selected expert @xmath4 for @xmath156 single games and receives loss @xmath157 .
( the points @xmath158 in basic time are defined recursively , see fig .
[ fig : foetilde ] . )
assume that the game has bounded instantaneous losses @xmath159 $ ] .
then the master algorithm s instantaneous losses are bounded by @xmath75 .
we denote the algorithm , which is completely specified in fig .
[ fig : foetilde ] , by @xmath160 .
then the following assertion is an easy consequence of the previous results .
[ cor : active ] assume @xmath160 plays a repeated game with bounded instantaneous losses @xmath161 $ ] .
choose @xmath132 , @xmath131 , @xmath162 and @xmath163 .
then for all experts @xmath4 and all @xmath164 , _
1:t & & _ 1:t^i+ o(()^22+k^i t^ ) 1-t^- + _ 1:t & & _
1:t^i+o(()^22+k^i t^ ) .
consequently , @xmath165 a.s .
the rate of convergence is at least @xmath166 , and it can be improved to @xmath167 at the cost of a larger power of @xmath168 .
this follows from changing the time scale from @xmath2 to @xmath154 in corollary [ cor : arbitrary ] : @xmath154 is of order @xmath169 .
consequently , the regret bound is @xmath170 .
broadly spoken , this means that @xmath171 performs asymptotically as well as the best expert .
asymptotic performance guarantees for the strategic experts algorithm have been derived in @xcite .
our results improve upon this by providing a rate of convergence
. one can give further corollaries , e.g. in terms of flexibility as defined in @xcite .
since we can handle countably infinite expert classes , we may specify a _ universal _ experts algorithm .
to this aim , let expert @xmath4 be derived from the @xmath4th ( valid ) program @xmath172 of some fixed universal turing machine .
the @xmath4th program can be well - defined , e.g. by representing programs as binary strings and lexicographically ordering them @xcite .
before the expert is consulted , the relevant input is written to the input tape of the corresponding program .
if the program halts , an appropriate part of the output is interpreted as the expert s recommendation . e.g. if the decision is binary , then the first bit suffices .
( if the program does not halt , we may for well - definedness just fill its output tape with zeros . )
each expert is assigned a prior weight by @xmath173 , where @xmath174 is the length of the corresponding program and we assume the program tape to be binary . this construction parallels the definition of solomonoff s _ universal prior _ @xcite .
[ cor : universal ] if @xmath160 is used together with a universal expert class as specified above and the parameters @xmath175 are chosen as in corollary [ cor : active ] , then it performs asymptotically at least as well as _ any computable expert _ @xmath4 .
the upper bound on the rate of convergence is exponential in the complexity @xmath98 and proportional to @xmath176 ( improvable to @xmath177 ) .
the universal prior has been used to define a universal agent aixi in a quite different way @xcite . note that like the universal prior and the aixi agent , our universal experts algorithm is not computable , since we can not check if a the computation of an expert halts .
on the other hand , if used with computable experts , the algorithm is computationally feasible ( at each time @xmath2 we need to consider only finitely many experts ) .
moreover , it is easy to impose an additional constraint on the computation time of each expert and abort the expert s computation after @xmath178 operations on the turing machine .
we may choose some ( possibly rapidly ) growing function @xmath178 , e.g.@xmath179 .
the resulting master algorithm is fully computable and has small regret with respect to all resource bounded strategies .
it is important to keep in mind that corollaries [ cor : active ] and [ cor : universal ] give assertions relative to the experts performance merely on the _ actual _ action - observation sequence . in other words ,
if we wish to assess how well @xmath160 does , we have to evaluate the actual _ value _ of the best expert @xcite .
note that the whole point of our increasing time construction is to cause this actual value to coincide with the value under _ ideal _ conditions . for passive tasks
, this coincidence always holds with any experts algorithm . with @xmath160 , the actual and the ideal value of an expert coincide in many further situations , such as finitely controllable tasks " . by this
we mean cases where the best expert can drive the environment into some optimal state in a fixed finite number of time steps .
an instance is the prisoner s dilemma with tit - for - tat being the opponent .
the following is an example for a formalization of this statement .
suppose @xmath160 acts in a ( fully or partially observable ) markov decision process .
let there be a computable strategy which is able to reach an ideal ( that is optimal w.r.t .
reward ) state sequence in a fixed number of time steps .
then @xmath160 performs asymptotically optimal .
this statement may be generalized to cases where only a close to optimal state sequence is reached with high probability .
however , we need assumptions on the closeness to optimality for a given target probability , which are compatible with the sampling behavior of @xmath160 . not all environments have this or similar nice properties . as mentioned above
, any version of @xmath12 would not perform well in the heaven - hell " example .
the following is a slightly more interesting variant of the heaven - hell task , where we might wish to learn optimal behavior , however @xmath12 will not .
consider the heaven - hell example from the beginning of this section , but assume that if at time @xmath2 i am in hell and i pray " for @xmath2 consecutive time steps , i will get back into heaven . then it is not hard to see that @xmath12 s exploration is so dominant that almost surely , @xmath12 will eventually stay in hell .
simulations with some @xmath180 matrix games show similar effects , depending on the opponent .
we briefly discuss the repeated game of chicken " , the learner is the column player , the opponent s loss matrix is the transpose , choosing the fist column means to defect , the second to cooperate .
hence , in the repeated game , it is socially optimal to take turns cooperating and defecting . ] . in this game
, it is desirable for the learner to become the dominant defector " , i.e. to defect in the majority of the cases while the opponent cooperates
. let s call an opponent primitive " if he agrees to cooperate after a fixed number of consecutive defecting moves of @xmath12 , and let s call him stubborn " if this number is high
. then @xmath12 learns to be the dominant defector against any primitive opponent , however stubborn . on the other hand
, if the opponent is some learning strategy which also tries to be the dominant defector and learns faster ( we conducted the experiment with aixi @xcite ) , then @xmath12 settles for cooperating , and the opponent will be the dominant defector .
interestingly however , aixi would not learn to defect against a stubborn primitive opponent . under this point of view
, it seems questionable that there is something like a universally optimal balance of exploration vs. exploitation in active learning at all .
as mentioned in the beginning of section [ sec : master ] , the analysis we gave uses a trick from @xcite .
such a trick seems necessary , as the basic fpl analysis only works for oblivious adversary .
the simple argument from @xcite which we used in the last paragraph of the proof of lemma [ lemma : ifplbeh ] works only for full observation games ( note that considering the estimated losses , we were actually dealing with full observations there ) . in order to obtain a similar result in the partial observation case , we may argue as follows .
we let the game proceed for @xmath6 time steps with independent randomization against an adaptive adversary .
then we analyze @xmath12 s performance _ in retrospective_. in particular , we note that for the losses assigned by the adversary , @xmath12 s expected regret coincides with the regret of another , virtual algorithm , which uses ( in its fpl subroutine ) identical perturbations @xmath108 .
performing the analysis for this virtual algorithm , we arrive at the desired assertion , however without needing lemma [ lemma : behbeh ] .
this results in tighter bounds as stated above .
the argument is formally elaborated in @xcite . in practice ,
the bounds we have proven seem irrelevant except for small expert classes , although asserting almost sure optimality and even a convergence rate .
the exponential of the complexity @xmath168 may be huge .
imagine for instance a moderately complex task and some good strategy , which can be coded with mere 500 bits .
then its prior weight is @xmath181 , a constant which is not distinguishable from zero in all practical situations .
thus , it seems that the bounds can be relevant at most for small expert classes with uniform prior .
this is a general shortcoming of bandit style experts algorithms : for uniform prior a lower bound on the expected loss which scales with @xmath182 ( where @xmath183 is the size of the expert class ) has been proven @xcite . in order to get a lower bound on @xmath12 s regret in the time @xmath6 ,
observe that @xmath12 is a _ label - efficient _ learner @xcite : according to the definition in @xcite , we may assume that in each exploration step , we incur maximal loss @xmath75 .
it is immediate that the same analysis then still holds . for label - efficient prediction , cesa - bianchi et al . @xcite
have shown a lower regret bound of @xmath184 . since according to the remark at the end of section [ sec : master ] , we have an upper bound of @xmath185 , this is almost tight except for the additive @xmath186 term .
it is an open problem to state a lower bound simultaneously tight in both @xmath168 and @xmath6 .
even if the bounds , in particular @xmath168 , seem not practical , maybe @xmath12 would learn sufficiently quickly in practice anyway ?
we believe that this is not so in most cases : the design of @xmath12 is too much tailored towards worst - case environments , @xmath12 is too _
defensive_. assume that we have a good " and a bad " expert , and @xmath12 learns this fact after some time .
then it still would spend a relatively huge fraction of @xmath132 to exploring the bad expert .
such defensive behavior seems only acceptable if we are already starting with a class of good experts . de farias , d.p . ,
megiddo , n. : how to combine expert ( and novice ) advice when actions impact the environment ? in thrun , s. , saul , l. , schlkopf , b. , eds . : advances in neural information processing systems 16 . mit press , cambridge , ma ( 2004 )
kalai , a. , vempala , s. : efficient algorithms for online decision . in : proc .
16th annual conference on learning theory ( colt-2003 ) .
lecture notes in artificial intelligence , berlin , springer ( 2003 ) 506521 mcmahan , h.b . , blum , a. : online geometric optimization in the bandit setting against an adaptive adversary . in : 17th annual conference on learning theory ( colt ) .
volume 3120 of lecture notes in computer science . , springer ( 2004 ) 109123 auer , p. , cesa - bianchi , n. , freund , y. , schapire , r.e . : gambling in a rigged casino : the adversarial multi - armed bandit problem . in : proc .
36th annual symposium on foundations of computer science ( focs 1995 ) , los alamitos , ca , ieee computer society press ( 1995 ) 322331 hutter , m. : towards a universal theory of artificial intelligence based on algorithmic probability and sequential decisions .
12@xmath187 european conference on machine learning ( ecml-2001 ) ( 2001 ) 226238 cesa - bianchi , n. , lugosi , g. , stoltz , g. : minimizing regret with label efficient prediction . in : 17th annual conference on learning theory ( colt ) .
volume 3120 of lecture notes in computer science . , springer ( 2004 ) 7792 | this paper shows how universal learning can be achieved with expert advice . to this aim , we specify an experts algorithm with the following characteristics : ( a ) it uses only feedback from the actions actually chosen ( bandit setup ) , ( b ) it can be applied with countably infinite expert classes , and ( c ) it copes with losses that may grow in time appropriately slowly . we prove loss bounds against an adaptive adversary . from this
, we obtain a master algorithm for reactive " experts problems , which means that the master s actions may influence the behavior of the adversary .
our algorithm can significantly outperform standard experts algorithms on such problems . finally , we combine it with a universal expert class .
the resulting universal learner performs in a certain sense almost as well as any computable strategy , for any online decision problem .
we also specify the ( worst - case ) convergence speed , which is very slow .
* keywords . *
prediction with expert advice , responsive environments , partial observation game , bandits , universal learning , asymptotic optimality .
tcs - tr - a-05 - 4 july 2005 idsia-15 - 05 |
in the past few years , the synthesis of ferromagnetic semiconductors has become a major challenge for spintronics .
actually , growing a magnetic and semiconducting material could lead to promising advances like spin injection into non magnetic semiconductors , or electrical manipulation of carrier induced magnetism in magnetic semiconductors @xcite . up to now , major efforts have focused on diluted magnetic semiconductors ( dms ) in which the host semiconducting matrix is randomly substituted by transition metal ( tm ) ions such as mn , cr , ni , fe or co @xcite .
however curie temperatures ( @xmath10 ) in dms remain rather low and tm concentrations must be drastically raised in order to increase @xmath10 up to room temperature .
that usually leads to phase separation and the formation of secondary phases .
it was recently shown that phase separation induced by spinodal decomposition could lead to a significant increase of @xmath10 @xcite .
for semiconductors showing @xmath10 higher than room temperature one can foresee the fabrication of nanodevices such as memory nanodots , or nanochannels for spin injection .
therefore , the precise control of inhomogeneities appears as a new challenge which may open a way to industrial applications of ferromagnetism in semiconductors .
the increasing interest in group - iv magnetic semiconductors can also be explained by their potential compatibility with the existing silicon technology . in 2002 , carrier mediated ferromagnetism was reported in mbe grown ge@xmath17mn@xmath18 films by park _
the maximum critical temperature was 116 k. recently many publications indicate a significant increase of @xmath10 in ge@xmath0mn@xmath1 material depending on growth conditions @xcite .
et al . _ reported a curie temperature as high as 285 k @xcite .
taking into account the strong tendency of mn ions to form intermetallic compounds in germanium , a detailed investigation of the nanoscale structure is required . up to now
, only a few studies have focused on the nanoscale composition in ge@xmath0mn@xmath1 films .
local chemical inhomogeneities have been recently reported by kang _
@xcite who evidenced a micrometer scale segregation of manganese in large mn rich stripes .
ge@xmath6mn@xmath7 as well as ge@xmath19mn@xmath20 clusters embedded in a germanium matrix have been reported by many authors . however
, curie temperatures never exceed 300 k @xcite .
ge@xmath6mn@xmath7 clusters exhibit a curie temperature of 296 k @xcite .
this phase frequently observed in ge@xmath0mn@xmath1 films is the most stable ( ge , mn ) alloy .
the other stable compound ge@xmath19mn@xmath20 has also been observed in nanocrystallites surrounded with pure germanium @xcite .
ge@xmath19mn@xmath20 and ge@xmath6mn@xmath7 phases are ferromagnetic but their metallic character considerably complicates their potential use as spin injectors .
recently , some new mn - rich nanostructures have been evidenced in ge@xmath0mn@xmath1 layers .
@xcite reported the formation of high mn content ( between 10 % and 20 % of mn ) amorphous ge@xmath0mn@xmath21 precipitates in a mn - free germanium matrix .
mn - rich coherent cubic clusters were observed by ahlers _
@xcite which exhibit a curie temperatures below 200 k. finally , high-@xmath10
( @xmath22 400 k ) mn - rich nanocolumns have been evidenced @xcite which could lead to silicon compatible room temperature operational devices . + in the present paper , we investigate the structural and magnetic properties of ge@xmath0mn@xmath21 thin films for low growth temperatures ( @xmath23 200@xmath4c ) and low mn concentrations ( between 1 % and 11 % ) . by combining tem , x - ray diffraction and squid magnetometry , we could identify different magnetic phases .
we show that depending on growth conditions , we obtain either mn - rich nanocolumns or ge@xmath15mn@xmath16 clusters embedded in a germanium matrix .
we discuss the structural and magnetic properties of these nanostructures as a function of manganese concentration and growth temperature .
we also discuss the magnetic anisotropy of nanocolumns and ge@xmath6mn@xmath7 clusters .
growth was performed using solid sources molecular beam epitaxy ( mbe ) by co - depositing ge and mn evaporated from standard knudsen effusion cells .
deposition rate was low ( @xmath24 0.2 .s@xmath25 ) .
germanium substrates were epi - ready ge(001 ) wafers with a residual n - type doping and resistivity of 10@xmath26 @xmath27 and 5 @xmath28 respectively .
after thermal desorption of the surface oxide , a 40 nm thick ge buffer layer was grown at 250@xmath4c , resulting in a 2 @xmath29 1 surface reconstruction as observed by reflection high energy electron diffraction ( rheed ) ( see fig .
next , 80 nm thick ge@xmath0mn@xmath1 films were subsequently grown at low substrate temperature ( from 80@xmath4c to 200@xmath4c ) .
mn content has been determined by x - ray fluorescence measurements performed on thick samples ( @xmath24 1 @xmath30 thick ) and complementary rutherford back scattering ( rbs ) on thin ge@xmath0mn@xmath1 films grown on silicon .
mn concentrations range from 1 % to 11% mn . for ge@xmath0mn@xmath1 films grown at substrate temperatures below 180@xmath4c , after the first monolayer ( ml ) deposition
, the 2 @xmath29 1 surface reconstruction almost totally disappears .
after depositing few mls , a slightly diffuse 1 @xmath29 1 streaky rheed pattern and a very weak 2 @xmath29 1 reconstruction ( fig .
1b ) indicate a predominantly two - dimensional growth . for growth temperatures above 180@xmath4c additional spots
appear in the rheed pattern during the ge@xmath0mn@xmath1 growth ( fig .
these spots may correspond to the formation of very small secondary phase crystallites .
the nature of these crystallites will be discussed below .
transmission electron microscopy ( tem ) observations were performed using a jeol 4000ex microscope with an acceleration voltage of 400 kv .
energy filtered transmission electron microscopy ( eftem ) was done using a jeol 3010 microscope equipped with a gatan image filter .
sample preparation was carried out by standard mechanical polishing and argon ion milling for cross - section investigations and plane views were prepared by wet etching with h@xmath6po@xmath31-h@xmath32o@xmath32 solution @xcite .
mn@xmath1 films : ( a ) 2 @xmath29 1 surface reconstruction of the germanium buffer layer .
( b ) 1 @xmath29 1 streaky rheed pattern obtained at low growth temperatures ( @xmath33180@xmath4c ) .
( c ) rheed pattern of a sample grown at @xmath34180@xmath4c .
the additional spots reveal the presence of ge@xmath6mn@xmath7 clusters at the surface of the film.,title="fig : " ] mn@xmath1 films : ( a ) 2 @xmath29 1 surface reconstruction of the germanium buffer layer .
( b ) 1 @xmath29 1 streaky rheed pattern obtained at low growth temperatures ( @xmath33180@xmath4c ) .
( c ) rheed pattern of a sample grown at @xmath34180@xmath4c .
the additional spots reveal the presence of ge@xmath6mn@xmath7 clusters at the surface of the film.,title="fig : " ] mn@xmath1 films : ( a ) 2 @xmath29 1 surface reconstruction of the germanium buffer layer .
( b ) 1 @xmath29 1 streaky rheed pattern obtained at low growth temperatures ( @xmath33180@xmath4c ) .
( c ) rheed pattern of a sample grown at @xmath34180@xmath4c .
the additional spots reveal the presence of ge@xmath6mn@xmath7 clusters at the surface of the film.,title="fig : " ]
mn@xmath1 film grown at 130@xmath4c and containing 6 % of manganese . ( a ) cross - section along the [ 110 ] axis : we clearly see the presence of nanocolumns elongated along the growth axis .
( b ) high resolution image of the interface between the ge@xmath0mn@xmath1 film and the ge buffer layer .
the ge@xmath0mn@xmath1 film exhibits the same diamond structure as pure germanium .
no defect can be seen which could be caused by the presence of nanocolumns .
( c ) plane view micrograph performed on the same sample confirms the columnar structure and gives the density and size distribution of nanocolumns .
( d ) mn chemical map obtained by energy filtered transmission electron microcopy ( eftem ) .
the background was carefully substracted from pre - edge images .
bright areas correspond to mn - rich regions.,title="fig : " ] mn@xmath1 film grown at 130@xmath4c and containing 6 % of manganese .
( a ) cross - section along the [ 110 ] axis : we clearly see the presence of nanocolumns elongated along the growth axis .
( b ) high resolution image of the interface between the ge@xmath0mn@xmath1 film and the ge buffer layer .
the ge@xmath0mn@xmath1 film exhibits the same diamond structure as pure germanium .
no defect can be seen which could be caused by the presence of nanocolumns .
( c ) plane view micrograph performed on the same sample confirms the columnar structure and gives the density and size distribution of nanocolumns .
( d ) mn chemical map obtained by energy filtered transmission electron microcopy ( eftem ) .
the background was carefully substracted from pre - edge images .
bright areas correspond to mn - rich regions.,title="fig : " ] mn@xmath1 film grown at 130@xmath4c and containing 6 % of manganese .
( a ) cross - section along the [ 110 ] axis : we clearly see the presence of nanocolumns elongated along the growth axis .
( b ) high resolution image of the interface between the ge@xmath0mn@xmath1 film and the ge buffer layer .
the ge@xmath0mn@xmath1 film exhibits the same diamond structure as pure germanium .
no defect can be seen which could be caused by the presence of nanocolumns .
( c ) plane view micrograph performed on the same sample confirms the columnar structure and gives the density and size distribution of nanocolumns .
( d ) mn chemical map obtained by energy filtered transmission electron microcopy ( eftem ) .
the background was carefully substracted from pre - edge images .
bright areas correspond to mn - rich regions.,title="fig : " ] mn@xmath1 film grown at 130@xmath4c and containing 6 % of manganese .
( a ) cross - section along the [ 110 ] axis : we clearly see the presence of nanocolumns elongated along the growth axis .
( b ) high resolution image of the interface between the ge@xmath0mn@xmath1 film and the ge buffer layer .
the ge@xmath0mn@xmath1 film exhibits the same diamond structure as pure germanium .
no defect can be seen which could be caused by the presence of nanocolumns .
( c ) plane view micrograph performed on the same sample confirms the columnar structure and gives the density and size distribution of nanocolumns .
( d ) mn chemical map obtained by energy filtered transmission electron microcopy ( eftem ) .
the background was carefully substracted from pre - edge images .
bright areas correspond to mn - rich regions.,title="fig : " ] in samples grown at 130@xmath4c and containing 6 % mn , we can observe vertical elongated nanostructures _
i.e. _ nanocolumns as shown in fig
nanocolumns extend through the whole ge@xmath0mn@xmath1 film thickness . from the high resolution
tem image shown in fig . 2b
, we deduce their average diameter around 3 nm .
moreover in fig .
2b , the interface between the ge buffer layer and the ge@xmath0mn@xmath1 film is flat and no defect propagates from the interface into the film .
the ge@xmath0mn@xmath1 film is a perfect single crystal in epitaxial relationship with the substrate . in fig .
2c is shown a plane view micrograph of the same sample confirming the presence of nanocolumns in the film . from this image
, we can deduce the size and density of nanocolumns .
the nanocolumns density is 13000 @xmath35 with a mean diameter of 3 nm which is coherent with cross - section measurements . in order to estimate the chemical composition of these nanocolumns
, we further performed chemical mapping using eftem . in fig .
2d we show a cross sectional mn chemical map of the ge@xmath0mn@xmath1 film .
this map shows that the formation of nanocolumns is a consequence of mn segregation .
nanocolumns are mn rich and the surrounding matrix is mn poor .
however , it is impossible to deduce the mn concentration in ge@xmath0mn@xmath1 nanocolumns from this cross section .
indeed , in cross section observations , the columns diameter is much smaller than the probed film thickness and the signal comes from the superposititon of the ge matrix and mn - rich nanocolumns . in order to quantify mn concentration inside the nanocolumns and inside the ge matrix , eels measurements ( not shown here )
have been performed in a plane view geometry @xcite .
these observations revealed that the matrix mn content is below 1 % ( detection limit of our instrument ) . measuring the surface occupied by the matrix and the nanocolumns in plane view tem images , and considering the average mn concentration in the sample ( 6 % )
, we can estimate the mn concentration in the nanocolumns .
the mn concentration measured by eels being between 0% and 1% , we can conclude that the mn content in the nanocolumns is between 30 % and 38 % . + for samples grown between 80@xmath36c and 150@xmath36c cross section and plane view tem observations reveal the presence of mn rich nanocolumns surrounded with a mn poor ge matrix . in order to investigate the influence of mn concentration on the structural properties of ge@xmath0mn@xmath1 films , ten samples
have been grown at 100@xmath36c and at 150@xmath36c with mn concentrations of 1.3 % , 2.3 % , 4 % , 7 % and 11.3 % .
their structural properties have been investigated by plane view tem observations .
c and 150@xmath4c respectively .
( a ) mn concentration dependence of the size distribution .
( b ) columns density as a function of mn concentration .
( c ) volume fraction of the nanocolumns as a function of mn concentration.,title="fig : " ] c and 150@xmath4c respectively .
( a ) mn concentration dependence of the size distribution .
( b ) columns density as a function of mn concentration .
( c ) volume fraction of the nanocolumns as a function of mn concentration.,title="fig : " ] c and 150@xmath4c respectively .
( a ) mn concentration dependence of the size distribution .
( b ) columns density as a function of mn concentration .
( c ) volume fraction of the nanocolumns as a function of mn concentration.,title="fig : " ] for samples grown at 100@xmath36c with mn concentrations below 5 %
the nanocolumns mean diameter is [email protected] nm .
the evolution of columns density as a fonction of mn concentration is reported in figure 3b . by increasing the mn concentration from 1.3 % to 4 %
we observe a significant increase of the columns density from 13000 to 30000 @xmath35 . for mn concentrations higher than 5 %
the density seems to reach a plateau corresponding to 35000 @xmath35 and their diameter slightly increases from 1.8 nm at 4 % to 2.8 nm at 11.3 % . by plotting the volume fraction occupied by the columns in the film as a function of mn concentration , we observe a linear dependence for mn contents below 5 % .
the non - linear behavior above 5 % may indicate that the mechanism of mn incorporation is different in this concentration range , leading to an increase of mn concentration in the columns or in the matrix .
for samples grown at 100@xmath36c , nanocolumns are always fully coherent with the surrounding matrix ( fig .
4a ) . increasing
the mn content in the samples grown at 150@xmath36c from 1.3 % to 11.3 % leads to a decrease of the columns density ( fig 3b ) . moreover
, their average diameter increases significantly and size distributions become very broad ( see fig .
3a ) . for the highest mn concentration ( 11.3 % )
we observe the coexistence of very small columns with a diameter of 2.5 nm and very large columns with a diameter of 9 nm . in samples
grown at 150@xmath36c containing 11.3 % of mn , the crystalline structure of nanocolumns is also highly modified . in plane view tem micrographs , one can see columns exhibiting several different crystalline structures . we still observe some columns which are fully coherent with the ge matrix like in the samples grown at lower temperature .
nevertheless , observations performed on these samples grown at 150@xmath36c and with 11.3% mn reveal some uniaxially @xcite or fully relaxed columns exhibiting a misfit of 4 % between the matrix and the columns and leading to misfit dislocations at the interface between the column and the matrix ( see fig .
thus we can conclude that coherent columns are probably in strong compression and the surrounding matrix in tension .
on the same samples ( t@xmath38=150@xmath4c , 11.3% mn ) , we also observe a large number of highly disordered nanocolumns leading to an amorphous like tem contrast(fig .
4c ) . c. the crystal structure is exactly the same as germanium .
( b ) partially relaxed nanocolumn .
one can see dislocations at the interface between the columns and the matrix leading to stress relaxation .
( c ) amorphous nanocolumn .
these columns are typical in samples grown at 150@xmath4c with high mn contents.,title="fig : " ] c. the crystal structure is exactly the same as germanium .
( b ) partially relaxed nanocolumn .
one can see dislocations at the interface between the columns and the matrix leading to stress relaxation .
( c ) amorphous nanocolumn .
these columns are typical in samples grown at 150@xmath4c with high mn contents.,title="fig : " ] c. the crystal structure is exactly the same as germanium .
( b ) partially relaxed nanocolumn .
one can see dislocations at the interface between the columns and the matrix leading to stress relaxation .
( c ) amorphous nanocolumn .
these columns are typical in samples grown at 150@xmath4c with high mn contents.,title="fig : " ] in conclusion , we have evidenced a complex mechanism of mn incorporation in mn doped ge films grown at low temperature .
in particular mn incorporation is highly inhomogeneous .
for very low growth temperatures ( below 120@xmath36c ) the diffusion of mn atoms leads to the formation of mn rich , vertical nanocolumns .
their density mostly depends on mn concentration and their mean diameter is about 2 nm . these results can be compared with the theoretical predictions of fukushima _ et al .
_ @xcite : they proposed a model of spinodal decomposition in ( ga , mn)n and ( zn , cr)te based on layer by layer growth conditions and a strong pair attraction between mn atoms which leads to the formation of nanocolumns .
this model may also properly describe the formation of mn rich nanocolumns in our samples .
layer by layer growth conditions can be deduced from rheed pattern evolution during growth .
for all the samples grown at low temperature , rheed observations clearly indicate two - dimensional growth .
moreover , ge / ge@xmath0mn@xmath1/ge heterostructures have been grown and observed by tem ( see fig .
ge@xmath0mn@xmath1/ge ( as well as ge / ge@xmath0mn@xmath1 ) interfaces are very flat and sharp thus confirming a two - dimensional , layer by layer growth mode . therefore we can assume that the formation of mn rich nanocolumns is a consequence of 2d - spinodal decomposition .
mn@xmath1/ge / ge@xmath0mn@xmath1/ge heterostructure .
this sample has been grown at 130 @xmath4c with 6% mn .
ge@xmath0mn@xmath1 layers are 15 nm thick and ge spacers 5 nm thick .
we clearly see the sharpness of both ge@xmath0mn@xmath1/ge and ge / ge@xmath0mn@xmath1 interfaces .
mn segregation leading to the columns formation already takes place in very thin ge@xmath0mn@xmath1 films . ] for growth temperatures higher than 160@xmath36c , cross section tem and eftem observations ( not shown here ) reveal the coexistence of two mn - rich phases : nanocolumns and ge@xmath15mn@xmath16 nanoclusters embedded in the germanium matrix .
a typical high resolution tem image is shown in figure 6 .
ge@xmath15mn@xmath16 clusters are not visible in rheed patterns for temperatures below 180@xmath36c .
to investigate the nature of these clusters , we performed x - ray diffraction in @xmath39 mode .
diffraction scans were acquired on a high resolution diffractometer using the copper k@xmath40 radiation and on the gmt station of the bm32 beamline at the european synchrotron radiation facility ( esrf ) .
three samples grown at different temperatures and/or annealed at high temperature were investigated .
the two first samples are ge@xmath0mn@xmath1 films grown at 130@xmath36c and 170@xmath36c respectively .
the third one has been grown at 130@xmath36c and post - growth annealed at 650@xmath36c . by analysing x - ray diffraction spectra , we can evidence two different crystalline structures . for
the sample grown at 130@xmath36c , the @xmath39 scan only reveals the ( 004 ) bragg peak of the germanium crystal , confirming the good epitaxial relationship between the layer and the substrate , and the absence of secondary phases in the film in spite of a high dynamics of the order of 10@xmath41 . for both samples grown at 170@xmath36c and
annealed at 650@xmath36c , @xmath39 spectra are identical .
in addition to the ( 004 ) peak of germanium , we observe three additional weak peaks .
the first one corresponds to the ( 002 ) germanium forbidden peak which probably comes from a small distortion of the germanium crystal , and the two other peaks are respectively attributed to the ( 002 ) and ( 004 ) bragg peaks of a secondary phase .
the @xmath42 lattice parameter of ge@xmath6mn@xmath7 hexagonal crystal is 5.053 @xcite which is in very good agreement with the values obtained from diffraction data for both ( 002 ) and ( 004 ) lines assuming that the @xmath42 axis of ge@xmath6mn@xmath7 is along the [ 001 ] direction of the ge substrate . c. we observe the coexistence of two different mn - rich phases : ge@xmath0mn@xmath1 nanocolumns and ge@xmath6mn@xmath7 clusters . ] in summary , in a wide range of growth temperatures and mn concentrations , we have evidenced a two - dimensional spinodal decomposition leading to the formation of mn - rich nanocolumns in ge@xmath0mn@xmath1 films .
this decomposition is probably the consequence of : @xmath8 a strong pair attraction between mn atoms , @xmath9 a strong surface diffusion of mn atoms in germanium even at low growth temperatures and @xmath12 layer by layer growth conditions .
we have also investigated the influence of growth parameters on the spinodal decomposition : at low growth temperatures ( 100@xmath4c ) , increasing the mn content leads to higher columns densities while at higher growth temperatures ( 150@xmath4c ) , the columns density remains nearly constant whereas their size increases drastically . by plotting the nanocolumns density as a function of mn content
, we have shown that the mechanism of mn incorporation in ge changes above 5 % of mn . finally , using tem observations and x - ray diffraction ,
we have shown that ge@xmath6mn@xmath7 nanoclusters start to form at growth temperatures higher than 160@xmath36c .
we have thoroughly investigated the magnetic properties of thin ge@xmath0mn@xmath1 films for different growth temperatures and mn concentrations . in this section ,
we focus on mn concentrations between 2 % and 11 % .
we could clearly identify four different magnetic phases in ge@xmath0mn@xmath1 films : diluted mn atoms in the germanium matrix , low @xmath10 nanocolumns ( @xmath10 @xmath11 170 k ) , high @xmath10 nanocolumns ( @xmath10 @xmath13 400 k ) and ge@xmath15mn@xmath16 clusters ( @xmath10 @xmath43 300 k ) .
the relative weight of each phase clearly depends on the growth temperature and to a lesser extend on mn concentration . for low growth temperature ( @xmath23 120@xmath4c ) ,
we show that nanocolumns are actually made of four uncorrelated superparamagnetic nanostructures . increasing t@xmath44 above 120@xmath4c
, we first obtain continuous columns exhibiting low @xmath10 ( @xmath23 170 k ) and high @xmath10 ( @xmath22 400 k ) for @xmath45130@xmath4c .
the larger columns become ferromagnetic _ i.e. _ @xmath46 .
meanwhile ge@xmath15mn@xmath16 clusters start to form .
finally for higher @xmath2 , the magnetic contribution from ge@xmath15mn@xmath16 clusters keeps increasing while the nanocolumns signal progressively disappears .
/mn ) of ge@xmath47mn@xmath48 samples for different growth temperatures .
the magnetic field is applied in the film plane .
the inset shows the temperature dependence of a sample grown at 130@xmath4c and annealed at 650@xmath4c for 15 minutes .
after annealing , the magnetic signal mostly arises from ge@xmath15mn@xmath16 clusters .
( b ) zfc - fc measurements performed on ge@xmath47mn@xmath48 samples for different growth temperatures .
the in - plane applied field is 0.015 t. the zfc peak at low temperature ( @xmath11150 k ) can be attributed to the superparamagnetic nanocolumns .
this peak widens and shifts towards high blocking temperatures when increasing growth temperature .
the second peak above 150 k in the zfc curve which increases with increasing growth temperature is attributed to superparamagnetic ge@xmath15mn@xmath16 clusters .
the increasing zfc - fc irreversibility at @xmath24 300 k is due to the increasing contribution from large ferromagnetic ge@xmath15mn@xmath16 clusters .
the nanocolumns signal completely vanishes after annealing at 650@xmath4c for 15 minutes.,title="fig : " ] /mn ) of ge@xmath47mn@xmath48 samples for different growth temperatures .
the magnetic field is applied in the film plane .
the inset shows the temperature dependence of a sample grown at 130@xmath4c and annealed at 650@xmath4c for 15 minutes .
after annealing , the magnetic signal mostly arises from ge@xmath15mn@xmath16 clusters .
( b ) zfc - fc measurements performed on ge@xmath47mn@xmath48 samples for different growth temperatures .
the in - plane applied field is 0.015 t. the zfc peak at low temperature ( @xmath11150 k ) can be attributed to the superparamagnetic nanocolumns .
this peak widens and shifts towards high blocking temperatures when increasing growth temperature .
the second peak above 150 k in the zfc curve which increases with increasing growth temperature is attributed to superparamagnetic ge@xmath15mn@xmath16 clusters .
the increasing zfc - fc irreversibility at @xmath24 300 k is due to the increasing contribution from large ferromagnetic ge@xmath15mn@xmath16 clusters .
the nanocolumns signal completely vanishes after annealing at 650@xmath4c for 15 minutes.,title="fig : " ] in fig .
7a , the saturation magnetization at 2 tesla in @xmath49/mn of ge@xmath0mn@xmath1 films with 7 % of mn is plotted as a function of temperature for different growth temperatures ranging from @xmath2=90@xmath4c up to 160@xmath4c .
the inset shows the temperature dependence of the magnetization at 2 tesla after annealing at 650@xmath4c during 15 minutes .
figure 7b displays the corresponding zero field cooled - field cooled ( zfc - fc ) curves recorded at 0.015 tesla . in the zfc - fc procedure , the sample is first cooled down to 5 k in zero magnetic field and the susceptibility is subsequently recorded at 0.015 tesla while increasing the temperature up to 400 k ( zfc curve ) .
then , the susceptibility is recorded under the same magnetic field while decreasing the temperature down to 5 k ( fc curve ) .
three different regimes can be clearly distinguished .
+ for @xmath50120@xmath4c , the temperature dependence of the saturation magnetization remains nearly the same while increasing growth temperature .
the overall magnetic signal vanishing above 200 k is attributed to the nanocolumns whereas the increasing signal below 50 k originates from diluted mn atoms in the surrounding matrix .
the mn concentration dependence of the saturation magnetization is displayed in figure 8 .
for the lowest mn concentration ( 4 % ) , the contribution from diluted mn atoms is very high and drops sharply for higher mn concentrations ( 7 % , 9 % and 11.3 % ) .
therefore the fraction of mn atoms in the diluted matrix decreases with mn concentration probably because mn atoms are more and more incorporated in the nanocolumns . in parallel , the curie temperature of nanocolumns increases with the mn concentration reaching 170 k for 11.3 % of mn .
this behavior may be related to different mn compositions and to the increasing diameter of nanocolumns ( from 1.8 nm to 2.8 nm ) as discussed in section [ structural ] .
/mn ) of ge@xmath0mn@xmath1 films grown at 100@xmath4c plotted for different mn concentrations : 4.1 % ; 7 % ; 8.9 % and 11.3 % . ]
zfc - fc measurements show that the nanocolumns are superparamagnetic .
the magnetic signal from the diluted mn atoms in the matrix is too weak to be detected in susceptibility measurements at low temperature . in samples containing 4 % of mn , zfc and fc curves
superimpose down to low temperatures . as we do not observe hysteresis loops at low temperature , we believe that at this mn concentration nanocolumns are superparamagnetic in the whole temperature range and the blocking temperature can not be measured . for higher mn contents
, the zfc curve exhibits a very narrow peak with a maximum at the blocking temperature of 15 k whatever the mn concentration and growth temperature ( see fig .
therefore the anisotropy barrier distribution is narrow and assuming that nanocolumns have the same magnetic anisotropy , this is a consequence of the very narrow size distribution of the nanocolumns as observed by tem . to probe the anisotropy barrier distribution ,
we have performed zfc - fc measurements but instead of warming the sample up to 400 k , we stopped at a lower temperature @xmath51 . ) of superparamagnetic nanostructures .
if magnetic anisotropy does not depend on the particle size , this distribution exactly reflects their magnetic size distribution . in this drawing the blocking temperature ( @xmath52 ) corresponds to the distribution maximum . at a given temperature @xmath51 such that 25@xmath53 falls into the anisotropy barrier distribution , the largest nanostructures with an anisotropy energy larger than 25@xmath53
are blocked whereas the others are superparamagnetic . ]
if this temperature falls into the anisotropy barrier distribution as depicted in fig .
9 , the fc curve deviates from the zfc curve .
indeed the smallest nanostructures have become superparamagnetic at @xmath51 and when decreasing again the temperature , their magnetization freezes along a direction close to the magnetic field and the fc susceptibility is higher than the zfc susceptibility .
therefore any irreversibility in this procedure points at the presence of superparamagnetic nanostructures .
the results are given in fig .
10a . zfc and fc curves clearly superimpose up to @xmath51=250 k thus the nanocolumns are superparamagnetic up to their curie temperature and no ge@xmath15mn@xmath16 clusters could be detected .
moreover for low @xmath51 values , a peak appears at low temperature in fc curves which evidences strong antiferromagnetic interactions between the nanocolumns @xcite .
mn@xmath54 sample grown at 115@xmath4c .
the in - plane applied field is 0.015 t. magnetization was recorded up to different t@xmath55 temperatures : 30 k , 50 k , 100 k , 150 k , 200 k and 250 k. curves are shifted up for more clarity .
( b ) zfc - fc curves for in - plane and out - of - plane applied fields ( 0.015 t).,title="fig : " ] mn@xmath54 sample grown at 115@xmath4c .
the in - plane applied field is 0.015 t. magnetization was recorded up to different t@xmath55 temperatures : 30 k , 50 k , 100 k , 150 k , 200 k and 250 k. curves are shifted up for more clarity .
( b ) zfc - fc curves for in - plane and out - of - plane applied fields ( 0.015 t).,title="fig : " ] in order to derive the magnetic size and anisotropy of the mn - rich nanocolumns embedded in the ge matrix , we have fitted the inverse normalized in - plane ( resp .
out - of - plane ) susceptibility : @xmath56 ( resp .
@xmath57 ) .
the corresponding experimental zfc - fc curves are reported in fig .
since susceptibility measurements are performed at low field ( 0.015 t ) , the matrix magnetic signal remains negligible . in order to normalize susceptibility data ,
we need to divide the magnetic moment by the saturated magnetic moment recorded at 5 t. however the matrix magnetic signal becomes very strong at 5 t and low temperature so that we need to subtract it from the saturated magnetic moment using a simple curie function . from fig .
10b , we can conclude that nanocolumns are isotropic . therefore to fit experimental data we use the following expression well suited for isotropic systems or cubic anisotropy : @xmath58 .
@xmath59 is the boltzmann constant , @xmath60 is the magnetic moment of a single - domain nanostructure ( macrospin approximation ) where @xmath61 is its magnetization and @xmath62 its volume . the in - plane magnetic field is applied along @xmath63 $ ] or @xmath64 $ ] crystal axes .
since the nanostructures curie temperature does not exceed 170 k , the temperature dependence of the saturation magnetization is also accounted for by writting @xmath65 .
antiferromagnetic interactions between nanostructures are also considered by adding an effective field estimated in the mean field approximation @xcite : @xmath66 .
the only fitting parameters are the maximum magnetic moment ( _ i.e. _ at low temperature ) per nanostructure : @xmath67 ( in bohr magnetons @xmath49 ) and the maximum interaction field ( _ i.e. _ at low temperature ) : @xmath68 .
mn@xmath54 sample grown at 115@xmath4c .
fits were performed assuming isotropic nanostructures or cubic anisotropy .
dashed line is for in - plane susceptibility and solid line for out - of - plane susceptibility . ] in fig .
11 , the best fits lead to @xmath691250 @xmath49 and @xmath70102 mt for in - plane susceptibility and @xmath691600 @xmath49 and @xmath7098 mt for out - of - plane susceptibility .
it gives an average magnetic moment of 1425 @xmath49 per column and an effective interaction field of 100 mt . using this magnetic moment and its temperature dependence ,
magnetization curves could be fitted using a langevin function and @xmath71 curves superimpose for @xmath72100 k. however , from the saturated magnetic moment of the columns and their density ( 35000 @xmath35 ) , we find almost 6000 @xmath49 per column .
therefore , for low growth temperatures , we need to assume that nanocolumns are actually made of almost four independent elongated magnetic nanostructures .
the effective field for antiferromagnetic interactions between nanostructures estimated from the susceptibility fits is at least one order of magnitude larger than what is expected from pure magnetostatic coupling .
this difference may be due to either an additional antiferromagnetic coupling through the matrix which origin remains unexplained or to the mean field approximation which is no more valid in this strong coupling regime .
as for magnetic anisotropy , the nanostructures behave as isotropic magnetic systems or exhibit a cubic magnetic anisotropy .
first we can confirm that nanostructures are not amorphous otherwise shape anisotropy would dominate leading to out - of - plane anisotropy .
we can also rule out a random distribution of magnetic easy axes since the nanostructures are clearly crystallized in the diamond structure and would exhibit at least a cubic anisotropy ( except if the random distribution of mn atoms within the nanostructures can yield random easy axes ) . since the nanostructures are in strong in - plane compression ( their lattice parameter is larger than the matrix one ) , the cubic symmetry of the diamond structure is broken and magnetic cubic anisotropy is thus unlikely .
we rather believe that out - of - plane shape anisotropy is nearly compensated by in - plane magnetoelastic anisotropy due to compression leading to a _ pseudo _
cubic anisotropy . from the blocking temperature ( 15 k ) and the magnetic volume of the nanostructures
, we can derive their magnetic anisotropy constant using @xmath73 : k@xmath2410 kj.m@xmath74 which is of the same order of magnitude as shape anisotropy .
mn@xmath48 sample grown at 122@xmath4c .
the in - plane applied field is 0.015 t. magnetization was recorded up to different t@xmath55 temperatures : 50 k , 100 k , 150 k , 200 k and 250 k. curves are shifted up for more clarity .
( b ) zfc - fc curves for in - plane and out - of - plane applied fields ( 0.015 t).,title="fig : " ] mn@xmath48 sample grown at 122@xmath4c .
the in - plane applied field is 0.015 t. magnetization was recorded up to different t@xmath55 temperatures : 50 k , 100 k , 150 k , 200 k and 250 k. curves are shifted up for more clarity .
( b ) zfc - fc curves for in - plane and out - of - plane applied fields ( 0.015 t).,title="fig : " ] for growth temperatures @xmath75120@xmath4c and mn concentrations @xmath13 7 % , samples exhibit a magnetic signal above 200 k corresponding to ge@xmath15mn@xmath16 clusters ( see fig .
7a ) . as we can see , squid measurements are much more sensitive to the presence of ge@xmath15mn@xmath16 clusters , even at low concentration , than tem and x - ray diffraction used in section [ structural ] .
we also observe a sharp transition in the zfc curve ( see fig .
12a and 12b ) : the peak becomes very large and is shifted towards high blocking temperatures ( the signal is maximum at @xmath7623 k ) .
this can be easily understood as a magnetic percolation of the four independent nanostructures obtained at low growth temperatures into a single magnetic nanocolumn .
therefore the magnetic volume increases sharply as well as blocking temperatures . at the same time
, the size distribution widens as observed in tem . in fig .
12a , we have performed zfc - fc measurements at different @xmath51 temperatures . the zfc - fc irreversibility is observed up to the curie temperature of @xmath24120 k meaning that a fraction of nanocolumns is ferromagnetic ( _ i.e. _ @xmath77 ) . in fig .
12b , in - plane and out - of - plane zfc curves nearly superimpose for @xmath78150 k due to the isotropic magnetic behavior of the nanocolumns : in - plane magnetoelastic anisotropy is still compensating out - of - plane shape anisotropy
. moreover the magnetic signal above 150 k corresponding to ge@xmath15mn@xmath16 clusters that start to form in this growth temperature range is strongly anisotropic .
this perpendicular anisotropy confirms the epitaxial relation : ( 0002 ) ge@xmath15mn@xmath16 @xmath79 ( 002 ) ge discussed in ref.@xcite .
the magnetic easy axis of the clusters lies along the hexagonal @xmath42-axis which is perpendicular to the film plane .
mn@xmath54 sample grown at 145@xmath4c .
the in - plane applied field is 0.015 t. magnetization was recorded up to different t@xmath55 temperatures : 50 k , 100 k , 150 k , 200 k , 250 k and 300 k. curves are shifted up for more clarity .
( b ) zfc - fc curves for in - plane and out - of - plane applied fields ( 0.015 t).,title="fig : " ] mn@xmath54 sample grown at 145@xmath4c .
the in - plane applied field is 0.015 t. magnetization was recorded up to different t@xmath55 temperatures : 50 k , 100 k , 150 k , 200 k , 250 k and 300 k. curves are shifted up for more clarity .
( b ) zfc - fc curves for in - plane and out - of - plane applied fields ( 0.015 t).,title="fig : " ] for growth temperatures @xmath75145@xmath4c the cluster magnetic signal dominates ( fig .
superparamagnetic nanostructures are investigated performing zfc - fc measurements at different @xmath51 temperatures ( fig .
the first zfc peak at low temperature _
i.e. _ @xmath11 150 k is attributed to low-@xmath10 nanocolumns ( @xmath80130 k ) .
this peak is wider than for lower growth temperatures and its maximum is further shifted up to 30 k. these results are in agreement with tem observations : increasing @xmath2 leads to larger nanocolumns ( _ i.e. _ higher blocking temperatures ) and wider size distributions .
zfc - fc irreversibility is observed up to the curie temperature due to the presence of ferromagnetic columns .
the second peak above 180 k in the zfc curve is attributed to ge@xmath15mn@xmath16 clusters and the corresponding zfc - fc irreversibility persisting up to 300 k means that some clusters are ferromagnetic .
we clearly evidence the out - of - plane anisotropy of ge@xmath15mn@xmath16 clusters and the isotropic magnetic behavior of nanocolumns ( fig .
13b ) . in this growth temperature range
, we have also investigated the mn concentration dependence of magnetic properties .
/mn ) of ge@xmath0mn@xmath1 films grown at 150@xmath4c plotted for different mn concentrations : 2.3 % ; 4 % ; 7 % ; 9 % ; 11.3 % .
( b ) zfc - fc measurements performed on ge@xmath0mn@xmath1 films grown at 150@xmath4c .
the in - plane applied field is 0.025 t for 2.3 % and 4 % and 0.015 t for 8 % and 11.3 % .
, title="fig : " ] /mn ) of ge@xmath0mn@xmath1 films grown at 150@xmath4c plotted for different mn concentrations : 2.3 % ; 4 % ; 7 % ; 9 % ; 11.3 % .
( b ) zfc - fc measurements performed on ge@xmath0mn@xmath1 films grown at 150@xmath4c .
the in - plane applied field is 0.025 t for 2.3 % and 4 % and 0.015 t for 8 % and 11.3 % .
, title="fig : " ] in fig .
14a , for low mn concentrations ( 2.3 % and 4 % ) the contribution from diluted mn atoms in the germanium matrix to the saturation magnetization is very high and nearly vanishes for higher mn concentrations ( 7 % , 9 % and 13 % ) as observed for low growth temperatures . above 7 % , the magnetic signal mainly comes from nanocolumns and ge@xmath15mn@xmath16 clusters .
we can derive more information from zfc - fc measurements ( fig .
indeed , for 2.3 % of mn , zfc and fc curves nearly superimpose down to low temperature meaning that nanocolumns are superparamagnetic in the whole temperature range .
moreover the weak irreversibility arising at 300 k means that some ge@xmath15mn@xmath16 clusters have already formed in the samples even at very low mn concentrations . for 4 % of mn
, we can observe a peak with a maximum at the blocking temperature ( 12 k ) in the zfc curve .
we can also derive the curie temperature of nanocolumns : @xmath2445 k. the irresversibility arising at 300 k still comes from ge@xmath15mn@xmath16 clusters .
increasing the mn concentration above 7 % leads to : higher blocking temperatures ( 20 k and 30 k ) due to larger nanocolumns and wider zfc peaks due to wider size distributions in agreement with tem observations ( see fig .
curie temperatures also increase ( 110 k and 130 k ) as well as the contribution from ge@xmath15mn@xmath16 clusters .
+ finally when increasing @xmath2 above 160@xmath4c , the nanocolumns magnetic signal vanishes and only ge@xmath15mn@xmath16 clusters and diluted mn atoms coexist .
the overall magnetic signal becomes comparable to the one measured on annealed samples in which only ge@xmath15mn@xmath16 clusters are observed by tem ( see fig .
+ the magnetic properties of high-@xmath10 nanocolumns obtained for @xmath2 close to 130@xmath4c are discussed in detail in ref.@xcite .
+ in conclusion , at low growth temperatures ( @xmath50120@xmath4c ) , nanocolumns are made of almost 4 independent elongated magnetic nanostructures . for @xmath75120@xmath4c ,
these independent nanostructures percolate into a single nanocolumn sharply leading to higher blocking temperatures .
increasing @xmath2 leads to larger columns with a wider size distribution as evidenced by zfc - fc measurements and given by tem observations . in parallel
, some ge@xmath15mn@xmath16 clusters start to form and their contribution increases when increasing @xmath2 .
results on magnetic anisotropy seems counter - intuitive . indeed ge@xmath15mn@xmath16 clusters exhibit strong out - of - plane anisotropy whereas nanocolumns which are highly elongated magnetic structures are almost isotropic .
this effect is probably due to compensating in - plane magnetoelastic coupling ( due to the columns compression ) and out - of - plane shape anisotropy .
in this paper , we have investigated the structural and magnetic properties of thin ge@xmath0mn@xmath1 films grown by low temperature molecular beam epitaxy .
a wide range of growth temperatures and mn concentrations have been explored .
all the samples contain mn - rich nanocolumns as a consequence of 2d - spinodal decomposition .
however their size , crystalline structure and magnetic properties depend on growth temperature and mn concentration . for low growth temperatures ,
nanocolumns are very small ( their diameter ranges between 1.8 nm for 1.3 % of mn and 2.8 nm for 11.3 % of mn ) , their curie temperature is rather low ( @xmath23 170 k ) and they behave as almost four uncorrelated superparamagnetic nanostructures .
increasing mn concentration leads to higher columns densities while diameters remain nearly unchanged . for higher growth temperatures ,
the nanocolumns mean diameter increases and their size distribution widens .
moreover the 4 independent magnetic nanostructures percolate into a single magnetic nanocolumn .
some columns are ferromagnetic even if curie temperatures remain quite low . in this regime , increasing mn concentration leads to larger columns while their density remains nearly the same . in parallel ,
ge@xmath15mn@xmath16 nanoclusters start to form in the film with their @xmath42-axis perpendicular to the film plane . in both temperature regimes ,
the mn incorporation mechanism in the nanocolumns and/or in the matrix changes above 5 % of mn and nanocolumns exhibit an isotropic magnetic behaviour due to the competing effects of out - of - plane shape anisotropy and in - plane magnetoelastic coupling . finally for a narrow range of growth temperatures around 130@xmath4c , nanocolumns exhibit curie temperatures
higher than 400 k. our goal is now to investigate the crystalline structure inside the nanocolumns , in particular the position of mn atoms in the distorted diamond structure , which is essential to understand magnetic and future transport properties in ge@xmath0mn@xmath1 films .
the authors would like to thank dr .
f. rieutord for grazing incidence x - ray diffraction measurements performed on the gmt station of bm32 beamline at the european synchrotron radiation facility . | we report on the structural and magnetic properties of thin ge@xmath0mn@xmath1 films grown by molecular beam epitaxy ( mbe ) on ge(001 ) substrates at temperatures ( @xmath2 ) ranging from @xmath3c to 200@xmath4c , with average mn contents between 1 % and 11 % .
their crystalline structure , morphology and composition have been investigated by transmission electron microscopy ( tem ) , electron energy loss spectroscopy and x - ray diffraction . in the whole range of growth temperatures and mn concentrations
, we observed the formation of manganese rich nanostructures embedded in a nearly pure germanium matrix .
growth temperature mostly determines the structural properties of mn - rich nanostructures . for low growth temperatures ( below 120@xmath4c ) , we evidenced a two - dimensional spinodal decomposition resulting in the formation of vertical one - dimensional nanostructures ( nanocolumns ) .
moreover we show in this paper the influence of growth parameters ( @xmath2 and mn content ) on this decomposition _
i.e. _ on nanocolumns size and density . for temperatures higher than @xmath5c
, we observed the formation of ge@xmath6mn@xmath7 clusters . for intermediate growth temperatures
nanocolumns and nanoclusters coexist . combining high resolution tem and superconducting quantum interference device magnetometry , we could evidence at least four different magnetic phases in ge@xmath0mn@xmath1 films : @xmath8 paramagnetic diluted mn atoms in the germanium matrix , @xmath9 superparamagnetic and ferromagnetic low-@xmath10 nanocolumns ( 120 k
@xmath11 @xmath10 @xmath11 170 k ) , @xmath12 high-@xmath10 nanocolumns ( @xmath10 @xmath13 400 k ) and @xmath14 ge@xmath15mn@xmath16 clusters . |
the recent impressive advances in the field of quantum simulators allowed to probe the many - body physics of strongly correlated systems at the level of the single quantum object . at present cold atoms trapped in optical lattices can be considered among the most promising examples of quantum simulators . by means of ultracold atomic and molecular gases , it is nowadays possible to reach a degree of control and accuracy in engineering the dynamics of many - body systems that were unimaginable in previously . as a consequence , the coherent quantum dynamics emerging from carefully tailored microscopic hamiltonians can now be tested experimentally @xcite .
it has been possible , just to recall one example , to implement the bose - hubbard ( bh ) model @xcite and to detect its zero - temperature superfluid ( sf ) to mott insulator ( mi ) quantum phase transition @xcite .
other models involving spinor gases , fermi systems , bose - fermi mixtures , or dipolar gases have been also devised and realized , providing an even richer phase diagram ( see for example the review @xcite ) .
we mention the stabilization of density - wave ( dw ) phases for bosons , as well as more peculiar topological or supersolid orderings , which can arise in the presence of finite - range interactions @xcite .
more recently a novel kind of many - body quantum simulator has been introduced , based on the idea to use single photons as quantum objects . since photons hardly interact in open space , the most natural way to significantly increase their interactions is to trap them into an optical qed cavity , and couple the field with atoms / molecules inside it in order to create an optical nonlinearity .
if the nonlinearity is sufficiently large , the so called photon blockade sets in @xcite , namely , the presence of a single photon inside a cavity prevents a second one to enter it . in the rotating - wave approximation ,
the simplest light - matter interaction scheme of this type can be accurately described by the jaynes - cummings ( jc ) model . by arranging an array of cavities coupled through the photon hopping , such to generate a competition between the hopping and the on - site nonlinearities
, one can devise a setup that is well described by the so called jaynes - cummings - hubbard ( jch ) model @xcite .
in many respects , if one ignores dissipation , the physics emerging from the jch hamiltonian resembles , at low - energies , that of an effective bh model .
probably the main difference between the two systems is that , instead of having neutral bosons as building blocks of the model , in the jch hamiltonian one has to think in terms of polaritons , _
i.e. _ , combined photonic / atomic excitations .
many different works already addressed the jch equilibrium phase diagram with analytical , as well as numerical methods , leading to a fairly complete theoretical understanding of the nature and the location of the emerging phases and phase transitions in terms of the parameters governing the system ( the field has been recently reviewed in , _
e.g. _ , refs .
@xcite ) .
additional interest in cavity arrays comes from the fact that these systems can be naturally considered as open - system quantum simulators .
some related features have been recently explored @xcite . in the following
we will not touch on this and consider only the `` equilibrium '' phase diagram .
this intense activity has been very recently boosted by the first experiments on qed cavity arrays @xcite . as of today
, the most concrete possibility to realize controllable and scalable quantum simulators with cavity arrays involves circuit - qed cavities @xcite .
so far the coupling between cavities has been mostly considered through photon hopping .
only few works started addressing more general schemes , where the cavity coupling can be induced also by means of non - linear elements @xcite .
such configurations include cross - kerr interactions and/or correlated hopping terms , which lead to generalizations of the jch model in a way similar to the extended bh ( ebh ) hamiltonian for atoms with large dipole momentum loaded in optical lattices @xcite . the underlying physical model is believed to possess a much richer structure , with the emergence of exotic phases of correlated polaritons .
it is particularly interesting to address these schemes in one - dimensional ( 1d ) systems , where interactions become crucial to stabilize exotic phases of matter @xcite .
these notably include a series of nontrivial density - wave ( dw ) states , which can arise in the strong coupling regime @xcite , as well as supersolidity and phase - separation effects @xcite .
extension to consider also counter - rotating terms in the ultrastrong coupling regime , thus leading to the so called rabi - hubbard model @xcite , have been investigated @xcite .
however we are not aware of numerical investigations of coupled cavity models beyond the jch and rabi - hubbard model . in all such situations , non - perturbative , either numerical or analytical calculations
are necessary . here
the density - matrix renormalization group ( dmrg ) algorithm @xcite has been employed to work out the quantitative zero - temperature phase diagram of the jch model @xcite .
this is a particularly efficient method for the statics of 1d many - body problems .
its key strategy consists in constructing a portion of the system ( called block ) and then recursively enlarge it . at each step
, the basis of the corresponding hamiltonian is truncated to a given value @xmath1 , so that one can manage the hamiltonian in an effective hilbert space of fixed dimensions , as the physical system grows .
this truncation is performed by retaining the eigenstates corresponding to the @xmath1 highest eigenvalues of the reduced density matrix of the block .
the aim of this paper is to quantitatively study a generalization of the jch hamiltonian , aimed at taking into account an effective nearest - neighbor nonlinearity between cavities mediated by an @xmath0-type four - level system as discussed for two cavities in ref .
the presence of this coupling leads to an effective cross - kerr non - linearity .
an analysis at the mean - field level of a dissipative open ebh as an effective model for nonlinearly coupled cavities has been performed , unveiling the emergence of novel photon crystal and supersolid phases @xcite .
here we do not resort to the effective ebh model and analyze the full model as introduced in @xcite . using the dmrg algorithm ,
we work out the 1d ground - state phase diagram .
we show that a physics similar to the ebh model appears , with a rich phase diagram including gapless sf , as well as mi and dw phases of polaritons .
we postpone the analysis of the interplay of driving and dissipation to a future work .
the paper is organized as follows . in the next two sections we introduce the model of coupled cavities of our interest ( sec .
[ sec : model ] ) and the quantities we are going to address , namely the energy gaps , and the staggered number - number correlations ( sec . [ sec : quantities ] ) . in sec .
[ sec : pd ] we discuss the zero - temperature equilibrium phase diagram , focusing on the mi / sf boundary and on the boundary separating the dw from the other phases . finally , in sec .
[ sec : summary ] we draw our conclusions .
-type atoms . the photon hopping between nearest - neighbor cavities has a strength @xmath2 .
each effective site is composed of a cavity and an atom ( dashed box ) .
( b ) level structure of the @xmath0-type atoms .
the transition @xmath3 is resonantly coupled to the cavity mode of its own site with strength @xmath4 , while @xmath5 is coupled to the cavity mode of its right nearest - neighbor site with strength @xmath6 , and has a detuning @xmath7 .
the transition @xmath8 is in resonance with an external laser field of strength @xmath9 . ]
let us consider a 1d array of qed cavities , where photons can hop between neighboring cavities .
moreover two adjacent resonators are also nonlinearly coupled to each other via a @xmath0-type four - level system , as shown in fig .
[ fig1](a ) . for the sake of clarity in our description
, we shall divide the 1d array into coupled effective sites composed of a cavity and an atom .
the four levels are denoted by @xmath10 , and are depicted in fig .
[ fig1](b ) .
an external laser with frequency @xmath9 resonantly drives the transition @xmath11 .
the transition @xmath3 is resonantly coupled to the cavity mode of the same site with strength @xmath4 , while the transition @xmath5 couples to the cavity mode of its right nearest - neighbor site with strength @xmath6 , and a detuning @xmath7 . the use of such @xmath0-type atom for generating large kerr nonlinearity has been extensively studied in the literature @xcite , however the vast majority of the scenarios only focused on a single - mode cavity .
our work is inspired by the idea of ref .
@xcite , where the cross - kerr nonlinearity is generated between two different and neighboring cavities , in circuit - qed systems . in practice , we use the unbalanced couplings of atomic transition @xmath12 with left cavity mode , and @xmath13 with right cavity mode respectively , in order to generate the local ( @xmath4 ) and nonlocal ( @xmath6 ) nonlinearities of our many - body system .
this kind of four - level artificial molecule can be realized using two josephson transmon qubits coupled by a superconducting quantum interference device .
using the interaction picture and in the rotating - wave approximation , the system hamiltonian reads @xmath14 \ , , \label{hamiltonian}\ ] ] where @xmath15 , and @xmath16 @xmath17 is the annihilation ( creation ) operator of the cavity mode .
the subscripts denote the site position along the 1d chain .
the first three terms in the r.h.s . of eq .
describe the local terms and the nonlinearities on each site . inside the latter brackets ,
the first term is the photon hopping , while the second term describes the coupling of the atom to its right neighboring cavity , which generates an effective nonlocal cross - kerr nonlinearity between the two cavities .
hereafter we concentrate on the 1d model in eq . at zero temperature , specifically addressing the case without dissipation with dmrg .
let us also fix the hamiltonian quantities in units of @xmath9 , set @xmath18 , and work with open boundary conditions .
we recall that , in the presence of dissipation , the problem becomes much more difficult to be handled numerically . for the system we are considering here , in the strong coupling regime atoms and photons
can not be considered as two separate entities .
it is thus natural to investigate the phase diagram in terms of combined atomic / photonic modes , named polaritons .
the polaritonic number operator on each site @xmath19 , representing the number of local excitations , is defined as @xmath20 . for the closed system described by the hamiltonian , the total number @xmath21 of such polaritons is a conserved quantity . in the following we work in the canonical ensemble for polaritons , and
focus on the integer filling situation .
the different nature of the various phases is sensitive to a number of properties which we are going to focus on .
here we are going to study quantities that resemble those characterizing the various phases of the ebh model @xcite .
first of all , the ground - state energy gap is an important indicator which characterizes the presence or absence of criticality in the model . in particular , in the critical sf phase , the _ charge gap _ vanishes in the thermodynamic limit . on the other side , in the insulating mi and dw phases
, such gap remains finite . in order to make connection with a similar notation in the ebh model , below we introduce the so called charge and neutral gaps referring respectively to the gaps corresponding to adding one extra particle ( charge " sector ) or remaining with the same number of particles ( neutral " sector ) .
we stress however that in the present model the excitation carry no real charge .
this has to be understood only as a convention .
the charge gap is defined as @xmath22 where , in the canonical ensemble , @xmath23 ( @xmath24 ) denotes the extra energy needed to add ( remove ) one particle , _ i.e. _ one polariton , in the system . in the specific , focusing on the unit filling , @xmath25 and @xmath26 , where @xmath27 is the ground - state energy per site of an @xmath28-sites cavity - array with exactly @xmath28 excitations , and @xmath29 ( @xmath30 ) is the corresponding energy per site with one excitation more ( less ) .
it is therefore possible to extrapolate @xmath31 by running three different dmrg simulations with fixed number of polaritons @xmath32
@xcite .
while the charge gap is able to detect particle - hole excitations , in some circumstances it is possible that the dominant low - energy excitations are of a different type .
their presence can be detected only by the so called _ neutral gap _ at a fixed number of particles , @xmath33 where , again working in the canonical ensemble , @xmath34 denotes the first excited energy per site of an @xmath28-site system with @xmath28 excitations . in the following
, we also focus on the analysis of the staggered diagonal order for the polaritons , in order to distinguish the dw from the other phases .
we do this by investigating the two - point correlation function @xmath35 where @xmath36 denotes the polariton fluctuation from the average filling @xmath37 .
the order parameter identifying the dw phase is thus given by : @xmath38 .
a finite value of @xmath39 indicates a tendency to establish , in the thermodynamic limit , a staggered occupation of the polaritons . on the other side , in the mi as well as the sf phases ,
@xmath40 vanishes exponentially with increasing distance @xmath41 .
the zero - temperature phase diagram of model , at unit polariton filling @xmath42 and in the @xmath43 plane , is summarized in fig .
we observe that three different phases can be stabilized .
their boundaries have been obtained by means of a finite - size scaling of the numerical data , for systems up to @xmath44 sites . in our simulations we imposed a cutoff photon number in each cavity ,
such that @xmath45 .
we also truncated the effective hilbert space dimension to a value @xmath46 in all the simulations , except for those shown in fig .
[ fig5 ] for the neutral energy gap ( see the discussion in sec . [
sec : neutral_gap ] ) .
we checked that , by increasing @xmath1 and the local fock - space truncation over the photon number , the results concerning the charge gap and the dw order parameter do not change on the scales shown here . for small photon hopping ( @xmath47 ) , by increasing the nonlocal nonlinearity @xmath6
the system exhibits a direct transition from the mi to the dw phase . on the other hand , for @xmath48 , the mi - to - dw transition
is mediated by an extended region appearing at intermediate @xmath6 values , where the system stabilizes into a gapless sf . in the following we are going to elucidate our finite - size scaling procedure and how we were able to distinguish between the different phases . in the limit of small @xmath6 and @xmath2 values , the dominant presence of the on - site interactions stabilize the system into a mi phase with exactly one polariton per cavity ( @xmath42 ) , and where the charge energy gap has a finite value .
as long as the hopping strength @xmath2 is progressively increased ( and for fixed @xmath49 ) , the system eventually enters a sf phase , with a vanishing gap .
the filled squares of fig .
[ fig2 ] denoting the mi / sf boundaries have been obtained by means of a finite - size scaling of the charge gap .
we performed simulations up to @xmath50 sites and analyzed whether the gap closes or remains finite in the thermodynamic limit @xmath51 . in fig .
[ fig3 ] , left panel , we highlight the size - dependence of @xmath31 as a function of @xmath52 for two points in the phase space close to the mi / sf transition ( see points along the dotted line in fig .
[ fig2 ] ) .
we expect to see a quadratic dependence @xmath53 ( dashed line ) at large @xmath28 @xcite , however a linear extrapolation ( solid line ) is already a good approximation to the scaling , and we can use it to determine @xmath31 in the thermodynamics limit . indeed , we observe that the difference between quadratic and linear extrapolation is tiny ( @xmath54 ) and does not produce any distinguishable modification on the scale of fig .
[ fig2 ] . in the specific case of fig .
[ fig3 ] , we fixed @xmath55 and chose two different values of @xmath56 corresponding to configurations in the gapped mi ( @xmath57 , triangles ) and in the gapless sf phase ( @xmath58 , squares ) .
the mi is signaled by an extrapolated finite value of @xmath59 , while in the sf this is zero .
in order to locate the critical @xmath6 for a given value of @xmath2 ( filled squares in fig . [ fig2 ] ) we perform a linear extrapolation of the charge gaps in the vicinity of the critical value of @xmath6 .
an example of such procedure is shown in the right panel of fig .
[ fig3 ] , where we plot @xmath31 as a function of @xmath6 , when this is close to the phase transition ( in the specific , here we set @xmath55 ) .
after a linear extrapolation , we get a critical @xmath6 value corresponding to @xmath60 .
an analogous procedure is repeated for all the filled squares shown in fig .
[ fig2 ] , thus identifying the mi / sf boundary .
the dw phase is characterized by a finite order parameter @xmath61 .
let us therefore look at the two - point staggered correlator in eq . .
since in dmrg simulations we are employing open boundary conditions , to minimize the border effects we analyze the correlations in such a way that the two points are taken symmetrically with respect to the center of the system , with @xmath62 , have been chosen such that @xmath63 for odd r , and @xmath64 for even r
( e.g. for @xmath65 sites , @xmath66 corresponds to @xmath67 ; @xmath68 corresponds to @xmath69 ; @xmath70 to @xmath71 , and so on ) ] .
the left panel of fig .
[ fig4 ] shows how differently such polariton correlations behave when the system goes from the mi to dw phase , for a fixed system size . to be more accurate , in the right panel we performed a finite - size scaling and showed that the staggered correlation @xmath72 approaches the zero value exponentially with @xmath28 , in the mi phase ( a similar behavior occurs in the sf region ) .
on the other hand , in the dw such correlator asymptotically converges to a finite value . in the specific ,
here we fix @xmath73 and show that for @xmath74 the dw order is exponentially suppressed with @xmath28 , while for @xmath75 it remains finite .
the @xmath61 order parameter reached for @xmath51 is displayed in the inset as a function of @xmath6 .
in order to determine the dw boundary in the phase diagram of fig .
[ fig2 ] , we adopted the following protocol . for a fixed value of @xmath76 , we start increasing @xmath6 from zero up to a finite value , with a fixed increment @xmath77 , and to compute the dw order parameter for all such values of @xmath6 . the boundary of dw phase in the @xmath43 plane ( filled circles in fig .
[ fig2 ] ) , for any fixed @xmath2 , is located by the @xmath78 that gives the first non - vanishing order parameter @xmath39 .
here we stress that , because of the arrangement of our 1d array [ see fig .
[ fig1](a ) ] and of the asymmetric coupling between the atom and its right / left cavity , the antiferromagnetic symmetry of the system is spontaneously broken .
in particular , the state @xmath79 of the @xmath28-th atom will be never occupied , since the transition @xmath5 does not couple to any cavity mode [ see eq . ] . as a consequence , in our simulations
we do not need any symmetry - breaking potential .
we can observe that the expectation value for the onsite number of polaritons explicitly exhibits a staggered behavior , in that the occupation of the @xmath80-th site is always higher than that of the @xmath81-th site ( for any integer value of @xmath82 ) .
finally we notice that such staggering persists at finite size , also for the set of parameters corresponding to the mi phase , although it is extremely tiny and decreases with @xmath28 .
this effect eventually disappears in the thermodynamic limit .
the extension of the dw phase depends on the cavity detuning @xmath7 .
in particular , the robustness of the order parameter increases with increasing the modulus of the detuning ( see fig . [ fundelta ] ) .
quite remarkably , we note that a positive @xmath7 will never stabilize an antiferromagnetic dw ordering . the analysis leading to the phase diagram in fig .
[ fig2 ] has been corroborated by a study of the neutral gap , which vanishes both in proximity of the phase transitions and in the entire superfluid region .
differently for the charge gap , it is able to detect the presence of excitations other than particle - hole , and thus locates the boundaries of insulating regions ( as the dw ) beyond the mi . the data displayed in fig .
[ fig5 ] show the behavior of @xmath83 as a function of @xmath6 , for a fixed value of @xmath84 .
in particular we analyzed a vertical cut in the phase diagram of fig .
[ fig2 ] ( see the rightmost vertical dotted line in that figure ) , where the system can be in three different phases according to the value of @xmath6 . with increasing @xmath6 , it goes from the mi phase ( nonzero @xmath83 , for @xmath85 ) to the sf phase ( zero @xmath83 , for @xmath86 ) , and then to the dw phase ( nonzero @xmath83 , for @xmath87 ) . while we can not see a clear signature of a finite gap for @xmath88 , the scaling with the size displayed in the left panel of fig .
[ fig5 ] seems to suggest the scenario depicted above .
it is however important to stress that the dmrg simulations needed to compute the neutral gap have to target the two lowest - lying eigenstates in a single run .
thus they generally require a larger dimension @xmath1 of the effective hilbert space , as compared to all the other ground - state calculations discussed before .
the analysis of the neutral gap requires a careful convergence test of the results with @xmath1 , which we provide in the right panel of fig .
we observe that the non monotonic features that are visible in the region @xmath86 have to be probably ascribed to the inaccuracy of the method at small @xmath1 values .
this signals the presence of the gapless sf phase there , in agreement with the results provided by the charge gap ( mi / sf boundary ) and for the dw order parameter ( sf / dw boundary ) .
using the density - matrix renormalization group with open boundary conditions , we studied the equilibrium phase diagram of a system of coupled qed cavities in one dimension .
we provided results beyond the standard model of couplings through photon hopping , and also considered nearest - neighbor cross - kerr nonlinearities .
our analysis is based on a finite - size scaling of the ground - state charge and neutral gaps , as well as of the density - wave order parameter , for systems up to 300 sites .
we showed that , beyond the conventional mott insulator and superfluid phases , the presence of a nearest - neighbor nonlinear coupling can also stabilize a density - wave ordering of polaritons .
the authors declare that they have no competing interests .
all the authors participated in the design of the research , analysis of the results , and writing of the paper .
the dmrg code used to run all the simulations of this research has been developed and written by d. rossini and coworkers @xcite ( see also ` www.dmrg.it ` ) .
the dmrg simulations were performed by j. jin .
we would like to acknowledge our previous collaboration with m. hartmann and m. leib which was inspiring for the present work .
this work was supported by italian miur via firb project rbfr12nlna and prin project 2010llkjbx , by eu through ip - siqs , and by national natural science foundation of china under grant no .
11175033 and no . | we study the zero - temperature phase diagram of a one - dimensional array of qed cavities where , besides the single - photon hopping , an additional coupling between neighboring cavities is mediated by an @xmath0-type four - level system . by varying the relative strength of the various couplings ,
the array is shown to exhibit a variety of quantum phases including a polaritonic mott insulator , a density - wave and a superfluid phase .
our results have been obtained by means of numerical density - matrix renormalization group calculations .
the phase diagram was obtained by analyzing the energy gaps for the polaritons , as well as through a study of two - point correlation functions . |
as the lhc is providing us with unprecedented high - energy data , top - quark physics has reached a mature level in both theory and experiment .
we are able to test properties and interactions of the top - quark with high precision in inclusive cross sections , as well as in specific kinematic regimes through differential distributions . probing
the top - antitop charge asymmetry is a precision test of quantum chromodynamics ( qcd ) .
it bares information on strong interactions of heavy quarks that can not be directly obtained from cross - section measurements @xcite .
phenomenologically , the charge asymmetry is often approached by comparing angular distributions of top- and antitop - quarks , commonly expressed in terms of rapidity differences ( for an overview of lhc observables , see @xcite ) .
precise qcd predictions of the rapidity asymmetry at the tevatron have been obtained at next - to - next - to - leading order ( nnlo ) in perturbation theory @xcite and from soft - gluon resummation @xcite . at the lhc ,
the rapidity asymmetry in top - pair production is very small for two reasons : it is induced by partonic quark - antiquark interactions , which are buried under the large ( charge - symmetric ) gluon - gluon background . and it is suppressed in qcd , since it arises only at the perturbative next - to - leading order ( nlo ) .
the atlas and cms collaborations have measured the rapidity asymmetry at center - of - mass ( cm ) energies of 7 and 8 tev @xcite . however
, statistical and systematic uncertainties currently strongly limit the sensitivity to the charge asymmetry through rapidity differences .
the ultimate sensitivity during the 13-tev phase of the lhc will crucially depend on whether it is possible to reduce the systematic uncertainties well below one percent .
can we find an observable of the charge asymmetry that is larger in magnitude and thus better suited for a precise measurement ? if the top - antitop pair is produced in association with a hard jet , the rapidity asymmetry appears already at the leading order ( lo ) in qcd , and it is indeed large and negative .
but nlo corrections are sizeable and positive and reduce the asymmetry to a quantity no larger than in inclusive top - pair production @xcite .
alternatively , the charge asymmetry can be observed in jet - associated top - pair production through top - antitop _ energy _
differences @xcite .
this so - called energy asymmetry is induced at lo qcd from partonic quark - gluon interactions , which are much more abundant than quark - antiquark states in proton - proton collisions at high energies .
the energy asymmetry is therefore less affected by the gluon - gluon background and larger than rapidity asymmetries .
since the definition of the energy asymmetry relies on the jet kinematics , it is important to know the impact of qcd corrections on the observable . in this work ,
we investigate fixed - order nlo corrections and parton - shower effects using monte - carlo methods .
we will show that and why these corrections are moderate and provide improved predictions of the energy asymmetry for the lhc at 13 tev collision energy . in section [ sec : observables ] , we review the definition of the energy asymmetry and a related optimized observable .
fixed - order nlo qcd corrections are discussed in section [ sec : nlo ] from a qualitative viewpoint and in section [ sec : lhc ] numerically for the 13-tev lhc .
parton - shower effects from multi - parton emission are the topic of section [ sec : shower ] . in section
[ sec : kinematics ] , we show how to enhance the energy asymmetry with kinematic cuts and present our final predictions for the lhc .
we conclude in section [ sec : conclusions ] with an outlook on the discovery prospects of the asymmetry .
the top - antitop energy asymmetry as an observable of the charge asymmetry was first introduced in reference @xcite . here
we briefly review the definition and properties of the energy asymmetry , as well as of an optimized observable @xcite . for details , we refer the reader to these two articles . the _ energy asymmetry _ in top - pair production requires the detection of an associated jet . in the process @xmath4 , the energy asymmetry is defined as @xcite @xmath5 here @xmath6 is the difference between the top and antitop energies . the angle @xmath7 is defined as the scattering angle of the hardest jet with respect to the direction of the incoming parton @xmath8 of the process @xmath9 . both @xmath10 and @xmath7
are defined in the @xmath11 rest frame , which at lo corresponds with the parton cm frame . at the parton level ,
the energy asymmetry is generated mainly from quark - gluon initial states .
an exemplary feynman diagram contributing to the energy asymmetry at lo is shown in fig . [
fig : diag1 ] @xmath12 . at this level ,
the energy asymmetry is equivalent to a forward - backward asymmetry of the quark - jet in the top - antitop rest frame .
contributions to @xmath13 from the quark - antiquark initial state are small and vanish when integrated over a range symmetric around @xmath14 .
the total energy asymmetry @xmath15 is obtained by integrating @xmath16 and @xmath17 in ( [ eq : ea ] ) over the range @xmath18 $ ] .
the jet angular distribution @xmath13 is shown in the upper right panel of figure [ fig : fo - lo - nlo ] .
as one can see , @xmath13 is symmetric in @xmath19 , with a minimum at @xmath14 . however , either of the two partonic contributions @xmath20 and @xmath21 has its minimum where the jet is emitted in the direction of the incoming quark , i.e. , for @xmath22 and @xmath23 , respectively .
we will `` guess '' the incoming quark s direction by exploiting the fact that the quark carries a larger fraction of the proton momentum along the beam axis than the gluon .
the final state in quark - gluon interactions thus tends to be boosted in the quark direction .
we quantify this boost in terms of the rapidity of the top - antitop - jet cm frame , @xmath24 , which at the lo is connected with the proton momentum fractions of @xmath8 and @xmath25 via @xmath26 .
this allows us to define an _
optimized energy asymmetry _ as @xcite @xmath27 where @xmath7 and @xmath28 are defined as in ( [ eq : ea ] ) , with an additional cut on the boost , @xmath24 . by associating forward ( @xmath29 ) and backward ( @xmath30 ) boosted final states with forward ( @xmath31 $ ] ) and backward ( @xmath32 $ ] ) scattered jets , we combine the maximum contributions from the @xmath33 and @xmath34 channels in the range @xmath31 $ ] . in the lower right panel of figure [ fig : fo - lo - nlo ] , we show the jet angular distribution of the optimized energy asymmetry .
the minimum of @xmath35 lies around @xmath36 and is deeper than for @xmath13 .
is tailored to the kinematics of the partonic quark - gluon channel , quark - antiquark contributions deplete its minimum ( see also reference @xcite ) . however
, this effect is small due to the low quark - antiquark abundance in proton - proton collisions at the lhc . ] in our numerical analysis in section [ sec : kinematics ] , we will focus on this optimized energy asymmetry .
occasionally , we will compare the energy asymmetry to the widely investigated _ rapidity asymmetries _ at the tevatron , @xmath37 and at the lhc , @xmath38 the top and antitop rapidities , @xmath39 and @xmath40 , are defined in the laboratory frame .
the script @xmath41 is used if the top - antitop pair is produced in association with a hard jet .
the rapidity asymmetry is generated from the quark - antiquark initial state and , unlike the energy asymmetry , does not require a hard jet in the final state . in the presence of an observed jet ,
however , @xmath42 is strongly reduced by qcd corrections at nlo @xcite , as we mentioned in the introduction .
this fact raises the question if a similar effect occurs for the energy asymmetry . in what follows
, we will discuss qcd corrections to the energy asymmetry in detail and explain why their impact is moderate . and @xmath43 ) and at lo ( @xmath12).,height=201 ]
the properties of the energy asymmetry at nlo are closely related to the kinematics of the observed jet .
we can obtain a qualitative understanding by analyzing the cross section and the asymmetry in the limit of soft and collinear jet emission .
the jet distribution of the cross section , @xmath44 , is displayed in the upper left panel of figure [ fig : fo - lo - nlo ] . [
cols="^,^ " , ] the effect of these cuts on the jet distribution of the optimized energy asymmetry and its normalization is displayed in figure [ fig : kin - cuts ] .
the cut on @xmath45 is very efficient in enhancing the asymmetry ( red curves ) .
however , this comes at the price of sacrificing a large portion of the cross section in the central - jet region around @xmath46 , where the asymmetry is largest and most precisely predicted . a lower cut on the boost @xmath47 ( blue curves ) , in turn , reduces the gluon - gluon background , ergo the cross section in the collinear regime @xmath48 .
by combining both cuts we can find an optimal balance of enhancing the absolute asymmetry and preserving a sizeable data set . in our exemplary set of combined cuts , @xmath49 and @xmath50 , the optimized energy asymmetry reaches @xmath51 in its minimum ( purple curves ) .
since the optimized energy asymmetry is largest near @xmath52 , a further cut on the jet scattering angle should be applied .
this cut can be expressed in terms of the jet rapidity in the @xmath11 frame , @xmath53 , which can be measured as the difference between the jet rapidity and the final - state boost in the laboratory frame @xcite , @xmath54 in table [ tab : predictions ] , we present our predictions of the optimized energy asymmetry @xmath55 and the corresponding cross section @xmath56 at nlo qcd for various combinations of kinematic cuts . in the three tables ,
the observables have been integrated over different ranges of the jet angle @xmath7 ( or the corresponding jet rapidity @xmath53 ) .
the errors correspond to scale variations @xmath57\,m_t$ ] .
notice that the scale uncertainty is much reduced when cuts on @xmath7 are applied .
this feature can be understood from figure [ fig : scale ] , where the small panels show that the relative scale uncertainty is largest in the regions around @xmath58 .
the improved precision of @xmath55 for @xmath46 is another strong argument for applying cuts on the jet direction . in our example with @xmath49 and @xmath50 , the relative scale uncertainty of the asymmetry in the range @xmath59
is no more than @xmath60 .
@xmath61 & no cut on @xmath24 & @xmath50 + ' '' '' height 1pt @a xhline no cut on @xmath10 & @xmath62 & @xmath63 + @xmath64 & @xmath65 & @xmath66 + @xmath67 & @xmath68 & @xmath69 + @xmath70 / @xmath71 & no cut on @xmath24 & @xmath50 + ' '' '' height 1pt @a xhline no cut on @xmath10 & @xmath72 & @xmath73 + @xmath64 & @xmath74 & @xmath75 + @xmath67 & @xmath76 & @xmath77 + @xmath59 / @xmath78 & no cut on @xmath24 & @xmath50 + ' '' '' height 1pt @a xhline no cut on @xmath10 & @xmath79 & @xmath80 + @xmath64 & @xmath81 & @xmath82 + @xmath67 & @xmath83 & @xmath84 + the discovery prospects of the energy asymmetry at the lhc will also depend on experimental aspects , such as the detector acceptance or reconstruction efficiency . in reference
@xcite , we have estimated the statistical uncertainty of the energy asymmetry at lo for various assumptions on the experimental luminosity and efficiency . since nlo effects on the asymmetry are moderate , the main conclusion from this analysis persists : a statistical significance of @xmath85 standard deviations , often considered the discovery level , is expected to be in reach within the current run of the lhc at 13 tev .
in this work , we have advocated the energy asymmetry as an observable of the charge asymmetry in jet - associated top - antitop production .
our main goal was to answer the question if the energy asymmetry is stable under qcd corrections .
to this end , we have calculated fixed - order nlo qcd contributions to the energy asymmetry .
corrections are moderate and most pronounced for jet emission collinear to the incoming partons ( see figure [ fig : fo - lo - nlo ] ) . in this region ,
new virtual nlo contributions enhance the asymmetry over its lo prediction . in the region of jet emission perpendicular to the incoming partons , where the absolute asymmetry is largest , nlo corrections are of a few percent only .
to estimate the uncertainty of our predictions , we have analyzed their dependence on the factorization and renormalization scales .
while the inclusion of nlo contributions reduces the scale dependence of the cross section , it enhances the scale uncertainty of the asymmetry near the endpoints of the jet distribution ( see figure [ fig : scale ] ) .
this is due to the fact that the renormalization scale dependence no longer cancels between the numerator and denominator of the asymmetry , once nlo contributions are taken into account .
the uncertainty is of a few percent only for perpendicular jet emission , but reaches several tens of percent in the collinear region .
these observations suggest to apply a cut on the jet angle @xmath7 that removes events with collinear jet emission .
the asymmetry can be further enhanced by applying kinematic cuts on the top - antitop energy difference , @xmath10 , and the boost of the final state , @xmath24 ( figure [ fig : kin - cuts ] ) .
our final results are shown in table [ tab : predictions ] , where we make predictions for the optimized energy asymmetry at the lhc with @xmath86 .
for instance , with @xmath87 , @xmath88 and @xmath50 , corresponding to a cross section of @xmath89 , an asymmetry of @xmath90 can be reached . with tighter cuts ,
the asymmetry can be enhanced even more , but at the cost of further reducing the data set .
experimentally , the sensitivity to the energy asymmetry around its minimum might be improved by using dedicated reconstruction techniques for boosted top - quarks ( which has recently been done for the rapidity asymmetry at the lhc @xcite ) . beyond fixed - order qcd
, we have investigated the effect of multi - parton emission on the energy asymmetry ( see figure [ fig : ps ] ) . due to the strong cut on the jet s transverse momentum ,
the impact of the parton shower on the observables is rather small in the region of perpendicular jet emission .
however , the shower affects the collinear region , where effects are generally expected to be larger . in this work ,
we have confined ourselves to a basic study of the parton shower .
a more comprehensive analysis would be very welcome , in particular to investigate the impact of the parton shower in the presence of kinematic cuts .
we hope that our predictions can serve as a basis for a measurement of the top charge asymmetry at the lhc .
the energy asymmetry surpasses rapidity - based asymmetries in magnitude by far .
this makes us confident that systematic uncertainties of the experimental analysis can be reduced to a level that facilitates a discovery during run ii .
obviously , the theory prediction of the energy asymmetry can be further improved by investigating the effects of top - quark decay @xcite and by including nnlo qcd corrections , as well as soft - gluon resummation .
the careful investigation of the rapidity asymmetries at the tevatron and the lhc has shown that such improvements are possible with the joint effort and ability of our community .
we wish to thank simone alioli and manfred kraus for contributing to this project at an early stage .
we thank efe yazgan and bugra bilin for discussions of experimental aspects , jan winter for helpful comments on the manuscript , marco zaro for his support with amc@nlo , and torben schell for help with the computing cluster .
sw acknowledges funding by the carl zeiss foundation through a _
junior - stiftungsprofessur_. we also acknowledge support by the state of baden wrttemberg through bwhpc .
s. dittmaier , p. uwer and s. weinzierl , phys .
lett . * 98 * , 262002 ( 2007 ) [ hep - ph/0703120 ] .
s. dittmaier , p. uwer and s. weinzierl , eur .
j. c * 59 * , 625 ( 2009 ) [ arxiv:0810.0452 [ hep - ph ] ] .
k. melnikov and m. schulze , nucl .
b * 840 * , 129 ( 2010 ) [ arxiv:1004.3284 [ hep - ph ] ] .
s. berge and s. westhoff , jhep * 1307 * , 179 ( 2013 ) [ arxiv:1305.3272 [ hep - ph ] ] .
s. alte , s. berge and h. spiesberger , jhep * 1409 * , 084 ( 2014 ) doi:10.1007/jhep09(2014)084 [ arxiv:1406.0501 [ hep - ph ] ] .
s. alioli , s. o. moch and p. uwer , jhep * 1201 * , 137 ( 2012 ) [ arxiv:1110.5251 [ hep - ph ] ] .
j. alwall _ et al .
_ , jhep * 1407 * , 079 ( 2014 ) doi:10.1007/jhep07(2014)079 [ arxiv:1405.0301 [ hep - ph ] ] . | the top - antitop energy asymmetry is a promising observable of the charge asymmetry in jet - associated top - quark pair production at the lhc .
we present new predictions of the energy asymmetry in proton - proton collisions at 13 tev , including qcd corrections at the next - to - leading perturbative order and from the parton shower .
the effect of qcd corrections on the observable is moderate .
scale uncertainties and parton - shower effects partially cancel in the normalized asymmetry . with suitable phase - space cuts ,
the asymmetry can be enhanced at the cost of reducing the cross section .
for instance , for a cross section of @xmath0 after cuts , we predict an energy asymmetry of @xmath1 at the next - to - leading order in qcd .
our results provide a sound basis for a measurement of the energy asymmetry at the lhc during run ii .
ttk-16 - 32 observing the top energy asymmetry at the lhc s. berge@xmath2 and s. westhoff@xmath3 + @xmath2 [email protected] + @xmath3 [email protected] |
in a constraint satisfaction problem ( csp ) we are given a set of variables , and the goal is to find an assignment of the variables subject to specified constraints .
a constraint is usually expressed as a requirement that combinations of values of a certain ( usually small ) set of variables belong to a certain relation .
csps have been intensively studied in both theoretical and practical perspectives . on the theoretical side
the key research direction has been the complexity of the csp when either the interaction of sets constraints are imposed on , that is , the hypergraph formed by these sets , is restricted @xcite , or restrictions are on the type of allowed relations @xcite . in the latter direction
the main focus has been on the so called _ dichotomy conjecture _
@xcite suggesting that every csp restricted in this way is either solvable in polynomial time or is np - complete .
this ` pure ' constraint satisfaction problem is sometimes not enough to model practical problems , as some constraint that have to be satisfied are not ` local ' in the sense that they can not be viewed as applied to only a limited number of variables
. constraints of this type are called _
global_. global constraints are very diverse , the current global constraint catalog ( see ) lists 348 types of such constraints . in this paper
we focus on _ global cardinality constraints _ @xcite .
a global cardinality constraint @xmath2 is specified for a set of values @xmath3 and a set of variables @xmath4 , and is given by a mapping @xmath5 that assigns a natural number to each element of @xmath3 such that @xmath6 .
an assignment of variables @xmath4 satisfies @xmath2 if for each @xmath7 the number of variables that take value @xmath8 equals @xmath9 . in a csp with global cardinality constraints ,
given a csp instance and a global cardinality constraint @xmath2 , the goal is to decide if there is a solution of the csp instance satisfying @xmath2 . the restricted class of csps with global cardinality constraints such that every instance from this class uses only relations from a fixed set @xmath1 of relations ( such a set is often called a _ constraint language _ ) is denoted by @xmath0 .
we consider the following problem : characterize constraint languages @xmath1 such that @xmath0 is solvable in polynomial time .
more general versions of global cardinality constraints have appeared in the literature , see , e.g. @xcite , where the number of variables taking value @xmath8 has to belong to a prescribed set of cardinalities ( rather than being exactly @xmath9 ) . in this paper
we call the csp allowing such generalized constraints _ extended csp with cardinality constraints_. as we discuss later , our results apply to this problem as well . the complexity of @xmath0 has been studied in @xcite for constraint languages @xmath1 on a 2-element set .
it was shown that @xmath0 is solvable in polynomial time if every relation in @xmath1 is _ width-2-affine _ , i.e. it can be expressed as the set of solutions of a system of linear equations over a 2-element field containing at most 2 variables , or , equivalently , using the equality and disequality clauses ; otherwise it is np - complete ( we assume p@xmath10np ) . in the 2-element case @xmath0
is also known as the @xmath11-ones@xmath12 problem , since a global cardinality constraint can be expressed by specifying how many ones ( the set of values is thought to be @xmath13 ) one wants to have among the values of variables .
the parameterized complexity of @xmath11-ones@xmath12 has also been studied @xcite , where @xmath11 is used as a parameter . in the case of a 2-element domain ,
the polynomial classes can be handled by a standard application of dynamic programming .
suppose that the instance is given by a set of unary clauses and binary equality / disequality clauses .
consider the graph formed by the binary clauses .
there are at most two possible assignments for each connected component of the graph : setting the value of a variable uniquely determines the values of all the other variables in the component .
thus the problem is to select one of the two assignments for each component .
trying all possibilities would be exponential in the number of components . instead ,
for @xmath14 , we compute the set @xmath15 of all possible pairs @xmath16 such that there is a partial solution on the first @xmath17 components containing exactly @xmath18 zeroes and exactly @xmath19 ones .
it is not difficult to see that @xmath20 can be computed if @xmath15 is already known .
we generalize the results of @xcite for arbitrary finite sets and arbitrary constraint languages .
as usual , the characterization for arbitrary finite sets is significantly more complex and technical than for the 2-element set . as a straightforward generalization of the 2-element case , we can observe that the problem is polynomial - time solvable if every relation can be expressed by binary mappings .
in this case , setting a single value in a component uniquely determines all the values in the component .
therefore , if the domain is @xmath3 , then there are at most @xmath21 possible assignments in each component , and the same dynamic programming technique can be applied ( but this time the set @xmath15 contains @xmath21-tuples instead of pairs ) . one might be tempted to guess that the class described in the previous paragraph is the only class where @xmath22 is polynomial - time solvable . however , it turns out that there are more general tractable classes .
first , suppose that the domain is partitioned into equivalence classes , and the binary constraints are mappings between the equivalence classes .
this means that the values in the same equivalence class are completely interchangeable .
thus it is sufficient to keep one representative from each class , and then the problem can be solved by the algorithm sketched in the previous paragraph .
again , one might believe that this construction gives all the tractable classes , but the following example shows that it does not cover all the tractable cases .
[ exa : subcomponent ] let @xmath23 .
we claim that @xmath24 is polynomial - time solvable .
consider the graph on the variables where two variables are connected if and only if they appear together in a constraint .
as before , for each component , we compute a set containing all possible cardinality vectors , and then use dynamic programming . in each component
, we have to consider only two cases : either every variable is in @xmath25 or every variable is in @xmath26 .
if every variable of component @xmath27 is in @xmath25 , then @xmath28 can be expressed by the unary constant relation @xmath29 , and the binary relation @xmath30 . the binary relations partition component @xmath27 into sub - components @xmath31 , @xmath32 , @xmath33 . since @xmath34 is a mapping ,
there are at most 2 possible assignments for each sub - component .
thus we can use dynamic programming to compute the set of all possible cardinality vectors on @xmath27 that use only the values in @xmath25 .
if every variable of @xmath27 is in @xmath26 , then @xmath28 can be expressed as the unary constant relation @xmath35 and the binary relation @xmath36 .
again , binary relation @xmath37 partitions @xmath27 into sub - components , and we can use dynamic programming on them . observe that the sub - components formed by @xmath34 and the sub - components formed by @xmath37 can be different : in the first case , @xmath38 and @xmath39 are adjacent if they appear in the second and third coordinates of a constraint , while in the second case , @xmath38 and @xmath39 are adjacent if they appear in the first and second coordinates of a constraint .
it is not difficult to make example [ exa : subcomponent ] more complicated in such a way that we have to look at sub - subcomponents and perform multiple levels of dynamic programming .
this suggests that it would be difficult to characterize the tractable relations in a simple combinatorial way .
we give two characterizations of finite @xmath22 , one more along the line of the usual approach to the csp , using polymorphisms , and another more combinatorial one .
the latter is more technical , but it is much more suitable for algorithms .
a polymorphism of a constraint language is an operation that preserves every relation from the language .
the types of polymorphisms we need here are quite common and have appeared in the literature many times . a ternary operation @xmath40 satisfying the equations @xmath41 is said to be _ majority _ , and a ternary operation @xmath42 satisfying @xmath43 is said to be _ maltsev_. an operation is _ conservative _ if it always takes a value equal to one ( not necessarily the same one ) of its arguments .
[ the : main - poly ] for a constraint language @xmath1 , the problem @xmath0 is polynomial time solvable if and only if @xmath1 has a majority polymorphism and a conservative maltsev polymorphism .
otherwise it is np - complete . observe that for constraint languages over a 2-element domain , theorem [ the : main - poly ] implies the characterization of creignou et al .
width-2 affine is equivalent to affine and bijunctive ( definable in 2sat ) , and over a 2-element domain , affine is equivalent to having a conservative maltsev polymorphism and bijunctive is equivalent to having a majority polymorphism .
the second characterization uses logical definability .
the right generalization of mappings is given by the notion of rectangularity .
a binary relation @xmath44 is called _ rectangular _ if @xmath45 implies @xmath46 .
we say that a pair of equivalence relations @xmath47 and @xmath48 over the same domain _ cross _ , if there is an @xmath49-class @xmath50 and a @xmath48-class @xmath3 such that none of @xmath51 , @xmath52 , and @xmath53 is empty .
a relation is _
2-decomposable _ if it can be expressed as a conjunction of binary relations .
we denote by @xmath54 the set of all relations that are primitive positive definable in @xmath1 .
a constraint language is said to be _ non - crossing decomposable _ if every relation from @xmath54 is 2-decomposable , every binary relation from @xmath54 is rectangular , and no pair of equivalence relations from @xmath54 cross . for detailed definitions and discussion
see section [ sec : preliminaries ] .
[ the : main - poly2 ] for a constraint language @xmath1 , the problem @xmath0 is polynomial time solvable if and only if @xmath1 is non - crossing decomposable . otherwise it is np - complete .
the equivalence of the two characterizations will be proved in section [ sec : equiv - char ] .
following @xcite , we also study the counting problem @xmath55 corresponding to @xmath0 , in which the objective is to find the number of solutions of a csp instance that satisfy a global cardinality constraint specified .
creignou et al .
@xcite proved that if @xmath1 is a constraint language on a 2-element set , say , @xmath13 , then @xmath55 are solvable in polynomial time exactly when @xmath0 is , that is , if every relation from @xmath1 is width-2-affine .
otherwise it is @xmath56-complete .
we prove that in the general case as well , @xmath55 is polynomial time solvable if and only if @xmath0 is .
however , in this paper we do not prove a complexity dichotomy , as we do not determine the exact complexity of the hard counting problems .
all such problems are np - hard as theorems [ the : main - poly ] and [ the : main - poly2 ] show ; and we do not claim that the np - hard cases are actually @xmath56-hard .
[ the : counting - main - poly ] for a constraint language @xmath1 , the problem @xmath55 is polynomial time solvable if and only if @xmath1 has a majority polymorphism and a conservative maltsev polymorphism ; or , equivalently , if and only if @xmath1 is non - crossing decomposable .
otherwise it is np - hard .
we also consider the so called _ meta - problem _ for @xmath0 and @xmath55 : suppose set @xmath3 is fixed .
given a finite constraint language @xmath1 on @xmath3 , decide whether or not @xmath0 ( and @xmath55 ) is solvable in polynomial time . by theorems [ the : main - poly ] and [ the : counting - main - poly ] it suffices to check if @xmath1 has a majority and a conservative maltsev polymorphism . since the set @xmath3 is fixed , this can be done by checking , for each possible ternary function with the required properties , whether or not it is a polymorphism of @xmath1 . to check
if a ternary operation @xmath57 is a polymorphism of @xmath1 one just needs for each relation @xmath58 to apply @xmath57 to every triple of tuples in @xmath44 .
this can be done in a time cubic in the total size of relations in @xmath1 .
[ the : meta - problem ] let @xmath3 be a finite set .
the meta - problem for @xmath0 and @xmath55 is polynomial time solvable .
note that all the results use the assumption that the set @xmath3 is fixed ( although the complexity of algorithms does not depend on a particular constraint language ) . without this assumption
the algorithms given in the paper become exponential time , and theorem [ the : meta - problem ] does not answer if the meta problem is polynomial time solvable if the set @xmath3 is not fixed , and is a part of the input .
the algorithm sketched above is then super - exponential .
our notation concerning tuples and relations is fairly standard .
the set of all tuples of elements from a set @xmath3 is denoted by @xmath59 .
we denote tuples in boldface , e.g. , @xmath60 , and their components by @xmath61,\ba[2],\ldots$ ] . an @xmath62-ary _ relation _ on set
@xmath3 is any subset of @xmath59 .
sometimes we use instead of relation @xmath44 the corresponding predicate @xmath63 .
a set of relations on @xmath3 is called a _ constraint language _ over @xmath3 . for a subset @xmath64 with @xmath65 and an @xmath62-tuple @xmath60 , by @xmath66
we denote the _ projection of _ @xmath60 _ onto _
@xmath67 , the @xmath11-tuple @xmath68\zd\ba[i_k])$ ] .
the _ projection _ @xmath69 of @xmath44 is the @xmath11-ary relation @xmath70 .
sometimes we need to emphasize that the unary projections @xmath71 , @xmath72 of a binary relation @xmath44 are sets @xmath73 and @xmath74 .
we denote this by @xmath75 .
pairs from equivalence relations play a special role , so such pairs will be denoted by , e.g. , @xmath76 . if @xmath49 is an equivalence relation on a set @xmath3 then @xmath77 denotes the set of @xmath49-classes , and @xmath78 for @xmath7 denotes the @xmath49-class containing @xmath8 .
we say that the equivalence relation @xmath49 on a set @xmath3 is _ trivial _ if @xmath3 is the only @xmath49 class .
let @xmath3 be a finite set ( throughout the paper we assume it fixed ) and @xmath1 a constraint language over @xmath3 .
an instance of the _ constraint satisfaction problem _ ( csp for short ) @xmath79 is a pair @xmath80 , where @xmath4 is a finite set of _ variables _ and @xmath81 is a set of _
every constraint is a pair @xmath82 consisting of an @xmath83-tuple @xmath84 of variables , called the _ constraint scope _ and an @xmath83-ary relation @xmath58 , called the _ constraint relation_. a solution of @xmath85 is a mapping @xmath86 such that for every constraint @xmath82 the tuple @xmath87 belongs to @xmath44 .
a _ global cardinality constraint _ for a csp instance @xmath85 is a mapping @xmath5 with @xmath6 .
a solution @xmath88 of @xmath85 satisfies the cardinality constraint @xmath2 if the number of variables mapped to each @xmath7 equals @xmath9 .
the variant of @xmath79 allowing global cardinality constraints will be denoted by @xmath0 ; the question is , given an instance @xmath85 and a cardinality constraint @xmath2 , whether there is a solution of @xmath85 satisfying @xmath2 .
[ exa : k - ones ] if @xmath1 is a constraint language on the 2-element set @xmath13 then to specify a global cardinality constraint it suffices to specify the number of ones we want to have in a solution .
this problem is also known as the @xmath11-ones@xmath12 problem @xcite .
[ exa : coloring ] let @xmath89 be the constraint language on @xmath90 containing only the binary disequality relation @xmath10 .
it is straightforward that @xmath91 is equivalent to the graph 3-colorability problem .
therefore @xmath92 is equivalent to the graph 3-colorability problem in which the question is whether there is a coloring with a prescribed number of vertices colored each color . sometimes it is convenient to use arithmetic operations on cardinality constraints .
let @xmath93 be cardinality constraints on a set @xmath3 , and @xmath94 .
then @xmath95 and @xmath96 denote cardinality constraints given by @xmath97 and @xmath98 , respectively , for any @xmath7 .
furthermore , we extend addition to sets @xmath99 , @xmath100 of cardinality vectors in a convolution sense : @xmath101 is defined as @xmath102 .
it is possible to consider an even more general csp with global cardinality constraints , in which every instance of @xmath79 is accompanied with a set of global cardinality constraints , and the question is whether or not there exists a solution of the csp instance that satisfies one of the cardinality constraints . sometimes such a set of cardinality constraints can be represented concisely , for example , all constraints @xmath2 with @xmath103 .
we denote such _ extended csp with global cardinality constraints _ corresponding to a constraint language @xmath1 by @xmath104 .
[ exa : extended - coloring ] the problem @xmath105 admits a wide variety of questions , e.g. whether a given graph admits a 3-coloring with 25 vertices colored 0 , and odd number of vertices colored 1 .
however , in our setting ( as @xmath21 is a fixed constant and we are investigating polynomial - time solvability ) the extended problems are not very interesting from the complexity point of view . [
lem : eccsp - to - ccsp ] for any constraint language @xmath1 the problem @xmath104 is turing reducible to @xmath0 .
since @xmath3 is fixed , for any instance @xmath85 of @xmath79 there are only polynomially many global cardinality constraints .
thus we can try each of the cardinality constraints given in an instance of @xmath104 in turn .
note that the algorithm in section [ sec : algorithm ] for @xmath22 actually finds the set of all feasible cardinality constraints .
thus @xmath106 can be solved in a more direct way than the reduction in lemma [ lem : eccsp - to - ccsp ] .
let @xmath1 be a constraint language on a set @xmath3 .
a relation @xmath44 is _ primitive positive _ _ ( pp- ) _ _ definable _ in @xmath1 if it can be expressed using ( a ) relations from @xmath1 , ( b ) conjunction , ( c ) existential quantifiers , and ( d ) the binary equality relations ( see , e.g. @xcite ) .
the set of all relations pp - definable in @xmath1 will be denoted by @xmath54 .
[ exa : product ] an important example of pp - definitions that will be used throughout the paper is the _ product _ of binary relations .
let @xmath107 be binary relations .
then @xmath108 is the binary relation given by @xmath109 in this paper we will need a slightly weaker notion of definability .
we say that @xmath44 is _ pp - definable _ in @xmath1 _ without equalities _ if it can be expressed using only items ( a)(c ) from above .
the set of all relations pp - definable in @xmath1 without equalities will be denoted by @xmath110 . clearly , @xmath111 .
the two sets are different only on relations with redundancies .
let @xmath44 be a ( say , @xmath62-ary ) relation .
a _ redundancy _ of @xmath44 is a pair @xmath112 of its coordinate positions such that , for any @xmath113 , @xmath114=\ba[j]$ ] .
[ exa : redundancy ] in some cases if a relation @xmath44 has redundancies , the equality relation is pp - definable in @xmath115 without equalities .
let @xmath44 be a ternary relation on @xmath90 given by ( triples , members of the relation , are written vertically ) @xmath116 then the equality relation is expressed by @xmath117 .
in other cases the equality relation can not be expressed that easily , but its restriction onto a subset of @xmath3 can .
let @xmath118 be a 4-ary on @xmath90 given by @xmath119 then the formula @xmath120 defines the equality relation on @xmath121 .
[ lem : redundancy ] for every constraint language @xmath1 , every @xmath122 without redundancies belongs to @xmath110 . consider a pp - definition of @xmath44 in @xmath1 .
suppose that the definition contains an equality relation on the variables @xmath18 and @xmath19 .
if none of @xmath18 and @xmath19 is bound by an existential quantifier , then the relation @xmath44 has two coordinates that are always equal , i.e. , @xmath44 is redundant .
thus one of the variables , say @xmath18 , is bound by an existential quantifier . in this case , replacing @xmath18 with @xmath19 everywhere in the definition defines the same relation @xmath44 and decreases the number of equalities used .
repeating this step , we can arrive to an equality - free definition of @xmath44 .
a _ polymorphism _ of a ( say , @xmath62-ary ) relation @xmath44 on @xmath3 is a mapping @xmath123 for some @xmath11 such that for any tuples @xmath124 the tuple @xmath125\zd\ba_k[1])\zd f(\ba_1[n]\zd\ba_k[n]))\ ] ] belongs to @xmath44 .
operation @xmath57 is a polymorphism of a constraint language @xmath1 if it is a polymorphism of every relation from @xmath1 .
there is a tight connection , a _ galois correspondence _
, between polymorphisms of a constraint language and relations pp - definable in the language , see @xcite . this connection has been extensively exploited to study the ordinary constraint satisfaction problems @xcite .
here we do not need the full power of this galois correspondence , we only need the following folklore result : [ lem : galois ] if operation @xmath57 is a polymorphism of a constraint language @xmath1 , then it is also a polymorphism of any relation from @xmath54 , and therefore of any relation from @xmath110 . for a ( say , @xmath62-ary ) relation @xmath44 over a set @xmath3 and a subset @xmath126 , by @xmath127 we denote the relation @xmath128 . for a constraint language @xmath1 over @xmath3
we use @xmath129 to denote the constraint language @xmath130 .
if @xmath57 is a polymorphism of a constraint language @xmath1 over @xmath3 and @xmath131 , then @xmath57 is not necessarily a polymorphism of @xmath129 .
however , it remains a polymorphism in the following special case .
a @xmath11-ary polymorphism @xmath57 is _ conservative , _ if @xmath132 for every @xmath133 .
it is easy to see that if @xmath57 is a conservative polymorphism of @xmath1 , then @xmath57 is a ( conservative ) polymorphism of @xmath129 for every @xmath134 .
polymorphisms help to express many useful properties of relations .
a ( say , @xmath62-ary ) relation @xmath44 is said to be _
2-decomposable _ if @xmath113 if and only if , for any @xmath135 , @xmath136 , see @xcite .
decomposability sometimes is a consequence of the existence of certain polymorphisms . a ternary operation @xmath40 on a set @xmath3 is said to be a _ majority operation _ if it satisfies equations @xmath41 for all @xmath137 . by @xcite
if a majority operation @xmath40 is a polymorphism of a constraint language @xmath1 then @xmath1 is 2-decomposable .
the converse is not true : there are 2-decomposable relations not having a majority polymorphism .
furthermore , 2-decomposability is not preserved by pp - definitions , thus we can not expect to characterize it by polymorphisms .
[ exa:2decomp ] consider the disequality relation @xmath138 over the set @xmath139 .
relation @xmath138 is trivially 2-decomposable , since it is binary .
let @xmath140 .
the binary projections of @xmath28 are @xmath141 , but @xmath28 is not @xmath142 : it does not allow that @xmath18 , @xmath19 , @xmath143 are all different . a binary relation @xmath44 is said to be _ rectangular _ if for any @xmath45
, the pair @xmath144 also belongs to @xmath44 .
rectangular relations and their generalizations play a very important role in the study of csp @xcite . a useful way to think about binary rectangular relations
is to represent them as _
thick mappings_. a binary relation @xmath75 is called a thick mapping if there are equivalence relations @xmath49 and @xmath48 on @xmath73 and @xmath74 , respectively , and a one - to - one mapping @xmath145 ( thus , in particular , @xmath146 ) such that @xmath147 if and only if @xmath148 . in this case
we shall also say that @xmath44 is a _ thick mapping with respect to _ @xmath49 , [ page : with - respect ] @xmath48 , and @xmath88 .
given a thick mapping @xmath44 the corresponding equivalence relations will be denoted by @xmath149 and @xmath150 . observe that @xmath151 and @xmath152 ; therefore @xmath153 .
thick mapping @xmath44 is said to be _ trivial _ if both @xmath149 and @xmath150 are the total equivalence relations @xmath154 and @xmath155 . in a graph - theoretical point of view , a thick mapping defines a bipartite graph where every connected component is a complete bipartite graph .
as with decomposability , rectangularity follows from the existence of a certain polymorphism . a ternary operation @xmath42 is said to be _ maltsev _ if @xmath156 for all @xmath137 .
the first part of following lemma is straightforward , while the second part is folklore [ lem : thick - mapping ] ( 1 ) binary relation @xmath44 is a thick mapping if and only if it is rectangular .
+ ( 2 ) if a binary relation @xmath44 has a maltsev polymorphism then it is rectangular .
a ternary operation @xmath42 satisfying equations @xmath157 for all @xmath137 is said to be a _
minority operation_. observe that every maltsev operation is minority , but not the other way round .
let us fix a constraint language @xmath1 on a set @xmath3 and let @xmath80 be an instance of @xmath79 .
a _ partial solution _ of @xmath85 on a set of variables @xmath158 is a mapping @xmath159 that satisfies every constraint @xmath160 where @xmath161 . here
@xmath162 denotes the subtuple of @xmath84 consisting of those entries of @xmath84 that belong to @xmath163 , and we consider the coordinate positions of @xmath44 indexed by variables from @xmath84 .
instance @xmath85 is said to be _ @xmath11-consistent _
if for any @xmath11-element set @xmath158 and any @xmath164 any partial solution on @xmath163 can be extended to a partial solution on @xmath165 , see @xcite . as we only need @xmath166 ,
all further definitions are given under this assumption .
any instance @xmath80 can be transformed to a 2-consistent instance by means of the standard 2-consistency algorithm .
this algorithm works as follows .
first , for each pair @xmath167 it creates a constraint @xmath168 where @xmath169 is the binary relation consisting of all partial solutions @xmath170 on @xmath171 , i.e. @xmath169 includes pairs @xmath172 .
these new constraints are added to @xmath81 , let the resulting instance be denoted by @xmath173 .
second , for each pair @xmath167 , every partial solution @xmath174 , and every @xmath175 , the algorithm checks if @xmath170 can be extended to a partial solution of @xmath176 on @xmath177 .
if not , it updates @xmath176 by removing @xmath170 from @xmath169 .
the algorithm repeats this step until no more changes happen . [
lem : consistency ] let @xmath80 be an instance of @xmath79 .
+ ( a ) the problem obtained from @xmath85 by applying 2-consistency is 2-consistent ; + ( b ) on every step of 2-consistency for any pair @xmath167 the relation @xmath169 belongs to @xmath110 .
\(a ) follows from @xcite .
\(b ) since after the first phase of the algorithm every relation @xmath169 is an intersection of unary and binary projections of relations from @xmath1 , it belongs to @xmath110 .
then when considering a pair @xmath167 and @xmath175 , the relation @xmath169 is replaced with @xmath178 , where @xmath118 is the set of all solution of the current instance on @xmath177 .
as every relation of the current instance belongs to @xmath110 , the relation @xmath118 is pp - definable in @xmath1 without equalities .
thus the updated relation @xmath169 also belongs to @xmath110 .
note that theorem [ lem : consistency](b ) implies that any polymorphism of @xmath179 is also a polymorphism of every @xmath180 .
if @xmath1 has a majority polymorphism , by theorem 3.5 of @xcite , every 2-consistent problem instance is _ globally consistent _ , that is every partial solution can be extended to a global solution of the problem .
in particular , if @xmath85 is 2-consistent , then for any @xmath167 , any pair @xmath181 can be extended to a solution of @xmath85 .
the same is true for any @xmath182 and any @xmath183 .
in this section , we prove that the two characterizations in theorems [ the : main - poly ] and [ the : main - poly2 ] are equivalent .
recall that two equivalence relations @xmath49 and @xmath48 over the same domain _ cross _ , if there is an @xmath49-class @xmath73 and a @xmath48-class @xmath74 such that none of @xmath184 , @xmath185 , and @xmath186 is empty .
a constraint language @xmath1 is said to be _ non - crossing decomposable _ if every relation from @xmath54 is 2-decomposable , every binary relation from @xmath54 is a thick mapping , and no pair of equivalence relations from @xmath54 cross .
one of the directions is easy to see .
[ lem : equiv0 ] if @xmath1 has a majority polymorphism @xmath40 and a conservative maltsev polymorphism @xmath42 , then @xmath1 is non - crossing decomposable . as @xmath40 is a polymorphism of every relation in @xmath187 , by @xcite
every relation in @xmath187 is 2-decomposable .
similarly , every binary relation in @xmath187 is invariant under the maltsev polymorphism , which , by lemma [ lem : thick - mapping ] , implies that the binary relation is a thick mapping .
finally , suppose that there are two equivalence relations @xmath188 over the same domain @xmath189 that cross .
this means that for some @xmath190 we have that @xmath191 , @xmath192 , but @xmath193 , @xmath194 .
let @xmath42 be a maltsev polymorphism of @xmath1 , and consider @xmath195 .
first of all , as @xmath42 is conservative , @xmath196 . then , since @xmath42 preserves @xmath49 and @xmath48 , @xmath197 which is impossible . to prove the other direction of the equivalence , we need to construct the two polymorphisms .
the following definition will be useful for this purpose . given a constraint language @xmath1 , we say that @xmath198 is true if @xmath54 contains an equivalence relation @xmath49 with @xmath199 in other words , the domain of the equivalence relation contains all three elements , @xmath200 and @xmath201 are in the same class , but @xmath8 is in a different class .
[ lem : triple ] let @xmath1 be a non - crossing decomposable constraint language over @xmath3 . for every @xmath202 , at most one of @xmath198 , @xmath203
, @xmath204 is true .
suppose that , say , both @xmath198 and @xmath203 are true ; let @xmath205 be the corresponding equivalence relations from @xmath187 .
let @xmath206 , @xmath207 be the domains of @xmath49 and @xmath48 , respectively .
we can consider @xmath206 and @xmath207 as unary relations , and they are in @xmath54 .
therefore , @xmath208 and @xmath209 and @xmath210 are also in @xmath54 . as is easily seen , @xmath211 and @xmath212 are over the same domain @xmath189 and they cross , a contradiction .
[ lem : triple2 ] let @xmath1 be a non - crossing decomposable constraint language over @xmath3 .
let @xmath213 be a binary relation such that @xmath214 , but @xmath215 for some @xmath216 and @xmath217 .
then @xmath218 if , say , @xmath8 and @xmath200 are in the same @xmath219 class , then @xmath201 has to be in a different class . since @xmath28 is a thick mapping , this means that @xmath220 and @xmath221 are in the same @xmath222 class and @xmath223 is in a different class
. therefore , both @xmath204 and @xmath224 are true , and by lemma [ lem : triple ] , none of the other statements can be true .
therefore , we can assume that @xmath8 , @xmath200 , @xmath201 are in different @xmath225 classes , and hence @xmath220 , @xmath221 , @xmath223 are in different @xmath222 classes .
suppose that @xmath198 is true ; let @xmath226 be the corresponding equivalence relation .
consider the relation @xmath227 , that is , @xmath228 , which has to be a thick mapping .
let @xmath229 be the equivalence relations of @xmath230 .
observe that @xmath231 .
we claim that @xmath232 and @xmath233 , showing that @xmath234 is true .
it is clear that @xmath235 : as @xmath236 , we have @xmath237 . to get a contradiction suppose that @xmath238 . since @xmath239 and @xmath230 is a thick mapping , the pairs @xmath240 has to belong to @xmath230 as well . that is , there are @xmath241 such that @xmath242 and @xmath243 .
now equivalence relation @xmath49 shows that @xmath244 is true ( since @xmath242 and @xmath245 ) and equivalence relation @xmath246 shows that @xmath247 is true ( since @xmath248 shows @xmath249 , @xmath250 shows @xmath251 , and we know that @xmath252 ) . by lemma [ lem : triple ] , @xmath244 and @xmath247 can not be both true , a contradiction .
let @xmath1 be a non - crossing decomposable language over @xmath3 .
let @xmath253 be @xmath8 if @xmath254
are all different or all the same , otherwise let it be the value that appears only once among @xmath254 .
similarly , let @xmath255 be @xmath8 if @xmath254
are all different or all the same , otherwise let it be the value that appears more than once among @xmath254 .
because of lemma [ lem : triple ] , the following two functions are well defined : @xmath256 [ lem : maltsev ] operations @xmath40 and @xmath42 are conservative majority and minority operations , respectively , and @xmath54 is invariant under @xmath40 and @xmath42 . it is clear that @xmath40 and @xmath42 are conservative . to show that @xmath40 is a majority operation , by definition of major , it is sufficient to consider the case when one of @xmath257 is true . if @xmath258 ( resp . ,
@xmath259 , @xmath260 ) , then only @xmath198 ( resp .
, @xmath203 , @xmath204 ) can be true , which means that @xmath261 ( resp .
, @xmath8 , @xmath200 ) , as required .
a similar argument shows that @xmath42 is a minority function . since every relation in @xmath54 is 2-decomposable , it is sufficient to show that every binary relation in @xmath54 is invariant under @xmath40 and @xmath42 .
we show invariance under @xmath42 , the proof is similar for @xmath40 .
let @xmath213 be a binary relation , which is a thick mapping by assumption .
. if @xmath254 are in the same @xmath149-class then @xmath263 are in the same @xmath150-class .
since @xmath264 and @xmath265 , it follows that @xmath266 . if @xmath254 are not all in the same @xmath149-class , and one of @xmath257 is true , then by lemma [ lem : triple2 ] , the corresponding statement from @xmath267 is also true .
now the pair @xmath268 has to be one of @xmath269 , @xmath270 , @xmath271 , hence it is in @xmath28
. if none of @xmath257 is true , then @xmath254 are in different @xmath272-classes . moreover ,
if none of @xmath234 , @xmath273 is true , then @xmath263 are in different @xmath274-classes . therefore @xmath275 .
[ rem : minority ] interestingly , lemma [ lem : maltsev ] ( along with lemma [ lem : equiv0 ] ) gives more than just the existence of a majority polymorphism , and a conservative maltsev polymorphism .
the operations @xmath40 and @xmath42 are both conservative , and @xmath42 is a minority operation , not just a maltsev one .
therefore , we have that a constraint language has a majority and conservative maltsev polymorphisms if and only if it has a majority and minority polymorphisms , both conservative
. the following two consequences of having a conservative maltsev polymorphism will be used in the algorithm .
[ lem : transitive ] let @xmath1 be a constraint language having a conservative maltsev polymorphism .
let @xmath276 be two nontrivial thick mappings such that @xmath277 .
then @xmath278 is also non - trivial . since @xmath44 , @xmath230
are nontrivial , there are values @xmath8 ,
@xmath220 , @xmath200 , @xmath221 such that @xmath279 , @xmath280 . if @xmath278 is trivial , then @xmath281 , which means that there are ( not necessarily distinct ) values @xmath282 such that @xmath283 let @xmath40 be a conservative maltsev polymorphism .
let @xmath284 .
applying @xmath40 on the pairs above , we get that @xmath285 and @xmath286 .
it is not possible that @xmath287 , since this would mean that @xmath288 .
it is not possible that @xmath289 either , since that would imply @xmath290 , a contradiction .
recall that if @xmath47 and @xmath48 are equivalence relations on the same set @xmath291 , then @xmath292 denotes the smallest ( in terms of the number of pairs it contains ) equivalence relation on @xmath291 containing both @xmath47 and @xmath48 .
it is well known that if @xmath49 and @xmath48 have a maltsev polymorphism then they _ permute _ , that is , @xmath293 .
[ lem : non - trivial ] let @xmath1 be a finite constraint language having a conservative maltsev polymorphism .
let @xmath294 be two nontrivial equivalence relations on the same set @xmath291 .
then @xmath295 is also non - trivial . since @xmath296 , by lemma [ lem : transitive ] , it is non - trivial .
we will also need the following observation .
[ lem : join - union ] let @xmath1 be a finite constraint language having a majority polymorphism and a conservative maltsev polymorphism .
let @xmath294 be two equivalence relations on the same set @xmath291
. then @xmath297 . by lemma [ lem : equiv0 ]
@xmath49 and @xmath48 are non - crossing
. then if @xmath191 and @xmath192 then @xmath298 . the result follows .
in this section we fix a constraint language @xmath1 that has conservative majority and minority polymorphisms .
we present a polynomial - time algorithm for solving @xmath0 and @xmath55 in this case . in this subsection
we prove several properties of instances of @xmath0 and @xmath55 that will be very instrumental for our algorithms .
first of all , we show that every such instance can be supposed binary , that is , that every its constraint is imposed only on two variables .
then we introduce a graph corresponding to such an instance , and show that if this graph is disconnected then a solution to the whole problem can be obtained by combining arbitrarily solutions for the connected components .
finally , if the graph is connected , the set of possible values for each variable can be subdivided into several subsets , so that if the variable takes a value from one of the subsets , then each of the remaining variables is forced to take values from a particular subset of the corresponding partition .
observe that if a constraint language @xmath1 satisfies the conditions of theorem [ the : main - poly ] then by remark [ rem : minority ] the constraint language @xmath299 obtained from @xmath1 by adding all unary relations also satisfies the conditions of theorem [ the : main - poly ] .
indeed , @xmath1 has conservative majority and minority polymorphisms that are also polymorphisms of @xmath299 .
therefore we will assume that @xmath1 contains all unary relations .
it will also be convenient to assume that @xmath1 contains all the binary relations from @xmath54 .
let @xmath1 be a constraint language and let @xmath80 be a 2-consistent instance of @xmath79 .
by @xmath300 we denote the instance @xmath301 such that @xmath302 is the set of all constraints of the form @xmath168 where @xmath167 and @xmath169 is the set of all partial solutions on @xmath171 . [
lem : binary - solutions ] let @xmath1 be a constraint language with a majority polymorphism
. then if @xmath85 is a 2-consistent instance of @xmath79 then @xmath300 has the same solutions as @xmath85 .
let us denote by @xmath303 the @xmath304-ary relations consisting of all solutions of @xmath85 and @xmath300 , respectively .
relations @xmath303 are pp - definable in @xmath1 without equalities , and @xmath305 . to show that @xmath306 we use the result from @xcite stating that , since @xmath1 has a majority polymorphism for any @xmath167 and any @xmath181 we have @xmath307 , i.e. @xmath308 .
since @xmath44 is 2-decomposable , if @xmath309 , that is , @xmath310 for all @xmath167 , then @xmath113 .
let @xmath80 be an instance of @xmath79 .
applying algorithm 2-consistency we may assume that @xmath85 is 2-consistent . by the assumption about @xmath1 , all constraint relations from @xmath85 are 2-decomposable , and @xmath300 has the same solutions as @xmath85 itself .
therefore , replacing @xmath85 with @xmath300 , if necessary , every constraint of @xmath85 can be assumed to be binary .
let constraints of @xmath85 be @xmath311 for each pair of different @xmath167 .
let @xmath312 , @xmath313 , denote the set of @xmath7 such that there is a solution @xmath88 of @xmath85 such that @xmath314 .
by @xcite , @xmath85 is globally consistent , therefore , @xmath315 for any @xmath316 , @xmath317 .
constraint @xmath318 is said to be _ trivial _ if @xmath319 , otherwise it is said to be _ non - trivial_. the _ graph of _
@xmath85 , denoted @xmath320 , is a graph with vertex set @xmath4 and edge set @xmath321 .
the 2-consistency of @xmath85 implies , in particular , the following simple property .
[ lem:2-consistency ] by the 2-consistency of @xmath85 , for any @xmath322 , @xmath323 .
therefore , by lemma [ lem : transitive ] , the graph @xmath320 is transitive , i.e. , every connected component is a clique . if @xmath320 is not connected , every combination of solutions for its connected components give rise to a solution of the entire problem .
more precisely , let @xmath324 be the connected components of @xmath320 , and let @xmath325 denote the instance @xmath326 where @xmath327 includes all the constraints @xmath311 for which @xmath328 .
we will use the following observation .
[ lem : disconnected ] let @xmath329 be solutions of @xmath330 . then the mapping @xmath331 such that @xmath332 whenever @xmath333 is a solution of @xmath85 .
we need to check that all constraints of @xmath85 are satisfied .
consider @xmath318 . if @xmath328 for a certain @xmath17 , then @xmath334 since @xmath335 is a solution to @xmath336 .
if @xmath337 belong to different connected components , then @xmath338 is trivial , and so @xmath339 is satisfied .
suppose that @xmath320 is connected and fix @xmath313 .
in this case , the graph is a clique , and therefore for any @xmath316 the constraint @xmath339 is non - trivial .
note that due to 2-consistency , every relation @xmath340 for @xmath341 is over the same set @xmath312 .
set @xmath342 ; as every @xmath343 is non - trivial , lemma [ lem : non - trivial ] implies that @xmath344 is non - trivial .
[ lem : variable - non - trivial ] suppose @xmath320 is connected .
equivalence relations @xmath344 and @xmath340 ( for any @xmath345 ) are non - trivial . [
lem : partition ] suppose @xmath320 is connected .
+ ( 1 ) for any @xmath167 there is a one - to - one correspondence @xmath346 between @xmath347 and @xmath348 such that for any solution @xmath88 of @xmath85 if @xmath349 , then @xmath350 . + ( 2 ) the mappings @xmath346 are consistent , i.e. for any @xmath351 we have @xmath352 for every @xmath18 .
\(1 ) let @xmath338 be a thick mapping with respect to a mapping @xmath353 , and @xmath354 , @xmath355 .
recall that @xmath353 is a one - to - one mapping from @xmath356 to @xmath357 .
suppose , to obtain a contradiction , that @xmath353 does not induce a one - to - one mapping between @xmath347 and @xmath348 .
then without loss of generality there are @xmath358 such that @xmath359 , but for certain @xmath360 we have @xmath361 and @xmath362 . since @xmath363 , @xmath364 , hence @xmath365 . by lemma [ lem
: join - union ] there is @xmath366 such that @xmath367 is a thick mapping with respect to some @xmath368 and @xmath369 .
therefore for some @xmath370 we have @xmath371 . since @xmath372 ,
there exist @xmath373 satisfying the conditions @xmath374 and @xmath375 .
the first pair of inclusions imply that @xmath376 , while the second one implies that @xmath377 . since @xmath363 , we obtain @xmath378 , a contradiction .
\(2 ) if for some @xmath322 there is a class @xmath379 such that @xmath380 then @xmath381 , a contradiction . fix a variable @xmath382 of @xmath85 and take a @xmath383-class @xmath73 .
let @xmath384 denote the problem instance over the same variables , where for every @xmath167 the set @xmath385 includes the constraint @xmath386 with @xmath387 .
[ lem : unary restrictions ] problem @xmath388 belongs to @xmath0 .
it suffices to show that @xmath389 for any @xmath167 . by lemma [ lem : consistency ] @xmath390 , and as we assumed that @xmath1 contains all binary relations from @xmath54 , we have @xmath391 . by the assumption made
, all unary relations including @xmath392 and @xmath393 belong to @xmath1 .
therefore relation @xmath394 is pp - definable in @xmath1 , and , as a binary relation , belongs to it .
we split the algorithm into two parts .
algorithm cardinality ( figure [ fig : alg - cardinality ] ) just ensures 2-consistency and initializes a recursive process .
the main part of the work is done by ext - cardinality ( figure [ fig : alg - vector ] ) .
algorithm ext - cardinality solves the more general problem of computing the set of all cardinality constraints @xmath2 that can be satisfied by a solution of @xmath85 .
thus it can be used to solve directly csp with extended global cardinality constraints , see preliminaries .
the algorithm considers three cases .
step 2 handles the trivial case when the instance consists of a single variable and there is only one possible value it can be assigned .
otherwise , we decompose the instance either by partitioning the variables or by partitioning the domain of the variables .
if @xmath320 is not connected , then the satisfying assignments of @xmath85 can be obtained from the satisfying assignments of the connected components .
thus a cardinality constraint @xmath2 can be satisfied if it arises as the sum @xmath395 of cardinality constraints such that the @xmath17-th component has a solution satisfying @xmath396 . instead of considering all such sums ( which would not be possible in polynomial time )
, we follow the standard dynamic programming approach of going through the components one by one , and determining all possible cardinality constraints that can be satisfied by a solution for the first @xmath17 components ( step 3 ) .
if the graph @xmath320 is connected , then we fix a variable @xmath382 and go through each class @xmath73 of the partition @xmath383 ( step 4 ) . if @xmath382 is restricted to @xmath73 , then this implies a restriction for every other variable @xmath397 .
we recursively solve the problem for the restricted instance @xmath388 arising for each class @xmath73 ; if constraint @xmath2 can be satisfied , then it can be satisfied for one of the restricted instances .
the correctness of the algorithm follows from the discussion above .
the only point that has to be verified is that the instance remains 2-consistent after the recursion .
this is obvious if we recurse on the connected components ( step 3 ) . in step 4
, 2-consistency follows from the fact that if @xmath398 can be extended by @xmath399 , then in every subproblem either these three values satisfy the instance restricted to @xmath177 or @xmath8 , @xmath200 , @xmath201 do not appear in the domain of @xmath39 , @xmath397 , @xmath38 , respectively . to show that the algorithm runs in polynomial time ,
observe first that every step of the algorithm ( except the recursive calls ) can be done in polynomial time . here
we use that @xmath3 is fixed , hence the size of the set @xmath99 is polynomially bounded .
thus we only need to bound the size of the recursion tree .
if we recurse in step 3 , then we produce instances whose graphs are connected , thus it can not be followed by recursing again in step 3 . in step 4 ,
the domain of every variable is decreased : by lemma [ lem : variable - non - trivial ] , @xmath400 is nontrivial for any variable @xmath397 .
thus in any branch of the recursion tree , recursion in step 4 can occur at most @xmath21 times , hence the depth of the recursion tree is @xmath401 .
as the number of branches is polynomial in each step , the size of the recursion tree is polynomial .
input : = an instance @xmath80 of @xmath0 with a cardinality constraint @xmath2 + output : yes if @xmath85 has a solution satisfying @xmath2 , no otherwise + _ step 1 . _ to @xmath85 + _ step 2 .
_ @xmath402ext - cardinality@xmath403 + _ step 3 . _ @xmath404 * output * yes + no input : = a 2-consistent instance @xmath80 of @xmath0 + output : the set of cardinality constraints @xmath2 such that @xmath85 has a solution that satisfies @xmath2 + _ step 1 . _
the graph @xmath405 + _ step 2 .
_ @xmath406 and the domain of this variable is a singleton @xmath407 * then do * + _ step 2.1 _ * set * @xmath408 where @xmath409 except for @xmath410 + _ step 3 .
_ @xmath320 is disconnected and @xmath411 are + its connected components * then do * + _ step 3.1 _ @xmath408 where @xmath5 is given by @xmath412 for @xmath7 + _ step 3.2 _ @xmath413 * to * @xmath11 * do * + _ step 3.2.1 _ * set * @xmath414 ext - cardinality@xmath415 + + + _ step 4 .
_ + _ step 4.1 _ @xmath313 * find * @xmath344 + _ step 4.2 _ @xmath416 * and set * @xmath417 + _ step 4.3 _ @xmath383-class @xmath73 * do * + _ step 4.3.1 _ @xmath418 where for every @xmath167 the set @xmath385 includes + the constraint @xmath419 + _ step 4.3.2 _ @xmath420ext - cardinality@xmath421 + + + _ step 4 .
_ @xmath99 in this section we observe that algorithm cardinality can be modified so that it also solves counting csps with global cardinality constraints , provided @xmath1 satisfies the conditions of theorem [ the : main - poly ] .
the counting algorithm works very similar to algorithm cardinality , except that instead of determining the set of satisfiable cardinality constraints , it keeps track of the number of solutions that satisfy every cardinality constraint possible .
it considers the same 3 cases . in the trivial case of a problem with one variable and one possible value for this variable
, the algorithm assigns 1 to the cardinality constraint satisfied by the only solution of the problem and 0 to all other cardinality constraints . in the case of disconnected graph @xmath320
if a cardinality constraint can be represented in the form @xmath422 , then solutions on the connected components of @xmath320 satisfying @xmath423 , respectively , contribute the product of their numbers into the number of solutions satisfied by @xmath2 .
we again use the dynamic programming approach , and , for each @xmath17 compute the number of solutions on @xmath424 satisfying every possible cardinality constraint .
observe , that the set of cardinality constraints considered is also changed dynamically , as the number of variables grows . finally ,
if @xmath320 is connected , then the different restrictions have disjoint sets of solutions , hence the numbers of solutions are computed independently .
input : = an instance @xmath80 of @xmath55 with a cardinality constraint @xmath2 + output : the number of solutions of @xmath85 that satisfy @xmath2 + _ step 1 . _ to @xmath85 + _ step 2 .
_ @xmath425#ext - cardinality@xmath403 + % @xmath426 is the number of solutions of @xmath85 satisfying cardinality constraint @xmath427 + _ step 3 .
_ @xmath428 + input : = a 2-consistent instance @xmath80 of @xmath55 + output : function @xmath353 that assigns to every cardinality constraint @xmath2 with @xmath6 , + the number @xmath428 of solutions of @xmath85 that satisfy @xmath2 + _ step 1 .
_ the graph @xmath405 + _ step 2 .
_ @xmath406 and the domain of this variable is a singleton @xmath407 * then do * + _ step 2.1 _ * set * @xmath429 where @xmath409 except for @xmath410 , and @xmath430 for all @xmath431 + with @xmath432 + _ step 3 .
_ @xmath320 is disconnected and @xmath411 are its + connected components * then do * + _ step 3.1 _
@xmath408 where @xmath5 is given by @xmath412 for @xmath7 , @xmath429 for @xmath404 + _ step 3.2 _ @xmath413 * to * @xmath11 * do * + _ step 3.2.1 _ * set * @xmath433 and @xmath434 # ext - cardinality@xmath415 + _ step 3.2.2 _ @xmath435 , @xmath436 for @xmath437 + _ step 3.2.3 _ @xmath404 * and * @xmath438 * set * @xmath439 + _ step 3.2.4 _ @xmath440 , @xmath441 + + + _ step 4 . _
+ _ step 4.1 _ @xmath313 * find * @xmath344 + _ step 4.2 _ @xmath416 * and set * @xmath442 for @xmath2 with @xmath6 + _ step 4.3 _
@xmath383-class @xmath73 * do * + _ step 4.3.1 _ @xmath418 where for every @xmath167 the set @xmath385 includes the constraint + @xmath419 + _ step 4.3.2 _ @xmath434#ext - cardinality@xmath421 + _ step 4.3.3 _
@xmath443 + + + _ step 4 .
we present two reductions that will be crucial for the proofs in section [ sec : hardness ] . in section [ sec : pp - definitions ] , we show that adding relations that are pp - definable ( without equalities ) does not make the problem harder , while in section [ sec : constants ] , we show the same for unary constant relations .
[ the : decision - pp - defin ] let @xmath1 be a constraint language and @xmath44 a relation pp - definable in @xmath1 without equalities .
then @xmath444 is polynomial time reducible to @xmath0 .
we proceed by induction on the structure of pp - formulas .
the base case of induction is given by @xmath58 .
we need to consider two cases .
@xmath445 .
observe that by introducing ` fictitious ' variables for predicates @xmath446 we may assume that both relations involved have the same arity .
a reduction from @xmath444 to @xmath0 is trivial : in a given instance of the first problem replace each constraint of the form @xmath447 with two constraints @xmath448 and @xmath449 .
@xmath450 .
let @xmath80 be a @xmath444 instance .
without loss of generality let @xmath451 be the constraints that involve @xmath44 .
instance @xmath176 of @xmath0 is constructed as follows . 1 .
variables : replace every variable @xmath39 from @xmath4 with a set @xmath452 of variables of size @xmath453 and introduce a set of @xmath21 variables for each constraint involving @xmath28 . more formally , @xmath454 2 .
non-@xmath44 constraints : for every @xmath455 with @xmath456 , introduce all possible constraints of the form @xmath457 , where @xmath458 for @xmath459 .
@xmath44 constraints : for every @xmath460 , @xmath461 , introduce all possible constraints of the form @xmath462 , where @xmath458 , @xmath463 .
if @xmath85 has a solution satisfying cardinality constraint @xmath2 then @xmath176 has a solution satisfying the cardinality constraint @xmath464 .
let @xmath88 be a solution of @xmath85 satisfying @xmath2 .
it is straightforward to verify that the following mapping @xmath170 is a solution of @xmath176 and satisfies @xmath427 : 1 .
for each @xmath313 and each @xmath465 set @xmath466 ; 2 . for each @xmath467 , where @xmath468 , set @xmath469 to be a value such that@xmath470 .
3 . for each @xmath461 and @xmath471 set @xmath472 to be such that @xmath473 . claim 2 .
if @xmath176 has a solution @xmath170 satisfying the cardinality constraint @xmath464 , then @xmath85 has a solution satisfying constraint @xmath2 .
let @xmath7 and @xmath474 .
observe first that if @xmath475 is a mapping such that @xmath476 for every @xmath313 ( i.e. , @xmath477 for at least one variable @xmath478 ) , then @xmath88 satisfies all the constraints of @xmath85 . indeed , consider a constraint @xmath479 of @xmath85 where @xmath480 .
let @xmath481 .
for every @xmath482 , there is a @xmath483 such that @xmath484 . by the way @xmath176 is defined
, it contains a constraint @xmath485 where @xmath486 .
now the fact that @xmath170 satisfies @xmath487 immediately implies that @xmath88 satisfies @xmath50 : @xmath488 .
the argument is similar if @xmath489 .
we show that it is possible to construct such a @xmath88 that also satisfies the cardinality constraint @xmath2 .
since @xmath490 , for any @xmath7 with @xmath491 , even if set @xmath492 contains all @xmath453 variables of the form @xmath467 and @xmath493 , it has to intersect at least @xmath9 sets @xmath452 ( as @xmath494 ) . consider the bipartite graph @xmath495 , where @xmath496 is a bipartition and 1 .
@xmath497 is the set of variables @xmath4 ; 2 .
@xmath498 is constructed from the set @xmath3 of values by taking @xmath9 copies of each value @xmath7 ; 3 .
edge @xmath499 , where @xmath220 is a copy of @xmath8 from @xmath498 , belongs to @xmath500 if and only if @xmath501 .
note that @xmath502 and a perfect matching @xmath503 corresponds to a required mapping @xmath88 : @xmath314 if @xmath504 for some copy @xmath220 or @xmath8 .
take any subset @xmath505 , let @xmath291 contains some copies of @xmath506 . then by the observation above , @xmath291 has at least @xmath507 neighbours in @xmath497 .
since @xmath291 contains at most @xmath508 copies of @xmath509 , @xmath510 by hall s theorem on perfect matchings in bipartite graphs , @xmath511 has a perfect matching , concluding the proof that the required @xmath88 exists .
let @xmath3 be a set , and let @xmath7 .
the _ constant relation _
@xmath512 is the unary relation that contains only one tuple , @xmath513 .
if a constraint language @xmath1 over @xmath3 contains all the constant relations , then they can be used in the corresponding constraint satisfaction problem to force certain variables to take some fixed values .
the goal of this section is to show that for any constraint language @xmath1 the problem @xmath514 is polynomial time reducible to @xmath0 . for the ordinary decision csp
such a reduction exists when @xmath1 does not have unary polymorphisms that are not permutations , see @xcite .
we make use of the notion of multi - valued morphisms , a generalization of homomorphisms , that in a different context has appeared in the literature for a while ( see , e.g. @xcite ) under the guise hyperoperation .
let @xmath44 be a ( say , @xmath62-ary ) relation on a set @xmath3 , and let @xmath57 be a mapping from @xmath3 to @xmath515 , the powerset of @xmath3 .
mapping @xmath57 is said to be a _
multi - valued morphism _ of @xmath44 if for any tuple @xmath516 the set @xmath517 is a subset of @xmath44 .
mapping @xmath57 is a multi - valued morphism of a constraint language @xmath1 if it is a multi - valued morphism of every relation in @xmath1 . for a multi - valued morphism @xmath57 and set @xmath518
, we define @xmath519 .
the product of two multi - valued morphisms @xmath520 and @xmath521 is defined by @xmath522 for every @xmath7 .
we denote by @xmath523 the @xmath17-th power of @xmath57 , with the convention that @xmath524 maps @xmath8 to @xmath407 for every @xmath525 .
[ the : adding - constants ] let @xmath1 be a finite constraint language over a set @xmath3 .
then @xmath514 is polynomial time reducible to @xmath0 .
let @xmath526 and @xmath527 .
we show that @xmath528 is polynomial time reducible to @xmath0 .
this clearly implies the result .
we make use of the following multi - valued morphism gadget @xmath529 ( i.e. a csp instance ) .
observe that it is somewhat similar to the _ indicator problem _ @xcite . 1 .
the set of variables is @xmath530 , where @xmath531 contains @xmath532 elements .
all sets @xmath531 are assumed to be disjoint .
the set of constraints is constructed as follows : for every ( say , @xmath533-ary ) @xmath58 and every @xmath534 we include all possible constraints of the form @xmath535 where @xmath536 for @xmath537 .
now , given an instance @xmath80 of @xmath528 , we construct an instance @xmath538 of @xmath0 . 1 .
let @xmath158 be the set of variables @xmath39 , on which the constant relation @xmath512 is imposed , that is , @xmath81 contains the constraint @xmath539 .
set @xmath540 .
the set @xmath541 of variables of @xmath176 is the disjoint union of the set @xmath542 of variables of @xmath529 and @xmath543 .
the set @xmath302 of constraints of @xmath176 consists of three parts : 1 .
@xmath544 , the constraints of @xmath529 ; 2 .
@xmath545 , the constraints of @xmath85 that do not include variables from @xmath163 ; 3 .
@xmath546 , for any constraint @xmath547 whose scope contains variables constrained by @xmath512 ( without loss of generality let @xmath548 be such variables ) , @xmath546 contains all constraints of the form @xmath549 , where @xmath550 .
we show that @xmath85 has a solution satisfying a cardinality constraint @xmath2 if and only if @xmath176 has a solution satisfying cardinality constraint @xmath427 given by @xmath551 suppose that @xmath85 has a right solution @xmath88 .
then a required solution for @xmath176 is given by @xmath552 it is straightforward that @xmath170 is a solution to @xmath176 and that it satisfies @xmath427 .
suppose that @xmath176 has a solution @xmath170 that satisfies @xmath427 .
since @xmath553 , there is @xmath554 such that @xmath555 .
thus the assignment @xmath556 is a satisfying assignment @xmath85 , but it might not satisfy @xmath2 .
our goal is to show that @xmath176 has a solution @xmath170 , where @xmath88 obtained this way satisfies @xmath2 .
observe that what we need is that @xmath170 assigns value @xmath557 to exactly @xmath558 variables of @xmath543 .
mapping @xmath57 taking every @xmath559 to the set @xmath560 is a multi - valued morphism of @xmath1 .
indeed , let @xmath561 , @xmath44 is an ( @xmath62-ary ) relation from @xmath1
. then by the construction of @xmath529 the instance contains all the constraints of the form @xmath447 with @xmath562 , @xmath563 .
therefore , @xmath564 claim 2 .
let @xmath57 be the mapping defined in claim 1 .
then @xmath565 defined by @xmath566 for every @xmath567 is also a multi - valued morphism of @xmath1 .
we show that for every @xmath559 , there is an @xmath568 such that @xmath569 for every @xmath570 . taking the maximum @xmath571 of all these integers
, we get that @xmath572 and @xmath573 ( since @xmath574 ) for every @xmath17 , proving the claim .
the proof is by induction on @xmath17 .
if @xmath575 , then we are done as we can set @xmath576 ( note that this is always the case for @xmath413 , since we observed above that @xmath170 has to assign value @xmath577 to a variable of @xmath578 ) .
so let us suppose that @xmath579 .
let @xmath580 and let @xmath581 be defined by @xmath582 .
observe that @xmath583 is well - defined , i.e. , @xmath584 : the set @xmath585 contains @xmath586 variables , while the number of variables which are assigned by @xmath170 values outside @xmath587 is @xmath588 .
let @xmath589 .
we claim that @xmath590 .
suppose that @xmath591 . by the definition of @xmath592 and the assumption @xmath579 , for every @xmath593 , the variables in @xmath594 can have values only from @xmath592 and from @xmath595 . the total number of variables in @xmath594 , @xmath596 is @xmath597 , while the total cardinality constraint of the values from @xmath598 is @xmath599 a contradiction .
thus @xmath590 , that is , there is a value @xmath600 such that @xmath601 and @xmath602 for some @xmath603 . by the induction hypothesis , @xmath604 for every @xmath605
, thus we have that @xmath606 if @xmath62 is at least @xmath607 .
this concludes the proof of claim 2 .
let @xmath608 ( resp . ,
@xmath609 ) be the set of those values @xmath559 that @xmath170 assigns to more than ( resp .
, less than ) @xmath610 variables of @xmath543 .
it is clear that if @xmath611 , then @xmath88 obtained from @xmath170 satisfies @xmath2 .
furthermore , if @xmath612 , then @xmath613 as well .
thus suppose that @xmath614 and let @xmath615 .
@xmath616 .
suppose @xmath617 .
every @xmath618 is assigned by @xmath170 to at least @xmath619 variables in @xmath543 , implying that @xmath170 assigns every such @xmath200 to at most @xmath620 variables in the gadget @xmath529 .
thus the total number of variables in the gadget having value from @xmath291 is at most @xmath621 ; in fact , it is strictly less than that since @xmath608 is not empty . by the definition of @xmath291 ,
only values from @xmath291 can be assigned to variables in @xmath594 for every @xmath622 . however , the total number of these variables is exactly @xmath621 , a contradiction .
by claim 3 , there is a value @xmath623 , which means that there is a @xmath624 and a sequence of distinct values @xmath625 , @xmath626 , @xmath32 , @xmath627 such that @xmath628 for every @xmath629 .
let @xmath630 be an arbitrary variable having value @xmath631 .
we modify @xmath170 the following way : 1 .
the value of @xmath39 is changed from @xmath631 to @xmath632 .
2 . for every @xmath633 , one variable in @xmath634 with value @xmath635
is changed to @xmath636 .
note that since @xmath637 and @xmath638 such a variable exists .
note that these changes do not modify the cardinalities of the values : the second step increases only the cardinality of @xmath625 ( by one ) and decreases only the cardinality of @xmath627 ( by one ) .
we have to argue that the transformed assignment still satisfies the constraints of @xmath176 .
since @xmath639 , the multi - valued morphism @xmath565 of claim 2 implies that changing @xmath631 to @xmath632 on a single variable and not changing anything else also gives a satisfying assignment . to see that the second step does not violate the constraints , observe first that constraints of type ( b ) are not affected and constraints of type ( c ) can not be violated ( since variables in @xmath578 are changed only to @xmath577 , and there is already at least one variable with value @xmath577 in @xmath578 ) . to show that constraints of type ( a ) are not affected ,
it is sufficient to show that the mapping @xmath640 described by the gadget after the transformation is still a multi - valued morphism .
this can be easily seen as @xmath641 , where @xmath565 is the multi - valued morphism of claim 2 .
thus the modified assignment is still a solution of @xmath176 satisfying @xmath427 .
it is not difficult to show that repeating this operation , in a finite number of steps we reach a solution where @xmath642 , i.e. , every value @xmath643 appears exactly @xmath644 times on the variables of @xmath543 .
as observed above , this implies that @xmath85 has a solution satisfying @xmath2 .
now we prove that if constraint language @xmath1 does not satisfy the conditions of theorem [ the : main - poly ] then @xmath0 is np - complete . first , we can easily simulate the restriction to a subset of the domain by setting to 0 the cardinality constraint on the unwanted values : for an instance @xmath173 of @xmath645 with a global cardinality constraint @xmath646 we construct an instance @xmath80 of @xmath0 such that for each @xmath647 we include @xmath648 into @xmath81 .
the cardinality constraint @xmath427 is replaced with @xmath5 such that @xmath649 for @xmath650 , and @xmath412 for @xmath651 .
it is straightforward that @xmath85 has a solution satisfying @xmath2 if and only if @xmath176 has a solution satisfying @xmath427 .
suppose now that a constraint language @xmath1 does not satisfy the conditions of theorem [ the : main - poly ] .
then one of the following cases takes place : ( a ) @xmath54 contains a binary relation that is not a thick mapping ; or ( b ) @xmath54 contains a crossing pair of equivalence relations ; or ( c ) @xmath54 contains a relation which is not 2-decomposable . since relations occurring in cases ( a ) , ( b ) are not redundant , and relations that may occur in case ( c ) can be assumed to be not redundant , by lemma [ lem : redundancy ] @xmath54 can be replaced with @xmath110 .
furthermore , by theorem [ the : adding - constants ] all the constant relations can be assumed to belong to @xmath1 .
we consider these three cases in turn .
furthermore , by a minimality argument , we can assume that if @xmath1 is over @xmath3 , then for every @xmath652 , constraint language @xmath653 satisfies the requirements of theorem [ the : main - poly ] .
one of the np - complete problems we will reduce to @xmath654 is the bipartite independent set problem ( or bis for short ) . in this problem
given a connected bipartite graph @xmath511 with bipartition @xmath655 and numbers @xmath656 , the goal is to verify if there exists an independent set @xmath291 of @xmath511 such that @xmath657 and @xmath658 .
the np - completeness of the problem follows for example from @xcite , which shows the np - completeness of the complement problem under the name constrained minimum vertex cover .
it is easy to see that the problem is hard even for graphs containing no isolated vertices . by representing the edges of a bipartite graph with the relation @xmath659
, we can express the problem of finding a bipartite independent set .
value @xmath200 ( resp . , @xmath8 ) represents selected ( resp . , unselected )
vertices in @xmath660 , while value @xmath201 ( resp . , @xmath661 )
represents selected ( resp . , unselected )
vertices in @xmath662 . with this interpretation ,
the only combination that relation @xmath44 excludes is that two selected vertices are adjacent . by lemma [ lem : thick - mapping ]
, if a binary relation is not a thick mapping , then it contains something very similar to @xmath44 .
however , some of the values @xmath8 , @xmath200 , @xmath201 , and @xmath661 might coincide and the relation might contain further tuples such as @xmath663 .
thus we need a delicate case analysis to show that the problem is np - hard for binary relations that are not thick mappings . since @xmath44 is not a thick mapping , there are @xmath45 such that @xmath665 . by lemma [
lem : restriction ] the problem @xmath666 , where @xmath667 , is polynomial time reducible to @xmath654 . replacing @xmath44 with @xmath230 if necessary we can assume that @xmath44 is a relation over @xmath668 ( note that some of those elements can be equal ) .
we suppose that @xmath44 is a ` smallest ' relation that is not a thick mapping , that is , for any @xmath230 definable in @xmath115 with @xmath669 , the relation @xmath230 is a thick mapping , and for any subset @xmath189 of @xmath3 the restriction of @xmath44 onto @xmath189 is a thick mapping .
let @xmath670 .
since @xmath671 , the unary relation @xmath74 is definable in @xmath44 .
if @xmath672 , by setting @xmath673 we get a binary relation @xmath230 that is not a thick mapping .
thus by the minimality of @xmath44 , we may assume that @xmath674 for any @xmath675 , and symmetrically , @xmath676 for any @xmath677 .
we claim that @xmath679 .
suppose , without loss of generality that @xmath680 appears in @xmath72 . if @xmath681 , then @xmath682 and @xmath665 .
therefore the restriction @xmath683 is not a thick mapping , contradicting the minimality of @xmath44 .
otherwise @xmath684 while @xmath685 .
hence @xmath686 is not a thick mapping .
thus we have @xmath687 and @xmath688 .
let @xmath689 be an instance of bis .
construct an instance @xmath80 by including into @xmath81 , for every edge @xmath690 of @xmath511 , the constraint @xmath691 , and defining a cardinality constraint as @xmath692 , @xmath693 , @xmath694 , @xmath695 .
it is straightforward that for any solution @xmath88 of @xmath85 the set @xmath696 is an independent set , @xmath697 , @xmath698 .
set @xmath699 satisfies the required conditions if and only if @xmath88 satisfies @xmath2 .
conversely , for any independent set @xmath700 mapping @xmath88 given by @xmath701 is a solution of @xmath85 that satisfies @xmath2 if and only if @xmath702 and @xmath703 . let us assume the converse .
let @xmath707 , @xmath708 ( as @xmath44 is over a 3-element set ) .
we consider two cases .
suppose first @xmath709 ( implying @xmath710 ) .
if @xmath685 , then the restriction of @xmath44 onto @xmath711 contains @xmath712 , but does not contain @xmath713 . thus @xmath714 is not a thick mapping , a contradiction with the minimality assumption . if @xmath715 then the set @xmath716 is definable in @xmath44 . observe that @xmath717 is not a thick mapping and definable in @xmath44 . a contradiction with the choice of @xmath44 .
now suppose that @xmath718 ( implying @xmath719 ) .
if @xmath715 , then the restriction @xmath714 is not a thick mapping , as @xmath720 and @xmath665 . otherwise let @xmath685 . by the assumption made @xmath721 , that is , @xmath722 .
we consider 4 cases depending on whether @xmath723 and @xmath724 are contained in @xmath44 . if @xmath725 , then , as @xmath726 , the relation @xmath727 is not a thick mapping ( recall that @xmath728 ) .
if @xmath729 , then we can restrict @xmath44 on @xmath730 ( note that @xmath731 ) .
finally , if @xmath732 [ or @xmath733 , then the relation @xmath734 [ respectively , @xmath735 is definable in @xmath44 .
it remains to observe that @xmath717 is not a thick mapping .
this concludes the proof of the claim .
thus we can assume that one of the projections @xmath71 or @xmath72 contains only 2 elements .
without loss of generality , let @xmath687 . in the remaining cases , we assume @xmath707 , where @xmath680 and @xmath18 may not be present . 1 .
@xmath737 , where all the sets @xmath662 and @xmath738 , @xmath739 , are disjoint , and @xmath740 .
2 . for any
@xmath741 the set @xmath81 contains all constraints of the form @xmath691 where @xmath742 .
the cardinality constraint @xmath2 is given by the following rules : 1 .
subcase 3a : + @xmath694 , @xmath743 , @xmath744 if @xmath745 , and + @xmath694 , @xmath746 , @xmath747 if @xmath748 .
subcase 3b : + @xmath695 , @xmath743 , @xmath749 if @xmath750 , and + @xmath695 , @xmath751 , @xmath747 if @xmath752 .
if @xmath511 has a required independent set @xmath291 , then consider a mapping @xmath753 given by @xmath754 for any @xmath755 , @xmath756 , either @xmath757 or @xmath758 .
in the first case @xmath759 and so @xmath760 . in the second case @xmath761 and @xmath762 .
again , @xmath760 .
finally it is straightforward that @xmath88 satisfies the cardinality constraint @xmath2 .
suppose that @xmath85 has a solution @xmath88 that satisfies @xmath2 .
since @xmath687 and we can assume that @xmath511 has no isolated vertices , for any @xmath756 , @xmath739 , we have @xmath763 . also if for some @xmath756 it holds that @xmath761 and @xmath764 for @xmath765 then @xmath766 .
we include into @xmath700 all vertices @xmath739 such that there is @xmath756 with @xmath761 . by the choice of the cardinality of @xmath738 and @xmath767
there are at least @xmath768 such vertices . in subcase 3a , we include in @xmath291 all vertices @xmath765 with @xmath764 .
there are exactly @xmath769 vertices like this , and by the observation above @xmath291 is an independent set . in subcase 3b , we include in @xmath291 all vertices @xmath765 with @xmath770 . by the choice of @xmath771
, there are at least @xmath769 such vertices . to verify that @xmath291 is an independent set it suffices to recall that in this case @xmath685 , and so @xmath772 .
next we show hardness in the case when there is a crossing pair of equivalence relations . with a simple observation
, we can obtain a binary relation that is not a thick mapping and then apply lemma [ lem : non - thick - mapping ] .
let @xmath205 be equivalence relations on the same domain @xmath3 .
this means that there are @xmath202 such that @xmath775 and @xmath776 .
let @xmath777 and @xmath778 .
clearly , @xmath779 by lemma [ lem : restriction ] , @xmath780 is polynomial - time reducible to @xmath774 .
consider the relation @xmath781 , that is , @xmath782 .
we have that @xmath654 is reducible to @xmath780 and @xmath783 observe that @xmath44 is not a thick mapping and by lemma [ lem : non - thick - mapping ] , @xmath654 is np - complete . finally , we prove hardness in the case when there is a relation that is not 2-decomposable .
an example of such a relation is a ternary boolean affine relation @xmath784 for @xmath785 or @xmath786 .
the csp with global cardinality constraints for this relation is np - complete by @xcite .
our strategy is to obtain such a relation from a relation that is not 2-decomposable .
however , as in lemma [ lem : non - thick - mapping ] , we have to consider several cases .
we can assume that every binary relation in @xmath790 is a thick mapping , and no pair of equivalence relations from this set cross , otherwise the problem is np - complete by theorems [ the : decision - pp - defin ] , [ the : adding - constants ] , and lemmas [ lem : non - thick - mapping ] , [ lem : non - non - crossing ] .
furthermore , we can choose @xmath44 to be the ` smallest ' non-2-decomposable relation in the sense that every relation obtained from @xmath44 by restricting on a proper subset of @xmath3 is 2-decomposable , and every relation @xmath791 that either have smaller arity , or @xmath669 , is also 2-decomposable .
we claim that relation @xmath44 is ternary .
indeed , it can not be binary by assumptions made about it .
suppose that @xmath792 is a tuple such that @xmath793 for any @xmath112 .
let @xmath794}(x_4)\wedge\ldots \wedge c_{\ba[n]}(x_n)).\end{aligned}\ ] ] it is straightforward that @xmath795,\ba[2],\ba[3])\not\in\rel'$ ] , while , since any proper projection of @xmath44 is 2-decomposable , @xmath796 , @xmath797 , @xmath798 , implying @xmath799,\ba[3])\in\pr_{23}\rel'$ ] , @xmath795,\ba[3])\in\pr_{13}\rel'$ ] , @xmath795,\ba[2])\in\pr_{12}\rel'$ ] , respectively .
thus @xmath230 is not 2-decomposable , a contradiction with assumptions made .
if @xmath805 is a thick mapping with respect to some equivalence relations @xmath806 ( see p. [ page : with - respect ] ) , @xmath807 is a thick mapping with respect to @xmath808 , and @xmath809 is a thick mapping with respect to @xmath810 , then @xmath811 , @xmath812 , and @xmath813 .
let the corresponding classes of @xmath814 , @xmath815 , and @xmath816 be @xmath817 , and @xmath818 , respectively .
then @xmath819 , @xmath820 , @xmath821 .
indeed , if one of these equalities is not true , since by lemma [ lem : classes ] the relations @xmath822 are pp - definable in @xmath44 without equalities , the relation @xmath823 is pp - definable in @xmath44 and the constant relations , is smaller than @xmath44 , and is not 2-decomposable .
next we show that @xmath824 for all @xmath825 .
if there is @xmath826 with @xmath827 then setting @xmath828 we have @xmath829 and @xmath830 .
thus @xmath831 is smaller than @xmath44 and is not 2-decomposable .
the same is true for @xmath200 and @xmath72 , and for @xmath201 and @xmath832 .
since every binary projection of @xmath44 is a thick mapping this implies that @xmath833 , @xmath834 , and @xmath835 . for each @xmath836 and every @xmath837 ,
the relation @xmath838 , where @xmath839 , is definable in @xmath44 and therefore is a thick mapping with respect to , say , @xmath840 .
our next step is to show that @xmath44 can be chosen such that @xmath841 does not depend on the choice of @xmath842 and @xmath17 .
if one of these relations , say , @xmath843 , equals @xmath844 , while another one , say @xmath845 does not , then consider @xmath846 .
we have @xmath847 .
moreover , since by the choice of @xmath44 relation @xmath845 is a non - trivial thick mapping there is @xmath848 such that @xmath849 , hence @xmath850 . therefore @xmath846 is not a thick mapping , a contradiction .
since @xmath851 does not equal @xmath844 ( and @xmath852 , @xmath853 ) , every @xmath841 is non - trivial .
observe that due to the equalities @xmath854 , we also have that the unary projections of @xmath855 are equal to those of @xmath44 for any @xmath842 ; and therefore all the equivalence relations @xmath856 , for a fixed @xmath857 , are on the same domain , @xmath858 .
let @xmath859 since for any non - crossing @xmath205 we have @xmath860 , the relation @xmath861 is pp - definable in @xmath44 and constant relations without equalities . since all the @xmath856 are non - trivial , @xmath861 is also non - trivial .
we set @xmath862 let @xmath863 be defined for @xmath230 in the same way as @xmath855 for @xmath44 . observe that since @xmath864 if and only if there is @xmath865 such that @xmath866 , @xmath867 , @xmath868 , the relations @xmath869 , @xmath870 , @xmath871 for @xmath872 are thick mappings with respect to the equivalence relations @xmath873 , relations @xmath874 , and relations @xmath875 , respectively .
all the binary projections of @xmath230 equal to the direct product of the corresponding unary projections , while @xmath876 are non - trivial , which means @xmath230 is not the direct product of its unary projections , and therefore it is not 2-decomposable .
we then can replace @xmath44 with @xmath230 .
thus we have achieved that @xmath841 does not depend on the choice of @xmath842 and @xmath17 .
next we show that @xmath44 can be chosen such that @xmath877 , @xmath878 , and for each @xmath836 there is @xmath879 such that @xmath880 is a reflexive relation .
if , say , @xmath881 , or @xmath882 , or @xmath883 is not reflexive for any @xmath884 , consider the following relation @xmath885 first , observe that @xmath886 for any @xmath887 .
then , for any fixed @xmath888 the relation @xmath889 is the product @xmath890 , that is , a non - trivial thick mapping . thus @xmath230 is not 2-decomposable .
furthermore , @xmath891 , for any @xmath892 the relation @xmath871 is a thick mapping with respect to @xmath893 , and @xmath894 is reflexive . repeating this procedure for the two remaining pairs of coordinate positions ,
we obtain a relation @xmath895 with the required properties .
observe that repeating the procedure does not destroy the desired properties where they are already achieved .
replacing @xmath44 with @xmath895 if necessary , we may assume that @xmath44 also has all these properties .
set @xmath896 and @xmath897 .
let @xmath898 be such that @xmath843 is reflexive .
let also @xmath899 be such that @xmath900 .
then @xmath901 while @xmath902 .
choose @xmath533 such that @xmath903 .
then the restriction of @xmath44 onto 3-element set @xmath904 is not 2-decomposable .
thus @xmath44 can be assumed to be a relation on a 3-element set . if @xmath905 is not the equality relation ,
say , @xmath906 , then as the restriction of @xmath44 onto @xmath907 is still a not 2-decomposable relation , @xmath44 itself is a relation on the set @xmath907 . identifying @xmath842 and
@xmath908 with 0 and 1 it is not hard to see that it is the affine relation @xmath909 on @xmath907 . the csp with global cardinality constraints for this relation is np - complete by @xcite .
suppose that @xmath905 is the equality relation .
since each of @xmath910 is a mapping and @xmath911 , it follows that the three relations are disjoint . as @xmath843 is the identity mapping ,
@xmath845 and @xmath912 are two cyclic permutations of ( the 3-element set ) @xmath74 .
hence either @xmath913 or @xmath914 belongs to @xmath845 .
let it be @xmath913 .
restricting @xmath44 onto @xmath907 we obtain a relation @xmath230 whose projection @xmath915 equals @xmath916 , which is not a thick mapping . a contradiction with the choice of @xmath44 .
p. idziak , p. markovic , r. mckenzie , m. valeriote , and r. willard .
tractability and learnability arising from algebras with few subpowers . in _ proceedings of the 22th annual ieee simposium on logic in computer science_. ieee computer society , 2007 . | in a constraint satisfaction problem ( csp ) the goal is to find an assignment of a given set of variables subject to specified constraints .
a global cardinality constraint is an additional requirement that prescribes how many variables must be assigned a certain value .
we study the complexity of the problem @xmath0 , the constraint satisfaction problem with global cardinality constraints that allows only relations from the set @xmath1 .
the main result of this paper characterizes sets @xmath1 that give rise to problems solvable in polynomial time , and states that the remaining such problems are np - complete
. [ section ] [ theorem]corollary [ theorem]lemma [ theorem]remark [ theorem]observation [ theorem]definition [ theorem]example = = = = = = = = = ) = # 1#1 # 1 # 1#2#1 _ 1#1 _ # 2 = \ # 1height7pt depth7pt width.4pt # 1/ # 1 # 1#2 ( c # 1 + # 2 ) # 1#2 ( c # 1 + # 2 ) # 1#2#3 ( c # 1 + # 2 + # 3 ) # 1#2#3#4 c # 1 + # 2 + # 3 + # 4 = # 1 - 1.7mm^#1 #
1 - 0.2mm^#1 = _ _ = = = = = = = = = = = = = = = eufm10 scaled 1200 = eufm6 12=12=12= = msbm10 = msbm8 11=11=11= = msbm10 scaled 1200 13=13= 13= = `` 0b4e = ' ' 0d3f |
slow dynamics of the magnetization reversal is a crucial requirement for potential applications of single - chain magnets ( scm s ) @xcite , and nanowires in general , in magnetic - memory manufacture . for nanowires with a biaxial anisotropy , provided that their length is much greater than the cross section diameter but smaller than exchange length , this phenomenon is governed by thermal nucleation and propagation of soliton - antisoliton pairs ; the associated characteristic time is expected to follow an arrhenius law @xcite . for genuine one dimensional ( 1d ) ising systems with single spin - flip stochastic dynamics , a slow relaxation of the magnetization was first predicted by glauber @xcite in 1963 . through glauber s approach ,
many physical systems were investigated , ranging from dielectrics @xcite to polymers @xcite .
more fundamentally , this model has been employed to justify the use of the kohlrauch - williams - watts function @xcite ( stretched exponential ) to fit the relaxation of generalized 1d spin systems .
also the universality issue of the dynamic critical exponent @xcite of the ising model @xcite , as well as strongly out - of - equilibrium processes ( magnetization reversal @xcite , facilitated dynamics @xcite , etc . ) have been studied moving from the basic ideas proposed by glauber . in this paper ,
single spin - flip glauber dynamics is used to investigate theoretically the slow dynamics of the magnetization reversal in molecular magnetic systems . in particular , we extend glauber s theory @xcite to the ising collinear ferrimagnetic chain , as well as to the case of a chain in which reciprocal non - collinearity of local anisotropy axes and the crystallographic ( laboratory ) frame is encountered .
such extensions are motivated by the fact that ( i ) in molecular - based realizations of scm s , antiferromagnetic coupling typically has a larger intensity than the ferromagnetic one ; in fact , the overlapping of magnetic orbitals , which implies antiferromagnetic exchange interaction between neighboring spins , can be more easily obtained than the orthogonality condition , leading to ferromagnetism @xcite ; ( ii ) non - collinearity between local anisotropy axes and the crystallographic ( laboratory ) frame takes place quite often in molecular spin chains . besides magnetization reversal , the dynamic response of the system to a weak magnetic field , oscillating in time at frequency @xmath0 , is also studied .
depending on the specific experimental geometry , selection rules are found for the occurrence of resonant behavior of the _ a.c . _
susceptibility as a function of temperature ( stochastic resonance ) at low frequencies , as well as for slow relaxation of the magnetization in zero field at low temperatures .
the paper is organized as follows . in sect .
ii we extend glauber s theory @xcite , originally formulated for a chain of identical and collinear spins , to the more general model of a chain with non - collinear spins , possibly with land factors that vary from site to site . in sect .
iii we use two different theoretical methods ( the generating functions approach and the fourier transform approach ) to investigate the relaxation of the magnetization after removal of an external static magnetic field , starting from two different initial conditions : fully saturated or partially saturated . in sect .
iv we calculate , in a linear approximation , the magnetic response of the system to an oscillating magnetic field . for a chain of @xmath1 spins ,
the _ a.c .
_ susceptibility is expressed as the superposition of @xmath1 contributions , each characterized by its time scale ; through a few simple examples , we show that , depending on the geometry of the system ( _ i.e. _ , the relative orientations of the local easy anisotropy axes and of the applied field ) , different time scales can be selected , possibly giving rise , for low frequencies , to a resonant peak in the temperature - dependence of the complex magnetic susceptibility . in sect . v we show that the theory provides a satisfactory account for the scm behavior experimentally observed in some magnetic molecular chain compounds , characterized by dominant antiferromagnetic exchange interactions and non - collinearity between spins . finally , in sect .
vi , the conclusions are drawn and possible forthcoming applications are also discussed .
in a celebrated paper @xcite , glauber introduced , in the usual 1d ising model @xcite , a stochastic dependence on the time variable @xmath2 : _ i.e. _ , the state of a spin lying on the @xmath3-th lattice site was represented by a two - valued stochastic function @xmath4 @xmath5 @xmath6 is the exchange coupling constant , that favors nearest neighboring spins to lie parallel ( @xmath7 , ferromagnetic exchange ) or antiparallel ( @xmath8 , antiferromagnetic exchange ) ; @xmath9 is the land factor of each spin , and @xmath10 the bohr magneton . in the original paper @xcite a 1d lattice of equivalent and collinear spins
was studied ; there the response to a time - dependent magnetic field h@xmath11 , applied parallel to the axis of spin quantization and oscillating with frequency @xmath0 , as in typical _ a.c . _
susceptibility experiments , was also considered . in order to investigate the phenomena of slow relaxation ( for h=0 ) and resonant behavior of the _ a.c .
_ susceptibility ( for h@xmath12 ) in molecular scm s , we are going to generalize the glauber model ( [ ising_hamiltonian_in_glauber ] ) accounting for non - collinearity of local anisotropy axes and crystallographic ( laboratory ) frame .
to this aim , we adopt the following model hamiltonian @xmath13 @xmath6 is an effective ising exchange coupling that can approximately be related to the hamiltonian parameters of a real scm @xcite : see later on the discussion in section v. like in the usual ising - glauber collinear model ( [ ising_hamiltonian_in_glauber ] ) , the spins in eq .
( [ non_collinear_h ] ) are described by classical , one - component vectors that are allowed to take two integer values @xmath14 , but now the magnetic moments may be oriented along different directions , @xmath15 , varying from site to site . within this scheme ,
the land tensor of a spin on the @xmath3-th lattice site has just a non - zero component , @xmath16 , along the local easy anisotropy direction @xmath17 . denoting by @xmath18 the direction of the oscilla.ting magnetic field , @xmath19=@xmath20 , we define the generalized land factor @xmath21 appearing in eq .
( [ non_collinear_h ] ) as @xmath22 _
i.e. _ , a scalar quantity that varies from site to site .
following glauber @xcite , we define the single spin expectation value @xmath23 , where brackets denote a proper ensemble average , and the stochastic magnetization along the direction of the applied field @xmath24 the basic equation of motion of the glauber model @xcite reads @xmath25 in which @xmath26 represents the probability per unit time to reverse the @xmath3-th spin , through the flip + @xmath27 . for a system of @xmath1 coupled spins
, this probability is affected by the interaction with the other spins , with the thermal bath and , possibly , with an external magnetic field . among all possible assumptions for the transition probability @xmath28 as a function of the @xmath29 variables @xcite @xmath30 , again following glauber @xcite we require @xmath28 to be independent of time and to depend only on the configuration of the two nearest neighbors of the @xmath3-th spin . in zero field , these requirements are fulfilled by @xmath31\ ] ] while in the presence of an external field @xmath32 is usually chosen ; the attempt frequency @xmath33 ( _ i.e. _ , the probability per unit time to reverse an isolated spin ) remains an undetermined parameter of the model ; @xmath34 accounts for the effect of the nearest neighbors ; the parameters @xmath35 have the role of stabilizing the configuration in which the @xmath3-th spin is parallel to the field , and destabilizing the antiparallel configuration .
thanks to the particular choices and for the transition probability , by imposing the detailed balance conditions @xcite it is possible to express @xmath34 and @xmath36 as functions of the parameters in the spin hamiltonian ( [ non_collinear_h ] ) @xmath37 where @xmath38 is the inverse temperature in units of boltzmann s constant .
another advantage of glauber s choices and is that the equation of motion takes a simple form . in particular , for @xmath39 , eq .
( [ glauber_eq_motion ] ) with the choice ( [ transition_prob_glauber ] ) becomes @xmath40 where @xmath41 denotes the vector of single - spin expectation values @xmath42 and @xmath43 is a square @xmath44 symmetric matrix , whose non - zero elements are a@xmath45=1 and a@xmath46=a@xmath47=@xmath48 , with a@xmath49=a@xmath50=@xmath48 if periodic boundary conditions are assumed for the @xmath1-spin chain .
a closed solution of this set of first - order differential equations can be obtained expressing the expectation value of each spin , @xmath51 , in terms of its spatial fourier transform ( ft ) @xmath52 @xmath53 substituting eq .
( [ xxx ] ) into eq .
( [ glauber_eq_motion_mat_ring ] ) , one readily obtains the dispersion relation @xmath54 with @xmath55 @xcite . for ferromagnetic coupling ( @xmath7 , hence @xmath56 )
the smallest eigenvalue @xmath57 occurs for @xmath58=0 , independently of the number of spins @xmath1 in the chain . for antiferromagnetic coupling ( @xmath8 , hence @xmath59 ) and @xmath1
even , the smallest eigenvalue @xmath60 occurs for @xmath58=@xmath61 ; while in the case of @xmath1 odd , the smallest eigenvalue corresponds to @xmath62 $ ] , thus depending on the number of spins in the antiferromagnetic chain @xcite .
the characteristic time scales of the system , @xmath63 , are given by @xmath64 at finite temperatures the characteristic times @xmath63 are finite because @xmath65 ; for @xmath66 one has that @xmath67 vanishes irrespectively of the sign of @xmath6 , because @xmath68 .
thus , for h@xmath69 , there is one diverging time scale in the @xmath66 limit : @xmath70 for ferromagnetic coupling and @xmath71 for antiferromagnetic coupling ( and even @xmath1 ) . in the presence of a non - zero , oscillating field @xmath72 , the equation of motion ( [ glauber_eq_motion ] ) with the choice ( [ transition_prob_field_glauber ] ) takes a form ( see eq . [ glauber_eq_motion_field_ring ] in section iv later on ) which can still be solved , though in an approximate way @xcite , for a sufficiently weak intensity of the applied magnetic field .
the original glauber model was formulated for a chain of collinear spins with the same land factors : _
i.e. _ , in eq .
( [ non_collinear_h ] ) one has @xmath73 , @xmath74 . assuming that the system has been fully magnetized by means of a strong external field , one can study how the system evolves if the field is removed abruptly .
this corresponds to take a fully saturated initial condition @xmath75 in ferrimagnetic chains , on the other hand , a partial " saturation can be reached , provided the antiferromagnetic coupling ( @xmath8 ) between nearest neighbors is
strong enough " , in a sense that will be clarified later on .
in fact , if the land factors for odd and even sites are not equal ( @xmath76 ) , through the application of an opportune field the sample can be prepared in a configuration with @xmath77 with respect to the case considered by glauber , it is convenient to separate explicitly the expectation values of the odd sites , @xmath78 , from those of the even sites , @xmath79 .
thus , for @xmath39 , the set of @xmath1 equations of motion ( [ glauber_eq_motion_mat_ring ] ) can be rewritten as @xmath80\\ \frac{d}{dt}s_{2r+1}=-\alpha \left[s_{2r+1}+\frac{1}{2}\gamma ( s_{2r}+s_{2r-2})\right ] \end{cases}\ ] ] in the following , the solutions of ( [ newform ] ) will be found using two different approaches that yield identical results .
the generating function approach , which closely follows the original glauber s paper , is exposed in detail in appendix . here , in order to distinguish between the ferromagnetic and ferrimagnetic relaxations , we specialize the general solution , eq . and eq . , to the two different kinds of initial conditions , eq .
( [ sat ] ) and eq .
( [ partial_sat ] ) . in both cases
, we will assume that the exchange coupling @xmath6 is negative .
the `` partially saturated '' configuration , eq .
( [ partial_sat ] ) , reflects a typical experimental situation , in which the antiferromagnetic coupling is much bigger ( @xmath81 100@xmath821000 k ) than the zeeman energy associated with accessible magnetic fields . on the other hand , the initial configuration with all the spins aligned in the same direction , eq .
( [ sat ] ) , clearly reflects the experimental situation of a fully saturated sample .
this condition is easily obtained for ferromagnetic coupling ( @xmath7 ) , while it may require very strong fields ( eventually unaccessible ) for antiferromagnetic coupling ( @xmath8 ) .
let us start from the saturated configuration , eq .
( [ sat ] ) . substituting the initial condition @xmath83 for all @xmath3 in both and
, we obtain @xmath84 \\
s_{2r+1}(t)=e^{-\alpha t}\overset{+\infty } { \underset{m=-\infty}\sum } \left [ \mathcal{i}_{2(r - m)}(\gamma \alpha t)+\mathcal{i}_{2(r - m)+1}(\gamma \alpha t)\right ] .
\end{cases}\ ] ] hence , exploiting the property of the bessel functions ( taking @xmath85 ) , and redefining the sums by a unique index @xmath86 , we get @xmath87 this means that , starting with all the spins aligned in the same direction , each spin expectation value ( both on even and odd sites ) decays obeying a mono - exponential law with relaxation time @xmath88^{-1}$ ] , which is just the characteristic time scale obtained as the inverse of the dispersion relation @xmath89 with zero wave number @xmath90 , see eq . .
notice that @xmath70 can diverge for @xmath66 only in the case of ferromagnetic coupling , @xmath7 ( @xmath56 ) .
let us now consider the partially saturated configuration , eq .
( [ partial_sat ] ) , in which @xmath83 for @xmath3 odd and @xmath91 for @xmath3 even . substituting these initial conditions in both and
, we obtain @xmath92 \\ s_{2r+1}(t)= e^{-\alpha t}\overset{+\infty } { \underset{m=-\infty}\sum } \left [ \mathcal{i}_{2(r - m)}(\gamma \alpha t)-\mathcal{i}_{2(r - m)+1}(\gamma \alpha t)\right ] \end{cases}\ ] ] and , still exploiting the property ( but now for @xmath93 ) , we get @xmath94 also in this case all the spins of the system relax with a mono - exponential law , but now the relaxation time is @xmath95^{-1}$ ] , which corresponds to the inverse of the eigenvalue @xmath96 with wave number @xmath97 , see eq .
( [ glauber_dispersion_pbc ] ) .
notice that @xmath71 can diverge for @xmath66 only in the case of antiferromagnetic coupling , @xmath8 ( @xmath59 ) .
summarizing , according to the sign of the exchange constant , both time scales @xmath70 ( for @xmath7 ) and @xmath71 ( for @xmath8 ) diverge in the low temperature limit @xmath66 , following an arrhenius law @xmath98 with energy barrier @xmath99 ( slow relaxing mode ) .
it is worth noting that the remaining relaxation times , given by the inverse of the eigenvalues in eq . with @xmath100 and @xmath101 ,
always remain of the same order of magnitude as @xmath102 ( fast relaxing modes ) .
this time scale is typically very small ( @xmath103 ps ) in real systems , and negligible with respect to the characteristic times involved in any experimental measurement we refer to . the solutions , and , to the set of equations ( [ newform ] )
can alternatively be deduced within the fourier transform ( ft ) formalism , which has already been exploited to obtain the dispersion relation . recalling the definition of @xmath51 and its spatial ft @xmath104 we evaluate @xmath105 at time @xmath106 , @xmath107 , for the two initial conditions of interest , ( [ sat ] ) and ( [ partial_sat ] ) .
starting from the all - spin - up configuration , eq .
( [ sat ] ) , we have @xmath108 hence the solution for the expectation value of a spin localized on the @xmath3 lattice site at time @xmath2 is @xmath109 which is identical to since @xmath110 . starting from the partially saturated configuration , eq .
( [ partial_sat ] ) , it is useful to rewrite it as @xmath111 , so that the ft at @xmath106 is @xmath112 hence the solution is readily obtained @xmath113 which is identical to since @xmath114 . finally , we observe that eqs . and hold even for a ring with a _
number @xmath1 of spins , while eqs . and
were obtained in the infinite - chain limit .
the expectation values , @xmath51 , of spins localized on the even and odd sites of a linear lattice at time @xmath2 , computed either with the generating functions or the fourier transform approach , have been shown to display a mono - exponential relaxation , see eqs . and , with different time scales , @xmath115^{-1}$ ] and @xmath116^{-1}$ ] respectively , depending on the different initial conditions , eqs .
( [ sat ] ) and ( [ partial_sat ] ) . as a consequence , also the macroscopic magnetization , expressed by eq . , displays the same mono - exponential relaxation as the single site quantities @xmath51 .
a chain in which all the magnetic moments are equal can be prepared only in the saturated initial configuration , eq .
( [ sat ] ) , with all the spin aligned in the same direction , through the application of an external field .
thus , when the field is abruptly removed , such a system will relax slowly at low temperature only if the exchange coupling is ferromagnetic ( @xmath7 ) .
in contrast , if the exchange coupling is antiferromagnetic ( @xmath8 ) and the chain is forced " in the saturated state by a strong applied magnetic field , the system will relax very fast ( in a typical time of the order of @xmath102 ) when the field is removed .
let us now discuss how these results , first obtained by glauber @xcite , are generalized to the case of a chain in which the magnetic moments are collinear , but _ not _ equal on each site . as pointed out in the introduction , a model with antiferromagnetic coupling ( @xmath8 ) but non - compensated magnetic moments on the two sublattices
is more akin to real scm s @xcite .
yet it is very interesting since , depending on the intensity of the applied magnetic field , the system can be prepared either in the saturated initial configuration , eq .
( [ sat ] ) , where all spins are parallel to each other , or in the partially saturated one eq .
( [ partial_sat ] ) , where nearest neighbors are antiparallel . in the former case , a very strong field
is required in order to overcome the antiferromagnetic coupling between nearest neighbors ; once the field is removed , the relaxation of the magnetization is expected to be fast at low temperatures , on the basis of the solution . in the latter case
, the partially saturated initial configuration ( [ partial_sat ] ) can easily be obtained through the application of a smaller , experimentally accessible magnetic field ; when the field is abruptly removed , the relaxation is expected to be slow according to the solution .
the solution justifies the observation of scm behavior in ferrimagnetic quasi-1d compounds like cophome @xcite ( see sect .
v ) . summarizing , we have found that when a collinear ferrimagnetic chain is prepared in an initial state fully or partially saturated depending on the intensity of the applied magnetic field once the field is removed abruptly , the spin system can show fast or slow relaxation , respectively .
fast relaxation corresponds to stronger fields ; unfortunately for the quasi-1d chain compound cophome @xcite , the antiferromagnetic exchange constant is so large ( @xmath117k ) that the realization of a fully saturated initial configuration would require a very high , almost unaccessible field ( @xmath118 koe ) .
thus this compound is not a good candidate for such a kind of experiments @xcite .
[ collinear ] in the presence of a magnetic field h , the transition probability to be put in the equation of motion is @xmath119 , defined in eq . .
one obtains @xmath120 \cr & + & \frac{\gamma\delta_k}{2}\left [ \left\langle \sigma_{k}\sigma_{k+1}\right\rangle_t + \left\langle \sigma_{k-1}\sigma_{k}\right\rangle_t \right ] -\delta_k \big\}\end{aligned}\ ] ] that differs from eq .
( [ newform ] ) , considered earlier for h@xmath69 , in the presence of both a non - homogeneous term , @xmath35 , and the time - dependent pair - correlation functions @xmath121 .
the latter ones , assuming that the field is so weak to induce just small departures from equilibrium , can be approximated by their time - independent counterparts @xcite @xmath122 as it is usual in _ a.c .
_ susceptibility measurements , we also assume the time - dependent magnetic field @xmath19=@xmath123 , oscillating at frequency @xmath0 , to be weak so that the @xmath35 parameters can be linearized @xmath124 the system of equations of motion then takes the form @xmath125\cr & + & \beta f(\beta j_i ) \mu_{b } g_k \mathrm{h}(t)\,,\end{aligned}\ ] ] where @xmath126 is a function of the reduced coupling constant @xmath127 and we have taken into account that @xmath128 . after a brief transient period , the system will reach the stationary condition in which the magnetic moment of each spin oscillates coherently with the forcing term at the frequency @xmath0 . expressing the expectation value of a spin on the @xmath3-th lattice site , @xmath51 , through its spatial ft , @xmath52
, the trial solution is @xmath129 substituting the latter in the system we get @xmath130 where @xmath131 is the ft of @xmath21 : @xmath132 the average of stochastic magnetization can readily be obtained from ( [ stochastic_mag_glauber ] ) as @xmath133 which accounts for non - collinearity of local anisotropy axes with respect to the field direction . performing the sum over all the lattice sites ( @xmath3 indices ) yields a factor @xmath134 in eq . ;
substituting the expression ( [ new ] ) for @xmath52 , one obtains @xmath135 then , considering that @xmath136 , the _ a.c .
_ susceptibility is finally obtained dividing by @xmath137 @xmath138 in principle , the _
_ susceptibility of a chain with @xmath1 spins admits @xmath1 poles , corresponding to the @xmath1 eigenvalues @xmath96 in eq . .
each mode is related to a different time scale @xmath139 . in practice ,
not all the time scales will be involved in the complex susceptibility @xmath140 , but _ only the ones selected by _ @xmath131 . a result
similar , at first glance , to eq .
( [ ac_susceptibility_geom_glauber ] ) was deduced by suzuki and kubo @xcite , but in their case the relationship was between the time scale @xmath63 and the wave - vector - dependent susceptibility @xmath141 .
in contrast , in an _ a.c . _ susceptibility experiment only the zero - wave - vector susceptibility @xmath142 is accessible ; the peculiarity of eq .
( [ ac_susceptibility_geom_glauber ] ) is that other time scales , different from @xmath70 , can be selected thanks to the dependence of the gyromagnetic factors @xmath21 and of the local anisotropy axes on the site position @xmath3 .
this is the main result of our study and will be clarified hereafter through a few examples .
let us start considering the case of a one - dimensional ising model with two kinds of spins ( aligned parallel on antiparallel to the chain axis ) alternating on the odd and even magnetic sites of the lattice with land factors @xmath143 and @xmath144 ( integer @xmath145 ) , respectively .
strictly speaking , a collinear ising ferrimagnet is characterized by an antiferromagnetic coupling @xmath8 , but also the case @xmath7 can be treated through eq .
( [ ac_susceptibility_geom_glauber ] ) .
in fact , since the local axis of anisotropy has the same direction for all the spins , the ft of the site - dependent land factor is @xmath146 \cr&=&(g_{e}+e^{iq}g_{o } ) { 1\over { n}}\sum_{r=1}^{n/2 } e^{-iq 2r}.\end{aligned}\ ] ] taking into account that , in the presence of periodic boundary conditions , one has @xmath147 it follows that the only non - zero values of @xmath131 are for @xmath90 and @xmath97 @xmath148.\ ] ] thus , according to eq .
, the parallel _ a.c .
_ susceptibility ( @xmath149 ) of a collinear ising chain with alternating spins is @xmath150 it appears that both the relaxation times obtained by suzuki and kubo @xcite for the ordinary and the staggered susceptibility of the usual ising model , namely @xmath151^{-1}$ ] and @xmath152^{-1}$ ] respectively , do coexist in the _ a.c .
_ susceptibility ( [ ac_susceptibility_ferrimagnet_glauber ] ) .
notice that , in the @xmath153 limit , the static susceptibility of the ising ferrimagnet in zero field @xcite is recovered from eq .
( [ ac_susceptibility_ferrimagnet_glauber ] ) , since one has @xmath154 .
as regards the dynamic response of the system to an oscillating magnetic field applied along the chain axis , depending on the sign of the effective exchange coupling constant @xmath6 , the ferromagnetic ( @xmath155 ) or the antiferromagnetic ( @xmath156 ) branch of the parallel susceptibility are characterized by a diverging time scale at low temperature . in particular , for a collinear ising ferrimagnet one has @xmath8 , so that @xmath71 is diverging , while @xmath70 is short ( of the order of @xmath102 , the attempt frequency of an isolated spin ) .
thus , for @xmath8 , a resonant behavior of the _ a.c . _
susceptibility versus temperature ( at low frequencies @xmath157 ) can only be observed in the case @xmath158 ( see fig .
1d ) when magnetic moments are _ uncompensated _ , while a broad peak is found in the case @xmath159 when the net magnetization is zero ( see fig .
clearly , for @xmath7 , a resonant peak is found in both cases ( see fig . 1a and 1c ) , because a net magnetization is always present in the system .
such a resonant behavior of the _ a.c . _
susceptibility versus @xmath160 , in ferromagnetic @xcite as well as in ferrimagnetic @xcite ising chains with single spin - flip glauber dynamics , is a manifestation of the stochastic resonance phenomenon @xcite : _ i.e. _ , the response of a set of coupled bistable systems to a periodic drive is enhanced in the presence of a stochastic noise when a matching occurs between the fluctuation - induced switching rate of the system and the forcing frequency . in a magnetic chain ,
the role of stochastic noise is played by thermal fluctuations and the resonant peak in the temperature - dependence of the _ a.c .
_ susceptibility occurs when the statistical time scale , associated to the slow decay of the magnetization , matches with the deterministic time scale of the applied magnetic field @xmath161 next , as an example of a non - collinear spin arrangement , we consider a system of spins with the local axes of anisotropy arranged on an @xmath58-fold helix ( see fig .
2 ) ; @xmath162 is the angle that the local axes form with @xmath163 , the unique axis of the helix ( _ i.e. _ , the chain axis ) . in this case
the land factors are equal on all lattice sites , but different spins experience different fields because of the geometrical arrangement of magnetic moments . in the following
we will make the approximation that the land tensor of a spin on the @xmath3-th lattice site has just a non - zero component @xmath9 along the easy anisotropy direction @xmath15 , so that @xmath164 ( see eq . ) . in the crystallographic frame ( @xmath165 ) ,
the directors @xmath15 read ( integer @xmath3 ) @xmath166+\cos\theta \hat{\mathbf{e}}_{z}.\ ] ] [ multihelix ] [ helix ] let us first consider the case of an oscillating magnetic field h applied parallel to @xmath163 , the helix axis .
all the spins actually undergo the same field , and since @xmath167 independently of the lattice site @xmath3 , the only peak in the ft @xmath131 occurs at @xmath90 @xmath168 following the same procedure as in the previous paragraph , the parallel _ a.c . _ susceptibility ( @xmath149 ) takes an expression ( valid for any value of the fold index @xmath58 of the helix ) @xmath169 that differs from glauber s result for the collinear ising chain @xcite only by the geometrical factor @xmath170 . for ferromagnetic coupling , @xmath7 , the relaxation time @xmath171^{-1}$ ]
diverges as @xmath172 , and a resonant behavior of the _ a.c . _
parallel susceptibility versus temperature is found , at low frequency , when the oscillating field is applied parallel to the helix axis , @xmath163 , along which spins are _ uncompensated _ : see fig .
3a , which refers to the case of a two - fold helix ( @xmath173 ) .
let us now consider the case of an oscillating magnetic field h applied perpendicularly to the chain axis . in this configuration , it is useful to distinguish the case @xmath173 from the general case @xmath174 .
* @xmath173 + in this case , it is worth noticing that for h parallel to @xmath175 , one has identically @xmath176 for any lattice site @xmath145 .
thus , @xmath177 and the corresponding _ a.c . _ susceptibility is identically zero @xmath178 ( not shown ) . in contrast , for h parallel to @xmath179 , one has @xmath180 on odd sites and @xmath181 on even sites .
the ft is @xmath182 where we have taken into account eq .
( [ boundary ] ) .
thus , for ferromagnetic coupling , @xmath7 , the relaxation time @xmath183^{-1}$ ] does not diverge as @xmath184 , and the perpendicular _ a.c .
_ susceptibility @xmath185 does not present a resonant behavior as a function of temperature ; rather , it presents a broad maximum ( see fig .
clearly , in the case of antiferromagnetic coupling , @xmath8 , the behavior of the susceptibility components is reversed : a broad maximum is found for the temperature dependence of the parallel susceptibility @xmath186 ( see fig .
3d ) , while a resonant behavior is found for the perpendicular susceptibility @xmath187 ( see fig .
3c ) . * @xmath188 + in this case , denoting by @xmath189 and @xmath190 the directors of the in - plane field , the ft s of @xmath21 are given by @xmath191 remarkably , as just @xmath192 appears in eq .
, the general result for the in - plane susceptibility turns out to be independent of the field direction .
thus , for @xmath174 , the perpendicular ( @xmath193 ) _ a.c .
_ susceptibility of the @xmath58-fold helix is given by @xmath194 -i ( \frac{\omega}{\alpha})},\end{aligned}\ ] ] where we have exploited the fact that @xmath195 for the term appearing at the denominator of eq . .
summarizing , in the general case of an ising chain with an @xmath58-fold helical spin arrangement ( @xmath196 ) , we have explicitly shown that a resonant behavior of the _ a.c . _
susceptibility versus temperature , similar to the one displayed by ferromagnetic @xcite and ferrimagnetic @xcite ising chains with collinear spins , is possible only for field applied in a direction where magnetic moments are _ uncompensated_. in contrast , a broad peak is found when there is no net magnetization along the field direction .
in this section we will apply the developed formalism to some real compounds as representative realizations of scm s ; for the three selected systems we know
this restriction is far from being exhaustive @xcite _ a.c .
_ susceptibility data on single crystal are available , which is a fundamental requirement for checking the proposed selection rules .
the considered systems @xcite are characterized by the alternation of two types of magnetic centers along the chain axis , so that at least two spins per cell have to be considered ; moreover , the magnetic moments are not collinear , the dominant exchange interactions are antiferromagnetic and a strong single - ion anisotropy is present , which favors magnetization alignment along certain crystallographic directions @xmath15 .
the static properties of these compounds , like magnetization and static susceptibility , are generally well described using a classical heisenberg model with an isotropic exchange coupling @xmath197 and a single - ion anisotropy @xmath198 .
thus , in order to describe the dynamic behavior in response to a weak , oscillating magnetic field by means of the previously developed theory , it is necessary to relate the hamiltonian parameters of such a classical spin model to the exchange constant @xmath6 of the effective ising model ( [ non_collinear_h ] ) . in the following
we will show , through a few examples on real systems , that indeed , depending on the geometry , selection rules are obeyed for the occurrence of slow relaxation of the magnetization at low temperatures ( @xmath199 ) , as well as for resonant behavior of the _ a.c .
_ susceptibility as a function of temperature at low frequencies . as regards the frequencies involved in an _
_ susceptibility experiment on real scm s , generally @xcite they lie in the range @xmath200 hz , while the attempt frequency @xmath201 is of the order of @xmath202 hz .
thus , for a typical experiment , a resonant peak in the _ a.c .
_ susceptibility can safely be observed provided that at least one of the characteristic time scales @xmath63 involved in diverges at low @xmath160 , in order for the condition ( [ cond_reson ] ) to be satisfied . in the one - dimensional molecular magnetic compound of formula [ mn(tpp)o@xmath204pphh]@xmath205h@xmath204o , obtained by reacting mn(iii ) acetate @xmath206tetraphenylporphyrin with phenylphosphinic acid @xcite , hereafter denoted by mn@xmath203-based scm
, the phenylphosphinate anion transmits a sizeable antiferromagnetic exchange interaction that , combined with the easy axis magnetic anisotropy of the mn@xmath203 sites , gives rise to a canted antiferromagnetic arrangement of the spins .
the static single - crystal magnetic properties were analyzed in the framework of a classical spins hamiltonian @xmath207 where @xmath208 is the antiferromagnetic nearest neighbor exchange interaction between @xmath209 spins .
@xmath210 is the uniaxial anisotropy favoring two different local axes , alternating along odd and even sites respectively ; both axes form an angle @xmath211 with the crystallographic @xmath212 axis , while they form opposite angles of modulus @xmath213 with the @xmath214 axis ( see fig .
4 ) . thus we can write @xmath215 and @xmath216
. a best fit of the static single - crystal magnetic susceptibilities , calculated via a monte carlo simulation @xcite provides @xmath217 k and @xmath218 k ; the gyromagnetic tensor @xmath219 is diagonal and isotropic with @xmath220 .
equivalent results can be obtained calculating the static properties of model ( [ heisenberg ] ) via a transfer matrix approach @xcite .
since the uniaxial anisotropy @xmath198 is rather strong with respect to the exchange coupling @xmath221 , as a first approximation one can assume the two sublattice magnetizations to be directed just along the two easy axes , @xmath222 and @xmath223 , so that the chain system ( [ heisenberg ] ) can be described by a non - collinear ising model formally identical to eq .
( [ non_collinear_h ] ) , with an effective @xcite ising exchange coupling @xmath6 and a generalized land factor @xmath21 defined as , respectively @xmath224 [ mn ] depending on the orientation of the oscillating magnetic field with respect to the crystallographic axes , the ft of the generalized land factor takes the following forms @xmath225 \cr & = & g^{\vert } \sqrt{s(s+1 ) } \begin{cases } \sin \theta_1 \cos
\phi_1 \delta_{q,0},~ { \rm h}\vert x \\ \sin \theta_1
\sin \phi_1 \delta_{q,\pi},~ { \rm h}\vert y \\ \cos \theta_1 \delta_{q,0 } , ~ { \rm h}\vert z~ { \rm(chain~ axis ) } \end{cases}\end{aligned}\ ] ] the corresponding _ a.c .
_ susceptibility takes the expression @xmath226\cr & \times & \begin{cases } \sin^2 \theta \cos^2 \phi { 1\over { ( 1-\gamma_i)-i({{\omega}\over { \alpha}})}},~ { \rm h}\vert x \\
\sin^2 \theta \sin^2 \phi { 1\over { ( 1+\gamma_i)-i({{\omega}\over { \alpha}})}},~ { \rm h}\vert y \\ \cos^2 \theta { 1\over { ( 1-\gamma_i)-i({{\omega}\over { \alpha } } ) } } , ~ { \rm h}\vert z~ { \rm(chain~ axis ) }
\end{cases}\end{aligned}\ ] ] taking into account that , for the mn@xmath203 scm under study , the true " exchange coupling , @xmath197 in eq .
( [ heisenberg ] ) , is antiferromagnetic , and that the angle between the two easy anisotropy axes @xmath227 and @xmath228 is @xmath229 ( see fig .
4 , right ) , from eq .
( [ bipartite ] ) it follows that also the effective ising exchange coupling is antiferromagnetic , @xmath8 . as a consequence , in the low temperature limit @xmath230 ,
the relaxation time @xmath71 diverges , while @xmath70 does not .
for low frequencies @xmath157 , the _ a.c .
_ susceptibility presents a resonant behavior only when the oscillating magnetic field is applied along the crystallographic @xmath175 axis , _ i.e. _ the direction , perpendicular to the chain axis , along which the magnetizations of the two sublattices are _ uncompensated _ ( see fig .
4 ) . in contrast
, when h is applied parallel to @xmath163 ( the chain axis ) or to @xmath179 , namely two directions along which the magnetizations of the two sublattices are exactly compensated , no resonant behavior is expected .
these theoretical predictions turn out to be in excellent agreement with experimental _ a.c .
_ susceptibility data @xcite obtained in a single crystal sample of [ mn(tpp)o@xmath204pphh]@xmath205h@xmath204o , thus confirming that such a mn@xmath203-based canted antiferromagnet is a _ bona fide _ scm .
the molecular magnetic compound of formula [ dy(hfac)@xmath231(nitphoph ) ] , hereafter denoted by dy@xmath232-based scm , belongs to a family of quasi one - dimensional magnets in which rare earth ions ( with spin @xmath233 ) and organic radical ions ( with spin @xmath234 ) alternate themselves along the chain axis , @xmath163 , which in this compound coincides with the crystallographic @xmath235 axis .
static measurements in single crystal samples suggest @xcite that there is an antiferromagnetic exchange interaction between neighboring dy@xmath203 ions , whose easy anisotropy axes are canted with respect to the chain axis in such a way to generate an uncompensated moment along @xmath235 , while the components in the @xmath236 plane are compensated .
thus , as far as the dominant exchange interaction @xmath208 between dy@xmath203 ions is taken into account , the spin hamiltonian of the system is quite similar to eq.([heisenberg ] ) .
however , with respect to the mn@xmath232-based chain ,
the crystal structure of the dy@xmath232-based scm is more complicated , not only owing to the presence of two kinds of magnetic centers ( the dy@xmath203 ions and the organic radical ions ) , but mainly because the system is formed by two different families of chains , with two almost orthogonal projections of the easy axes in the @xmath236 plane , perpendicular to the chain axis : this accidental " ( in the sense that it is not imposed by symmetry ) orthogonality is the reason for the nearly isotropic magnetic behavior displayed by the system within such a plane @xcite .
[ dy ] we adopt a simplified model formally equivalent to eq .
( [ heisenberg ] ) . taking into account only the dominant antiferromagnetic exchange interaction ( @xmath208 ) between neighboring
dy@xmath203 ions ( which indeed are next nearest neighbors in the real system ) and their uniaxial anisotropy ( @xmath210 ) , the system can approximately be described by the classical spins hamiltonian ( [ heisenberg ] ) , where now @xmath237 . by means of a classical transfer matrix calculation ,
the static properties of the dy@xmath203-based scm turn out to be satisfactorily fitted @xcite by @xmath238 k , @xmath239 k , @xmath240 , with the two easy anisotropy axes @xmath222 , @xmath223 forming equal angles @xmath241 with the chain axis @xmath163 .
( notice that the latter property holds true for both families of chains . ) also in the case of the dy@xmath203-based scm , the uniaxial anisotropy @xmath198 turns out to be sufficiently strong with respect to the exchange coupling @xmath242 in order to assume , as a first approximation @xcite , the two sublattice magnetizations of dy@xmath203 to be directed just along the two easy axes . thus one can define an equivalent non - collinear ising model ( [ non_collinear_h ] ) , where the effective ising exchange coupling @xmath6 and the generalized land factor @xmath243 are now defined as @xmath244 depending on the orientation of the oscillating magnetic field with respect to the crystallographic axes , the ft of the generalized land factor takes the form @xmath245
\cr & \propto & g^{\vert } \begin{cases } \cos \theta ~\delta_{q,0},~~ { \rm h}\vert z~~{\rm(chain~axis ) } \\
\sin \theta ~ \delta_{q,\pi},~~ { \rm h}\perp z \end{cases}.\end{aligned}\ ] ] it is important to notice that this result holds true for both families ( a , b ) of chains .
next , we observe that since in the dy@xmath203-based scm , the spins on opposite sublattices are coplanar with the chain axis , the angle between @xmath222 and @xmath223 is just @xmath246 .
taking into account that the true " exchange constant in eq . ( [ heisenberg ] ) is antiferromagnetic , @xmath208 , from eq .
( [ geii ] ) it follows that the effective ising exchange coupling is now ferromagnetic , @xmath7 ( see fig . 5 ,
top ) . as a consequence , in the low temperature limit @xmath247 ,
the relaxation time @xmath70 diverges , while @xmath71 does not .
thus , the _ a.c .
_ susceptibility @xmath248 is expected to have a resonant behavior , for low frequencies @xmath157 , only when the oscillating magnetic field is applied parallel to the chain axis , @xmath163 , along which the magnetizations of the two sublattices are _ uncompensated _ ( see fig . 5 , top ) .
such a theoretical prediction turns out to be in excellent agreement with the experimental _ a.c .
_ susceptibility data@xcite obtained in a single crystal sample of [ dy(hfac)@xmath231(nitphoph)]@xmath249 , thus confirming that also the dy@xmath203-based canted antiferromagnet is a _ bona fide _ scm .
the only qualitative difference , with respect to the mn@xmath203-based chain is that , due to the different geometry of the spin arrangement and of the local anisotropy axes with respect to the chain axis , the resonant behavior of the _ a.c . _
susceptibility is now observed for field applied parallel to the chain axis , rather than perpendicular to it . in the molecular magnetic compound of formula [ co(hfac)@xmath204nitphome ] , hereafter denoted by cophome
@xcite , the magnetic contribution is given by cobalt ions , with an ising character and effective @xmath251 , and by nitphome organic radical ions , magnetically isotropic and with @xmath234 .
the spins are arranged on a helical structure , schematically depicted in fig .
6 , right , whose projections in a plane perpendicular to the helix axis @xmath163 ( coincident with the crystallographic @xmath212 axis ) , are represented in fig .
the primitive magnetic cell is made up of three cobalts ( black arrows ) and three organic radicals ( red arrows ) .
although the effective spins of the two types of magnetic centers have the same value , the gyromagnetic factors are different : @xmath252 ; thus , since the nearest neighbor ( cobalt - radical ) exchange interaction is negative ( and strong , @xmath253 k ) @xcite , the sublattice magnetizations are not compensated along @xmath163 , whereas they are compensated within the @xmath254 plane perpendicular to the chain axis @xmath163 . for this compound , which was the first to display scm behavior @xcite , static measurements on single - crystal samples has not been interpreted in terms of a simple model yet , due to the complexity of the system itself .
thus , a relationship such as and , which associate the ising hamiltonian parameters ( @xmath6 and @xmath21 ) with those of a more realistic hamiltonian , is still missing .
however , the dynamic behavior has been thoroughly investigated treating - for the sake of simplicity - both the co@xmath250 and the organic radical spins as ising variables , with @xmath255 .
the effective ising hamiltonian reads
@xmath256 + e^{-i\omega t } \mu_b { \rm h } \cr & & \left [ g_r \sigma_{l,2m-1 } ( \hat{\mathbf{z}}_{2m-1 } \cdot \hat{\mathbf{e}}_{\rm h } ) + g_{co } \sigma_{l,2 m } ( \hat{\mathbf{z}}_{2 m } \cdot \hat{\mathbf{e}}_{\rm h})\right ] \}\end{aligned}\ ] ] with @xmath257 magnetic cell index and @xmath258 site label with boundary conditions @xmath259 . [ cophome ] since all the local axes @xmath15 ( @xmath260 ) form the same angle @xmath261 with the @xmath163 axis , when a magnetic field is applied along @xmath163 , the ft of the generalized land factor is simply given by @xmath262\end{aligned}\ ] ] which , except for the prefactor @xmath263 , is quite similar to eq .
( [ collinear_altern ] ) for the collinear ising chain with alternating spins .
thus , the parallel _ a.c . _ susceptibility is @xmath264\,.\end{aligned}\ ] ] taking into account that the effective exchange coupling of cophome is negative ( @xmath8 ) , the antiferromagnetic branch of the parallel susceptibility is characterized by a diverging time scale @xmath265^{-1}$ ] at low temperature , so that , for low frequencies @xmath157 , @xmath266 displays a resonant behavior .
let us now consider the case of a field applied in the plane perpendicular to the chain axis . for h@xmath267 ( see fig .
6 , left ) one has , letting @xmath268 @xmath269\cr & & g^x_{2r}=g_{co } \sin\theta \cos\left[k_0 2r \right]\end{aligned}\ ] ] so that the ft takes the form @xmath270\big)\cr & = & { 1\over 4}\sin\theta [ ( g_{co}+g_r e^{i(q - k_0)})(\delta_{q , k_0}+\delta_{q,\pi+k_0})\cr & + & ( g_{co}+g_r e^{i(q+k_0)})(\delta_{q ,- k_0}+\delta_{q,\pi - k_0})]\end{aligned}\ ] ] where , as usual , we have exploited eq .
( [ boundary ] ) .
thus it follows that @xmath271 the corresponding relaxation times are @xmath272 and @xmath273 so that , summing the four contributions we obtain the perpendicular _ a.c . _ susceptibility @xmath274.\end{aligned}\ ] ] in conclusion , for the six - fold helix model with alternating spins and ising exchange coupling in eq .
( [ ising_co ] ) , the parallel and perpendicular components of the _ a.c .
_ susceptibility , @xmath266 and @xmath275 , display a behavior similar to that of a ferrimagnetic chain with alternating spins ( see ) and of an @xmath58-fold helical spin arrangement with equivalent spins ( see ) , respectively . in spite of the approximations involved in model ( [ ising_co ] ) to describe the real cophome molecular magnetic chain , the two calculated susceptibilities and ,
qualitatively reproduce the dynamic behavior of this compound @xcite .
in fact , no out - of - phase _ a.c .
_ susceptibility ( imaginary part ) is observed when the field is applied in the plane perpendicular to the chain axis , @xmath163 , for the experimental frequencies ( @xmath276 hz ) @xcite .
in contrast , when the oscillating field is parallel to @xmath163 , a resonant behavior is observed as a function of temperature .
even though our theoretical treatment holds only for small deviations from equilibrium , it is worth mentioning that the absence of slow relaxation for fields applied in the perpendicular plane is evidenced in the low temperature magnetization curve as well : at low enough temperatures , a finite - area hysteresis loop is present only when a static field is applied parallel to the chain axis , while no hysteresis is observed in the in - plane magnetization curve @xcite .
in conclusion , in the framework of a one - dimensional ising model with single spin - flip glauber dynamics , taking into account reciprocal non - collinearity of local anisotropy axes and the crystallographic ( laboratory ) frame , we have investigated : ( i ) the dynamics of magnetization reversal in zero field , and ( ii ) the response of the system to a weak magnetic field , oscillating in time .
we have shown that scm behavior is not only a feature of collinear ferro- and ferrimagnetic , but also of canted antiferromagnetic chains .
in particular , we have found that resonant behavior of the _ a.c . _
susceptibility versus temperature in response to an oscillating magnetic field is possible , at low frequency , only for fields applied in a direction where magnetic moments are uncompensated .
in contrast , a broad peak is expected when there is no net magnetization along the field direction .
the role played by geometry in selecting the time scales involved in a process is an important and peculiar result , typical of magneto - molecular approach to low - dimensional magnetism .
in fact , magnetic centers with uniaxial anisotropy usually correspond to building blocks with low symmetry @xcite , which in turn often crystallize in more symmetric space groups , realizing a reciprocal non - collinearity between local anisotropy axes as a natural consequence @xcite .
thus the family of real scm s , to which our model applies , does not restrict to _ ad - hoc _ synthesized compounds but , instead , is expected to grow larger in the future @xcite . as a validity check of our selection rules ( as well as a tutorial exemplification ) , we have shown how our theory applies successfully to three different molecular - based spin chains ; when possible , we have put the parameters of our model hamiltonian in relationship with those of more general models , typically used to fit the static properties of the corresponding compounds . needless to say that the possibility of schematizing the chosen three compounds with hamiltonian relies on the fact that at low enough temperatures they behave as chains consisting of two - level units coupled by a fully anisotropic exchange interaction .
the latter assumption is expected to hold also for spin larger than one - half in the presence of strong single ion anisotropy , provided that domain walls still remain sharp @xcite . in this case
each single magnetic center follows a thermally activated dynamics , with an energy barrier @xmath277 , and well established heuristic arguments @xcite suggest to replace the attempt frequency by @xmath278 . +
a naive application of our 3-fold - helix results and to the recently synthesized non - collinear dy@xmath279 cluster would prevent the observation of _ slow _ relaxation , while single - molecule - magnet dynamics is indeed there observed even in the presence of compensated magnetic moments @xcite .
however , such a behavior in the classical regime , _
i.e. _ far from level crossings where underbarrier processes of quantum origin are important , is observed for dy@xmath231 in non - zero field and the resonant behavior is due to a change of the relative population between the lowest and the first excited kramers doublets of each dy ion : for sure this mechanism can not be accounted for when dealing with two - level elementary variables , like @xmath280 in hamiltonian .
an extension of our model to multivalued @xmath280 s definitely deserve to be considered in the next future . beyond molecular spin chains , our approach
might also be used to model monatomic nanowires showing slow relaxation of the magnetization at low temperatures @xcite and , possibly , one - dimensional spin glasses @xcite ( provided that quenched disorder is somehow taken into account ) . in this regard ,
the question of distinguishing between scm and spin - glass behavior in quasi-1d systems is still a hot topic of discussion @xcite .
after the successful organization of single - molecule magnets onto surfaces @xcite , the grafting of properly functionalized scm s on substrates represents a foreseeable goal as well as a fundamental step for their possible use as magnetic - memory units @xcite .
technologies employing more traditional materials but based on alternative geometrical arrangement of magnetic anisotropy axes with respect to the switching field , such as in perpendicular recording @xcite or processional switching @xcite , are already at the stage of forthcoming implementation in devices . were scm s to be considered as a possible route to tackle the main issues of high - density magnetic storage
_ i.e. _ optimization of the signal - noise ratio , thermal stability and writability @xcite the proposed selection rules for slow relaxation , and related bistability , might find an application in magnetic - memory manufacture as well .
we wish to thank r. sessoli and a. rettori for stimulating discussions , and j. villain for interest and fruitful suggestions in the early stages of this research work .
financial support from ethz , the swiss national foundation , and italian national research council is also acknowledged .
assuming periodic boundary conditions and defining the two generating functions @xmath281 eqs .
( [ newform ] ) take the form of two differential equations ( with the dot indicating the first derivative with respect to the adimensional variable @xmath282 ) @xmath283 the system can be decoupled through the substitution @xmath284 from which we directly get @xmath285 with @xmath286 .
the solutions of these two equations are @xmath287 and @xmath288 that , exploiting the property @xmath289 = \sum ^{+\infty } _ { k=-\infty } y ^{k}\ , \mathcal{i}_{k}(x)\ ] ] of the bessel functions of imaginary argument @xmath290 , can be rewritten as @xmath291 performing the inverse transformation of , the solutions to the system can be obtained @xmath292 \cr\cr & & \mathcal{g}(y , t)=\frac{1}{2}e^{-\alpha t}\times \cr & & \overset{+\infty } { \underset{k=-\infty}\sum } y ^{k}\ , \left [ \mathcal{u}(y , 0)\mathcal{i}_{k}(\gamma \alpha t)-(-1)^{k}\mathcal{w}(y , 0)\mathcal{i}_{k}(\gamma \alpha t)\right]\,.\end{aligned}\ ] ] then , separating the @xmath3-odd from the @xmath3-even terms in both sums , we get @xmath293\cr & & \mathcal{g}(y , t)=e^{-\alpha t}\overset{+\infty } { \underset{r=-\infty}\sum } [ y ^{2r}\mathcal{g}(y , 0)\mathcal{i}_{2r}(\gamma \alpha t)\cr & & + y ^{2r-1}\mathcal{l}(y , 0)\mathcal{i}_{2r-1}(\gamma \alpha t)]\,.\end{aligned}\ ] ] according to , now @xmath294 and @xmath295 can be expressed again in terms of the initial single spin expectation values @xmath296 and @xmath297 respectively @xmath298\end{aligned}\ ] ] putting @xmath299 we have @xmath300\,.\end{aligned}\ ] ] comparing this latter result with the definition of @xmath301 and requiring for the terms corresponding to the same power of @xmath175 to be equal , an explicit function for the odd spin expectation values is readily obtained @xmath302\,.\end{aligned}\ ] ] substituting @xmath303 e @xmath295 in the solution found for @xmath304 and performing similar passages , we obtain the expectation value for even sites @xmath305\,.\end{aligned}\ ] ] it is worth noticing that , once a good candidate were found , an experimental verification of the theoretical prediction might be not so easy since the considered relaxation is a strongly out - of - equilibrium process , while in our theory small departures from equilibrium were assumed . for a _ collinear _ heisenberg ferromagnet with exchange @xmath197 and anisotropy @xmath198 ,
the energy cost of a domain wall was calculated and compared with the energies of a sharp wall and of a soliton , and a the crossover between the sharp wall " regime ( @xmath306 ) and the broad wall " regime ( @xmath307 ) was found to occur @xcite for @xmath308 . in principle , a similar calculation should be performed also for the non - collinear model ( [ heisenberg ] ) , in order to find the limits of validity for the approximation made in eq .
( [ bipartite ] ) . | the magnetic behavior of molecular single - chain magnets is investigated in the framework of a one - dimensional ising model with single spin - flip glauber dynamics .
opportune modifications to the original theory are required in order to account for reciprocal non - collinearity of local anisotropy axes and the crystallographic ( laboratory ) frame .
the extension of glauber s theory to the case of a collinear ising ferrimagnetic chain is also discussed . within this formalism ,
both the dynamics of magnetization reversal in zero field and the response of the system to a weak magnetic field , oscillating in time , are studied .
depending on the geometry , selection rules are found for the occurrence of slow relaxation of the magnetization at low temperatures , as well as for resonant behavior of the _ a.c . _
susceptibility as a function of temperature at low frequencies .
the present theory applies successfully to some real systems , namely mn- , dy- , and co - based molecular magnetic chains , showing that single - chain - magnet behavior is not only a feature of collinear ferro- and ferrimagnetic , but also of canted antiferromagnetic chains . |
the cluster category was introduced independently in @xcite for type @xmath0 and in @xcite for the general case .
let @xmath7 be the bounded derived category of the finitely generated modules over a finite dimensional hereditary algebra @xmath8 over a field @xmath9 . in @xcite
the cluster category was defined as the orbit category @xmath10 $ ] , where @xmath11 is the auslander - reiten translation and [ 1 ] the suspension functor .
the cluster - tilted algebras are the algebras of the form @xmath12 , where @xmath1 is a cluster - tilting object in @xmath3 .
see @xcite .
let @xmath13 be a quiver with no multiple arrows , no loops and no oriented cycles of length two .
mutation of @xmath13 at vertex @xmath14 is a quiver @xmath15 obtained from @xmath13 in the following way . 1 .
add a vertex @xmath16 .
2 . if there is a path @xmath17 , then if there is an arrow from @xmath18 to @xmath19 , remove this arrow .
if there is no arrow from @xmath18 to @xmath19 , add an arrow from @xmath19 to @xmath18 .
3 . for any vertex @xmath19 replace all arrows from @xmath19 to @xmath14 with arrows from @xmath16 to @xmath19 , and replace all arrows from @xmath14 to @xmath19 with arrows from @xmath19 to @xmath16 .
4 . remove the vertex @xmath14 .
we say that a quiver @xmath13 is mutation equivalent to @xmath15 , if @xmath15 can be obtained from @xmath13 by a finite number of mutations .
the mutation class of @xmath13 is all quivers mutation equivalent to @xmath13 .
it is known from @xcite that the mutation class of a dynkin quiver @xmath13 is finite .
if @xmath20 is a cluster - tilted algebra , then we say that @xmath20 is of type @xmath0 if it arises from the cluster category of a path algebra of dynkin type @xmath0 .
let @xmath13 be a quiver of a cluster - tilted algebra @xmath20 . from @xcite
, it is known that if @xmath15 is obtained from @xmath13 by a finite number of mutations , then there is a cluster - tilted algebra @xmath21 with quiver @xmath15 .
moreover , @xmath20 is of finite representation type if and only if @xmath21 is of finite representation type @xcite .
we also have that @xmath20 is of type @xmath0 if and only if @xmath21 is of type @xmath0 . from @xcite we know that a cluster - tilted algebra is up to isomorphism uniquely determined by its quiver .
see also @xcite .
it follows from this that to count the number of cluster - tilted algebras of type @xmath0 , it is enough to count the mutation class of any quiver with underlying graph @xmath0 .
we recall some results from @xcite .
let @xmath23 be a positive integer and let @xmath24 be a regular polygon with @xmath22 vertices .
a diagonal is a straight line between two non - adjacent vertices on the border .
a triangulation is a maximal set of diagonals which do not cross . if @xmath25 is any triangulation of @xmath24 , we know that @xmath25 consists of exactly @xmath23 diagonals .
let @xmath26 be a diagonal between vertex @xmath27 and vertex @xmath28 on the border of @xmath24 . in @xcite
a _ pivoting elementary move _
@xmath29 is an anticlockwise move of @xmath26 to another diagonal @xmath30 about @xmath27 .
the vertices of @xmath30 are @xmath27 and @xmath31 , where @xmath28 and @xmath31 are vertices of a border edge and rotation is anticlockwise .
a _ pivoting path _ from @xmath26 to @xmath30 is a sequence of pivoting elementary moves starting at @xmath26 and ending at @xmath30 .
fix a positive integer @xmath23 .
categories of diagonals of regular @xmath32-polygons were introduced in @xcite .
let @xmath33 be the category with indecomposable objects all diagonals of the polygon , and we take as objects formal direct sums of these diagonals .
morphisms from @xmath26 to @xmath30 are generated by elementary pivoting moves modulo the mesh relations , which are defined as follows .
let @xmath26 and @xmath34 be diagonals , with @xmath35 and @xmath36 the vertices of @xmath26 and @xmath37 and @xmath38 the vertices of @xmath34 .
suppose @xmath39 takes @xmath26 to @xmath34 .
then @xmath40 . furthermore ,
if one of the intermediate edges in a pivoting elementary move is a border edge , this move is zero .
it is shown in @xcite that this category is equivalent to the cluster category defined in section 1 in the @xmath0 case .
we have the following from @xcite .
* the irreducible morphisms in @xmath33 are the direct sums of pivoting elementary moves .
* the auslander - reiten translation of a diagonal is given by clockwise rotation of the polygon .
* @xmath41 if and only if @xmath26 and @xmath42 do not cross .
it follows that a tilting object in @xmath3 corresponds to a triangulation of @xmath24 . for any triangulation @xmath25 of @xmath24 , it is possible to define a quiver @xmath43 with @xmath23 vertices in the following way .
the vertices of @xmath43 are the midpoints of the diagonals of @xmath25 .
there is an arrow between @xmath19 and @xmath18 in @xmath43 if the corresponding diagonals bound a common triangle .
the orientation is @xmath44 if the diagonal corresponding to @xmath18 is obtained from the diagonal corresponding to @xmath19 by rotating anticlockwise about their common vertex .
it is known from @xcite that all quivers obtained in this way are quivers of cluster - tilted algebras of type @xmath0 .
we defined the mutation of a quiver of a cluster - tilted algebra above .
we also define mutation of a triangulation at a given diagonal , by replacing this diagonal with another one .
this can be done in one and only one way .
let @xmath43 be a quiver corresponding to a triangulation @xmath25 .
then mutation of @xmath43 at the vertex @xmath19 corresponds to mutation of @xmath25 at the diagonal corresponding to @xmath19 .
it follows that any triangulation gives rise to a quiver of a cluster - tilted algebra , and that a quiver of a cluster - tilted algebra can be associated to at least one triangulation .
let @xmath45 be the mutation class of @xmath0 , i.e. all quivers obtained by repeated mutation from @xmath0 , up to isomorphisms of quivers .
let @xmath46 be the set of all triangulations of @xmath24 .
we can define a function @xmath47 , where we set @xmath48 for any triangulation @xmath25 in @xmath46 . note that @xmath49 is surjective .
if @xmath35 and @xmath36 are vertices on the border of a regular polygon , we say that the _ distance _ between @xmath35 and @xmath36 is the smallest number of border edges between them .
let us say that a diagonal from @xmath35 to @xmath36 is _ close to the border _ if the distance between @xmath35 and @xmath36 is exactly @xmath50 .
for a quiver @xmath43 corresponding to a triangulation @xmath25 , let us always write @xmath51 for the vertex of @xmath43 corresponding to the diagonal @xmath26 .
[ sinksourcecycle ] if a diagonal @xmath26 of a triangulation @xmath25 is close to the border , then the corresponding vertex @xmath52 in @xmath48 is either a source , a sink or lies on a cycle ( oriented of length @xmath53 ) .
all cycles are oriented and of length @xmath53 in the @xmath0 case .
suppose that @xmath26 is a diagonal in @xmath25 which is close to the border .
there are only three cases to consider , shown in figure [ figsinksourcecycle ] .
see the proof of lemma [ sinksourcecycle ] .
sink , source and cycle.,title="fig:",width=124 ] see the proof of lemma [ sinksourcecycle ] .
sink , source and cycle.,title="fig:",width=124 ] see the proof of lemma [ sinksourcecycle ] .
sink , source and cycle.,title="fig:",width=124 ] in the first case @xmath26 corresponds to a sink .
there is no other vertex adjacent to @xmath52 but @xmath54 , or else the corresponding diagonal of this vertex would cross @xmath34 .
we have the same for the second case where @xmath26 is a source . in the third case @xmath52
lies on a cycle .
[ lemconnected ] let @xmath25 be a triangulation and let @xmath48 be the corresponding quiver .
a quiver @xmath15 obtained from @xmath43 by factoring out a vertex @xmath52 is connected if and only if the corresponding diagonal @xmath26 is close to the border .
suppose @xmath26 is close to the border .
by lemma [ sinksourcecycle ] , @xmath26 corresponds to a sink , a source or a vertex on a cycle .
if @xmath51 is a sink or a source then @xmath51 has only one adjacent vertex , so factoring out @xmath51 does not disconnect the quiver .
suppose @xmath51 lies on a cycle .
then we are in the case shown in the third picture in figure [ figsinksourcecycle ] .
we see that there can be no other vertex adjacent to @xmath51 except @xmath55 and @xmath56 , since else the corresponding diagonal would cross @xmath34 or @xmath57 . hence factoring out @xmath51
does not disconnect the quiver .
next , suppose that factoring out @xmath52 does not disconnect the quiver . if @xmath52 is a source or a sink with only one adjacent vertex , then @xmath52 is close to the border . if not
, first suppose @xmath52 does not lie on a cycle .
then it is clear that factoring out @xmath52 disconnects the quiver , so we may assume that @xmath52 lies on a cycle .
then @xmath26 is an edge of a triangle consisting of only diagonals ( i.e. no border edges ) , say @xmath34 and @xmath57 .
suppose there is a vertex @xmath58 adjacent to @xmath52 , with @xmath59 and @xmath60
. then @xmath58 can not be adjacent to @xmath54 or @xmath56 , since then we would have two cycles sharing one arrow .
we also see that @xmath58 can not be adjacent to any vertex @xmath61 from which there exists a path to @xmath54 or @xmath56 not containing @xmath52 , or else there would be a cycle of length greater than @xmath53 . therefore factoring out
@xmath52 would disconnect the quiver , and this is a contradiction , thus there can be no other vertices adjacent to @xmath52 .
it follows that @xmath26 can not be adjacent to any other diagonal but @xmath34 and @xmath57 , hence @xmath26 is close to the border .
let @xmath25 be a triangulation of @xmath24 and let @xmath26 be a diagonal close to the border .
the triangulation @xmath62 of @xmath63 obtained from @xmath25 by factoring out @xmath26 is defined as the triangulation of @xmath63 by letting @xmath26 be a border edge and leaving all the other diagonals unchanged .
we write @xmath64 for the new triangulation obtained
. see figure [ figfactoring ] .
[ lemfactoring ] let @xmath25 be a triangulation and @xmath48 .
factoring out a vertex in @xmath43 such that the resulting quiver is connected , corresponds to factoring out a diagonal of @xmath25 close to the border .
factoring out a vertex @xmath52 in @xmath13 such that the resulting quiver is connected , implies that @xmath26 is close to the border by lemma [ lemconnected ] .
then consider all cases shown in figure [ figsinksourcecycle ] .
[ factor out ] let @xmath13 be a quiver of a cluster - tilted algebra of type @xmath0 , with @xmath66 .
let @xmath15 be obtained from @xmath13 by factoring out a vertex such that @xmath15 is connected .
then @xmath15 is the quiver of some cluster - tilted algebra of type @xmath67 .
it is already known from @xcite that @xmath15 is the quiver of a cluster - tilted algebra .
suppose @xmath25 is a triangulation of @xmath24 such that @xmath68 .
such a @xmath25 exists since @xmath49 is surjective .
it is enough , by lemma [ lemconnected ] , to consider vertices corresponding to a diagonal close to the border . by lemma [ lemfactoring ] , factoring out
a vertex corresponding to a diagonal @xmath26 close to the border , corresponds to factoring out @xmath26 .
then the resulting triangulation of @xmath69 corresponds to a quiver of a cluster - tilted algebra of type @xmath67 , since it is a triangulation .
now we want to do the opposite of factoring out a vertex close to the border .
if @xmath25 is a triangulation of @xmath24 , we want to add a diagonal @xmath26 such that @xmath26 is a diagonal close to the border and such that @xmath70 is a triangulation of @xmath71 .
consider any border edge @xmath72 on @xmath24 .
then we have one of the cases shown in figure [ figbordercases ] .
it follows that for a given diagonal @xmath34 , there are at most three ways to extend the polygon with a diagonal @xmath26 such that @xmath26 is adjacent to @xmath34 , and it is easy to see that these extensions gives non - isomorphic quivers .
for a triangulation @xmath25 of @xmath24 , let us denote by @xmath73 the triangulation obtained from @xmath25 by rotating @xmath25 @xmath19 steps in the clockwise direction .
we define an equivalence relation on @xmath46 , where we let @xmath74 for all @xmath19 .
we define a new function @xmath75 induced from @xmath49 .
this is well defined , for if @xmath76 for an @xmath19 , then obviously @xmath77 in @xmath45 . and hence since @xmath49 is a surjection , we also have that @xmath78 is a surjection .
we actually have the following .
it is easy to check that @xmath84 is injective .
suppose @xmath85 is injective .
let @xmath26 be a diagonal close to the border in @xmath25 , with image @xmath52 in @xmath13 , where @xmath13 is a representative for @xmath86 .
then the diagonal @xmath42 in @xmath87 corresponding to @xmath52 in @xmath13 is also close to the border .
we have @xmath88 by lemma [ lemfactoring ] , and hence , by hypothesis , @xmath89 in @xmath83 .
we can obtain @xmath25 and @xmath87 from @xmath89 by extending the polygon at some border edge . fix a diagonal @xmath34 in @xmath25 such that @xmath52 and @xmath54 are adjacent .
this can be done since @xmath13 is connected .
let @xmath90 be the diagonal in @xmath62 corresponding to @xmath54 . by the above
there are at most three ways to extend @xmath64 such that the new diagonal is adjacent to @xmath34 .
it is clear that these extensions will be mapped by @xmath78 to non - isomorphic quivers .
also there are at most three ways to extend @xmath91 such that the new diagonal is adjacent to @xmath57 , and all these extensions are mapped to non - isomorphic quivers , thus @xmath82 in @xmath83 .
now , let @xmath1 be a cluster - tilting object of the cluster category @xmath3 .
this object corresponds to a triangulation @xmath25 of @xmath24 , and all tilting objects obtained from rotation of @xmath25 gives the same cluster - tilted algebra .
no other triangulation gives rise to the same cluster - tilted algebra .
the number @xmath97 of non - isomorphic basic cluster - tilted algebras of type @xmath0 is the number of triangulations of the disk with @xmath23 diagonals , i.e. @xmath98 where @xmath94 is the @xmath19th catalan number and the second term is omitted if @xmath99 is not an integer and the third term is omitted if @xmath100 is not an integer .
let @xmath25 be the triangulation of @xmath24 corresponding to @xmath1 and let @xmath62 be the triangulation corresponding to @xmath2 . if @xmath104 for any @xmath19 , then @xmath62 is not obtained from @xmath25 by a rotation , and hence @xmath4 is not isomorphic to @xmath5 by theorem [ quivers not iso ] .
let @xmath1 be a tilting object in @xmath3 corresponding to the triangulation @xmath25 .
denote by @xmath73 the rotation of @xmath25 @xmath19 steps in the clockwise direction .
let @xmath106 be the smallest number of rotations needed to obtain the same triangulation @xmath25 , i.e. the smallest @xmath107 such that @xmath108 .
it is clear from the above that @xmath109 , where @xmath2 corresponds to @xmath110 with @xmath111 , hence @xmath107 is the number of non - isomorphic tilting objects giving the same cluster - tilted algebra .
now we only need to show that @xmath107 divides @xmath22 , but this is clear . the proof of the following is easy and is left to the reader .
first recall from ( * ? ? ?
* proposition 3.8 ) that there are exactly @xmath112 non - isomorphic tilting objects in the cluster category for type @xmath0 , where @xmath112 denotes the @xmath23th catalan number .
* there are always at least @xmath50 non - isomorphic cluster - tilting objects giving the same cluster - tilted algebra .
* there are at most @xmath22 non - isomorphic cluster - tilting objects giving the same cluster - tilted algebra .
* let @xmath20 be a cluster - tilted algebra of type @xmath0 . if @xmath22 is prime , there are exactly @xmath22 non - isomorphic cluster - tilting objects giving @xmath20 . in this case
there are @xmath113 non - isomorphic cluster - tilted algebras , where @xmath112 denotes the @xmath23th catalan number . | the purpose of this paper is to give an explicit formula for the number of non - isomorphic cluster - tilted algebras of type @xmath0 , by counting the mutation class of any quiver with underlying graph @xmath0 .
it will also follow that if @xmath1 and @xmath2 are cluster - tilting objects in a cluster category @xmath3 , then @xmath4 is isomorphic to @xmath5 if and only if @xmath6 . |
about 50% of well - localized grbs show optical transients ( ots ) successive to the prompt gamma - ray emission , whereas an x - ray counterpart is present in 90% of cases .
it is possible that late and shallow observations could not detect the ots in some cases ; several authors argue that dim and/or rapid decaying transients could bias the determination of the fraction of truly obscure grbs @xcite . however , recent reanalysis of optical observations @xcite has shown that grbs without ot detection ( usually dark grbs , foas failed optical afterglows , or ghosts , gamma ray burst hiding an optical source transient ) have had on average weaker optical counterparts , at least 2 magnitudes in the r band , than grbs with ots . therefore , they appear to constitute a different class of objects , albeit there could be a fraction undetected for bad imaging .
two hypothesis have been put forward to explain the behaviour of ghosts .
first , they are similar to the other bright grbs , except for the fact that their lines of sight pass through large and dusty molecular clouds , that cause high absorption .
second , they are more distant than grbs with ot , at @xmath3 @xcite , so that the lyman break is redshifted into the optical band . these grbs might be associated with the explosion of ancient population iii , high mass stars .
nevertheless , the distances of a few dark grbs have been determined and they do not imply high redshifts @xcite goal of this paper is an analysis of a complete sample of bepposax x - ray afterglows in order to distinguish between these various scenarios , including all x - ray fast observation from the launch to february 2001 . in 2 and 3 we present the data analysis of the afterglows and we show the results , whose implications are discussed in 4 .
finally , we summarize our conclusions in 5 .
we have analyzed all the 31 fast bepposax observations of grb x - ray afterglows taken by the low energy ( 0.1 - 10 kev ) and medium energy ( 1.6 - 10 kev ) concentrator spectrometer ( lecs and mecs respectively , see parmar et al .
1997 , boella et al .
1997 ) up to grb010222 .
we excluded only grb960720 for the late follow up , grb990705 due to its high contamination of a nearby x - ray source , and grb980425 due to its peculiarity .
x - ray follow up observations usually start @xmath4 hours after the high energy event and the typical observation time is @xmath5 seconds for mecs and @xmath6 for lecs . the exposure - or integration - lasts @xmath7 of the observation . in order to find out the grb x - ray afterglows ,
we first have built up the images of each grb with the mecs and selected sources with @xmath8 significance within the wfc error box .
successively , we have built the light curves of these sources to recognize afterglows through their typical fading emission .
the counts have been collected within a circle centered on source with radius @xmath9 arcmin .
then we have subtracted the background collected in annuli around the extraction area and 5 times more extended .
local background have been used in order to take into account possible time fluctuations .
toos successive to the first one ( typically @xmath10 day after ) have been used , if available .
we have fitted the light curves with simple power law @xmath11 ( where @xmath12 is counts per second ) and 26 sources with decaying index @xmath13 ( at @xmath14 confidence level ) have been recognized as grb afterglows . in the case of grb970111 , grb991106 and grb000615 we have detected 1 source within wfc error box that does not show a significant fading behaviour
. we will refer to them as `` candidate '' afterglows .
we have calculated the probability to have serendipitous sources with their flux within the wfc error box , adopting the log n - log s distribution for bepposax released by @xcite .
the probability is @xmath15 for each one , while the probability that all of them are not afterglows is @xmath16 .
the mecs integration time for grb990907 was only 1070 seconds , so that the presence of a fading flux could not be verified .
the x - ray source detected was recognized as the grb afterglow because the probability to have a serendipitous source with flux @xmath17 erg @xmath18 sec@xmath19 ( see further ) in the wfc error box was @xmath20 .
finally , in the case of grb990217 and grb010220 , we have not detected any source with @xmath21 significance .
to obtain flux , we have produced spectra for the afterglows from lecs and mecs first too data . for absorption and spectral index ,
we have selected those with more than 150 photons in the mecs ( background subtracted ) . 5 ghosts and
9 otgrbs passed this criterion .
we have generally taken the lecs data between 0.1 and 4.0 kev and the mecs data between 1.6 and 10 kev .
the backgrounds we have used are the library ones because they have a very good signal - to - noise ratio , due to long exposition .
however , we have taken the minimum energy for lecs to be 0.4 kev if the galactic column density was @xmath22 @xmath18 because in this case the low energy backgrounds differ from the library ones , which have been taken at high galactic latitudes and lower column densities @xcite .
if we had not adopted this criterion , our analysis would have led to overestimate the true absorption at the source .
the standard spectrum model to fit the data consists of a constant , galactic absorption , extragalactic absorption ( i.e. _ in situ _ ) and a power law .
the constant has been included because lecs and mecs observe a decaying source at different times .
its value is allowed to vary within a range , obtained in each case by fitting lecs and mecs data in the 1.6 - 4 kev interval ( to avoid absorption effects ) with a simple power law model .
the redshift in our fits has been forced to be 1 for all bursts .
this value corresponds roughly to the average redshift of otgrbs .
we have adopted this `` working hypothesis '' to obtain a homogeneous set that allows us to compare the absorption properties of dark grbs in the assumption that they are at the same distance .
we have calculated the 1.6 - 10 kev flux of dark and bright grbs 11 hours after the burst trigger .
we have chosen this time to avoid effects of changes in decaying slope .
the average count rate in the mecs has been associated with the flux given by the spectrum .
successively , we have taken the count rate at 11 hours , which is given by light curves , to compute the flux at that time . in most cases ,
observations include it . in a few cases ( e.g. grb000926 ) the flux has been extrapolated . for grb990907 ,
the counts collected were very few and we have not been able to do any spectral analysis .
we have estimated the flux assuming a spectral index @xmath23 .
for the two non - detections , we have calculated the @xmath8 upper limits on counts and converted them to flux adopting again @xmath24 . in all successive analysis ,
upper limits have been included as true afterglows as well as candidate afterglows . as a first assessment of our study , we can say that x - ray afterglows follow the prompt gamma emission in 26 of 31 cases , which constitute 84% of the sample .
if all doubtful sources are considered as afterglows , then the fraction of x - ray afterglows increases up to @xmath25 . instead
, optical afterglows are 11 and constitute only @xmath26 of the sample .
we note that all these fractions are in agreement with published data .
we do not know any optical study on grb980515 .
we have calculated its x - ray flux but this burst has not been included in our successive analysis .
the data we have obtained are the result of the convolution of the intrinsic distribution with the measurement error distribution . under the assumption that both are gaussian
, it is possible to deconvolve the two distribution .
we have followed a maximum likelihood method @xcite to gather jointly the best estimates of parent distribution mean and standard deviation .
we have used these best estimates ( hereafter , indicated with index _ m _ ) for successive analysis , but we have calculated and shown also the weighted mean and standard deviation of our data .
the complete set of fit parameters is given in table 1 and plotted in figures 1 and 2 .
for ghosts , the weighted mean and the standard deviation of the measured energy indexes are @xmath27 ( hereafter errors are at @xmath28 , unless otherwise indicated ) and @xmath29 respectively .
the best estimates for the parent population are @xmath30 , @xmath31 . in the case of otgrbs , @xmath32 , @xmath33 and @xmath34 , @xmath35 for
the observed and the parent distribution respectively .
energy indexes of dark and optically bright burst are compatible at @xmath36 level . the mean value and the standard ( linear ) deviation of the measured absorption ( hereafter , in units of @xmath37 ) are , respectively , @xmath38 , @xmath39 for dark grbs , and @xmath40 , @xmath41 for otgrbs .
the best estimates for the parent population are @xmath42 , @xmath43 for dark grbs , and @xmath44 , @xmath45 ( see also stratta et al .
2002 ) for otgrbs . _
the amount of absorption does not appear statistically different for optically bright and dark grbs in the assumption that they lie at the same average @xmath46 .
_ the logarithmic weighted means and the standard deviations of the observed x - ray fluxes ( c.g.s .
units ) are @xmath47 log @xmath48 @xmath49 @xmath50 , @xmath51 for dark grbs and + @xmath47log @xmath48 @xmath49 @xmath52 , @xmath53 for otgrbs .
best estimates for the parent population are @xmath47log @xmath48 @xmath54 = @xmath55 , @xmath56 for dark grbs and @xmath47 log @xmath57 = @xmath58 , @xmath59 for grbs with ots .
the ghost mean flux is likely overestimated , because we have considered upper limits as detections .
the logarithm of ratio between the mean fluxes of the two parent populations is @xmath60 , which corresponds to 4.8 in linear units .
a k - s test performed on the flux distributions shows that the probability that optically bright and dark grbs derive from the same population is @xmath61 .
this is a conservative result , because it has been obtained by including the upper limits and the non - fading sources as true afterglows in the set of dark grbs .
if we were to substitute the the non - fading source fluxes with the @xmath8 upper limits of their wfc error boxes , then the distributions of dark and optically bright grbs would be even more different because limits are lower .
_ our analysis shows that dark grbs have on average weaker x - ray flux than bright grbs . _ then , we could simply explain why we miss their optical detection by assuming that dark bursts are weaker than ot grbs in the optical band by the same ratio .
dark bursts should have had ots at least 2 magnitudes fainter than ot grbs ; the 4.8 flux ratio that we have found corresponds to @xmath62 magnitudes . in order to check the viability of this hypothesis , we have calculated the optical flux density in the r band and hence the optical - to - x flux ratios ( hereafter @xmath63 ) of each otgrb and ghost 11 hours after the burst ( lazzati et al .
2002 , fynbo et al .
2001 and reference therein ; vreeswijck et al .
1999 ; masetti et al . 2001 ) .
upper limits on optical fluxes of ghosts have been extrapolated from the tightest constraint available and adopting an optical flux decaying index @xmath64 .
results are shown in table 1 and plotted in figures 3 , 4 and 5 .
we note that the optical and x ray fluxes of otgrbs _ are correlated : _ the higher is the the x - ray flux , the more luminous is the optical counterpart .
the probability that it occurs by chance is only @xmath65 .
the logarithmic standard deviation of @xmath63 s is @xmath66 , which corresponds to a multiplicative factor of 2.6 , while the logarithmic mean is @xmath47 log @xmath67 @xmath54 = @xmath68 if the x - ray and optical fluxes are expressed in @xmath69 erg @xmath18 sec@xmath19 and @xmath70jy respectively .
we have fitted the distribution of x - ray and optical fluxes with the function log @xmath71 = k + @xmath72 log(f@xmath73 .
the best fit values are @xmath74 , k = @xmath75 .
we have also calculated the average optical - to - x spectral index , @xmath76 , as function of @xmath47 log @xmath77 @xmath54 , by adopting the x - ray and optical density flux at 2 kev and r band respectively and x - ray spectral index @xmath78 .
our result is @xmath79 . if we exclude grb980519 , which seems to be the only afterglow explained by interaction of a jet outflow with a star wind medium @xcite , the correlation is strengthened : @xmath80 , which corresponds to a multiplicative factor of 1.9 ; the probability of a chance occurrence is @xmath81 ; @xmath47 log @xmath82 = @xmath83 .
the best fit values are @xmath84 , @xmath85 .
we can immediately recognize that @xmath86 of the ghosts of our sample ( 14 of 19 ) have optical flux upper limits consistent with ot detections ( see fig .
3 ) , so that they may not be actually `` dark '' .
optical follow - ups conducted for these bursts would not have been deep enough to detect the faintest ots in our set .
a similar fraction has been found out by @xcite and @xcite , comparing sets of non - detections with the light curves of the dim afterglows of grb000630 and grb020124 .
it is worth noting that the @xmath63 upper limits of these 14 objects are quite similar to otgrbs ones : a k - s test performed shows that the probability they belong to the same population is not marginal is not constrained .
however , wherever it is , it would not affect much the result . ] .
this result gives support to the fact that they could be faint sources with optical properties assimilable to x - ray ones .
the remaining 5 objects , which constitute @xmath87 of ghosts , have optical flux upper limits lower than all otgrbs in our set ( see fig .
2 ) . furthermore , their optical emission must have been even 2 - 3 times fainter than the dim afterglow of grb000630 and 4 - 6 fainter than grb020124 @xcite , which had an r band flux @xmath88 jy and @xmath89 jy respectively .
these two optical afterglows , however , are not the weakest ever occurred . in our set ,
the ot of grb980613 is even dimmer ( see table 1 ) and establishes a more stringent test .
we wonder if the hypothesis of weaker flux at all wavelenghts can hold for these 5 ghosts ( hereafter , we will refer to them as `` the darkest '' or the `` most obscure '' , etc . for simplicity ) .
if so , we would expect that their x - ray fluxes were proportionally very weak like the optical fluxes , so that their @xmath63 s should be not very different from otgrbs . in 4 cases of 5
the @xmath63 s are lower than all otgrbs and @xmath90 times lower than the average optical - to - x flux ratio of otgrbs .
the exception is grb990217 , which has upper limits both in optical and in x - rays flux .
if we use both of them , then we get log @xmath91 which is much more similar to @xmath47 log @xmath67 @xmath54 .
we have performed a k - s test on @xmath63 between all these darkest bursts and all otgrbs .
the probability that they are drawn from the same distribution is @xmath92 ( @xmath93 confidence level ) . the average optical - to - x spectral index of these objects @xmath94 , well below that of otgrbs .
therefore , we have a strong indication that for these bursts the spectrum is depleted in the optical band , by @xmath95 magnitude on average .
the x - ray mean flux of these 5 ghosts is @xmath96 .
the logarithmic ratio between the otgrb x - ray flux mean and this mean is log @xmath97 , which corresponds to a 4.3 factor .
a hypothesis for the absence of ot and fainter x - ray flux is that of very high redshift .
the grb prompt emission and x - ray afterglow of the strongest bursts ( e.g. grb990123 and grb990510 ) could be detectable even they occur at @xmath98 @xcite .
however , if ghosts were at @xmath99 , then extragalactic hydrogen clouds would entirely wash out optical emission @xcite . to estimate the average redshift of the most obscure grbs
, we shall use the formula @xcite : @xmath100 where @xmath101 is the spectral index , @xmath102 is the decaying ( temporal ) index and @xmath103 is the comoving distance .
we assume the cosmological parameters values @xmath104 , @xmath105 , @xmath106 .
the average of the known redshifts of otgrbs in our set is @xmath107 . in the simplest model of grb afterglows , @xmath108
, so @xmath109 .
for such parameters , the average redshift of the darkest grbs should be @xmath110 under the assumption that the lower mean flux were only due to their larger distances and not to an intrinsic difference in their luminosity .
using the best estimate of @xmath111 calculated for otgrbs and the average @xmath112 of the strongest bursts of our sample , we obtain @xmath113 .
we should also expect a distribution of burst redshifts around @xmath114 .
these facts make the high redshift scenario for the most obscure ghosts still plausible .
adopting the hypothesis grbs are the final result of very massive star evolution , an interesting issue to address is what might be the progenitors of grbs at very high redshifts .
currently , we observe only old and low - mass population ii stars , but even high mass stars could have formed .
theories suggest that the first stars of the universe - the so - called population iii - might have very large mass , so they could possibly be good candidates .
recent calculations suggest @xcite that the star formation rate has two peaks .
the first one , at @xmath115 is due to population iii stars .
the second one , due to population ii , is higher and much broader and it is at a redshift in the range @xmath116 .
also the number of stars ( i.e. the star formation rate time - dilated and weighted by the comoving volume of the universe ) shows two peaks at @xmath117 and @xmath118 . in a few cases ,
however , the redshifts of some dark grbs have been almost securely found , e.g. grb970828 at @xmath119 @xcite , grb 000210 at @xmath120 @xcite , grb 000214 at @xmath121 @xcite , while grb981226 is also likely to have not occurred at very high redshift @xcite , because the candidate host galaxy is still detected in the r band . with present statistics ,
at least @xmath122 of the examined dark grbs are not at very large redshift .
it should be noticed that two of them are included in the list of most obscure objects in our set . an hypothesis to explain the lack of the optical emission , alternative to the very high redshift scenario , may be strong absorption @xcite .
so far , we have collected many indications that grbs take place in dense environments , like the giant molecular clouds ( hereafter gmcs ) @xcite .
gmcs are very rich in dust , which extinguishes very efficiently the optical and uv light .
@xcite argues that in the case of grb000210 the lower limit on amount of obscuration is 1.6 magnitude in the r band .
this value has been obtained extrapolating a power law spectrum , described by the fireball model , from the x - ray band to the optical band and comparing the expected flux with the upper limits .
we find a similar result through our model - independent analysis of optical - to - x flux ratios .
the @xmath63 upper limits of the burst is 3.8 times lower than @xmath123 , which corresponds to @xmath124 magnitude depletion .
the measurements of chandra x - ray observatory showed that the amount of local absorbtion is able to explain this obscuration , under the assumption that the dust - to - gas ratio of the intervening medium is the same of the galaxy .
however , we note that in the case of otgrbs the dust - to - gas ratio seems not to be consistent with the galactic one @xcite .
similarly , @xcite derived extinction in the case of ghost grb970828 @xcite , for which a significative amount of x - ray absorption was detected .
if the most obscure ghosts were similar to grbs with ot except for higher absorption , we would expect to see differences in values of @xmath125 . from our results
, we can not affirm that @xmath125 in these bursts shows this tendency , also due to considerable errors ( see table 1 and figure 1 ) .
for those with good statistics , we do not find any absorption value @xmath126 higher than the galactic value but marginal evidence ( @xmath127 ) . on the other hand , we can not rule out the hypothesis of obscuring gmcs altogether .
the upper limits on @xmath125 , a few @xmath128 , are in fact the typical column densities of gmcs . _ the optical absorption , however , does not imply that most obscure ghosts have x - ray flux weaker than otgrbs , as we have found in our analysis , _ because x - ray absorption is almost negligible at energy larger than 1.6 kev .
reichart & yost ( 2001 ) try to reconcile this fact with the hypothesis of dusty birthplaces for grbs and , in particular , they considered the effect of variously beamed grb fireballs on their dusty environments .
the energetics of grbs are more or less the same for all events @xcite , but the beaming angles differ , being narrower for stronger bursts .
the larger the beaming angle is , the more difficult it is for the diluited prompt uv and x - ray emission to destroy dust along the line of sight ( waxman & draine 2000 ; fruchter et al .
2001 ; draine & hao 2002 ; perna & lazzati 2002 ) , so that we see a weak grbs without ot . with a narrower beaming angle ,
the prompt emission will destroy a larger fraction of dust and the grbs will appear strong and with ot . if this hypothesis is correct , on the basis of our results we have to assume that the average beaming angle of the darkest ghosts is @xmath10 times wider than the otgrb one . according to frail et al .
, the average beaming angle of bepposax ot grbs is @xmath129 , so that the average beaming angle of the darkest grbs should be @xmath130 .
this prediction is important , because it can be experimentally tested by observing and timing the presence of achromatic breaks in the light curves .
another consequence of dark grb occurrence in high density environments should be the detection of semi - ionized absorber in the low energy x - ray spectrum .
so far , this kind of feature has not been found .
the ionization front , however , should be rather sharp ( see e.g. draine & hao 2002 ) , and therefore it would be hard to detect signatures of semi - ionized species in the x - ray spectra of the bursts .
we have discussed the issue of grbs with x - ray but no optical afterglows .
we have performed a standard temporal and spectral analysis of a complete sample of 31 grb x - ray follow up observations of bepposax , i.e. all the fast observations from the launch until february 2001 .
we have found that x - ray afterglows follow the prompt gamma emission in @xmath131 of the cases .
we have obtained the 1.6 - 10 kev fluxes 11 hours after the trigger for each grb and the values of n@xmath132 at @xmath119 to compare the absorption properties for strong x - ray afterglows . while absorption of optically bright and dark grbs does not appear to be significatively different , the fluxes of grbs with ot are on average about 5 times stronger than ghost ones .
the probability that ghosts and optically bright grbs belong to the same population in fluxes is @xmath133 .
from the very fact that x - ray flux of dark grbs is 5 times lower than otgrbs , the optical flux could be @xmath10 magnitude lower under the assumption that the shape of the optical - to - x spectrum is the same of otgrbs .
this difference could explain the non - detection of the optical transient . in order to test this hypothesis ,
we have calculated the optical - to - x flux ratios of otgrbs and upper limits for ghosts .
otgrbs show a tight correlation of optical and x - ray fluxes .
the mean for otgrbs is @xmath47 log @xmath134 @xmath54 = @xmath135 and @xmath136 ; the probability of a chance correlation is a marginal @xmath137 .
we find that @xmath86 of ghosts have @xmath63 upper limits similar to otgrb ones ; however , the remaining @xmath138 of dark bursts are fainter in optical than in x - rays , being their average optical - to - x flux ratio @xmath47 log @xmath139 @xmath54 @xmath140 .
thus , we have a strong indication that for these bursts the spectra are different from otgrbs .
this result is significant at @xmath141 level .
two different interpretations for this effect can be given : 1 ) location at @xmath142 ; 2 ) higher absorption .
in the very high redshift scenario , the optical flux of the sample is extinguished by the intervening ly-@xmath101 systems , while the x - ray flux lower than otgrbs is understood in terms of a higher distance .
however , given the fact that some ghosts of the sample almost certainly do not lie at very high redshift , we have considered the alternative possibility of occurrence in dusty and dense environments like gmcs .
we have not found that these bursts have a higher absorption than optically bright grbs , but we note that upper limits on @xmath125 are consistent with those of giant clouds . in the case of grb000210
our model - independent analysis has shown a depletion in the optical which is compatible with the x - ray absorption measured by chandra , assuming a dust - to - gas ratio similar to that of our galaxy . in the near future , a key role will be played by fast and deep follow up x - ray and optical observations of grbs , which will allow us to constrain better their spectral properties . in particular , observations in the ir band are a really important tool because they are less sensitive to dust and to ly-@xmath101 extinction .
they will enable us to investigate dark grb properties like distance , that is a crucial piece of information to disclose the nature of these objects .
m.d.p . is grateful for support from university `` la sapienza '' grant .
bepposax was a joint program of italian space agency ( asi ) and netherlands agency for aerospace programs ( nivr ) .
we warmly thank all members of bepposax scientific operation center and operation control center . | we have examined the complete set of x - ray afterglow observations of dark and optically bright grbs performed by bepposax until february 2001 .
x - ray afterglows are detected in @xmath0 of the cases .
we do not find significant differences in the x - ray spectral shape , in particular no higher x - ray absorption in grbs without optical transient ( dark grbs ) compared to grbs with optical transient ( otgrbs ) .
rather , we find that the 1.6 - 10 kev flux of otgrbs is on average about 5 times larger than that of the dark grbs .
a k - s test shows that this difference is significant at 99.8% probability . under the assumption that dark and otgrb have similar spectra
, this could suggest that the first are uncaught in the optical band because they are just faint sources . in order to test this hypothesis ,
we have determined the optical - to - x ray flux ratios of the sample .
otgrbs show a remarkably narrow distribution of flux ratios , which corresponds to an average optical - to - x spectral index @xmath1 .
we find that , while 75% of dark grbs have flux ratio upper limits still consistent with those of ot grbs , the remaining 25% are 4 - 10 times weaker in optical than in x - rays .
the significance of this result is @xmath2 .
if this sub - population of dark grbs were constituted by objects assimilable to otgrbs , they should have shown optical fluxes higher than upper limits actually found .
we discuss the possible causes of their behaviour , including a possible occurrence in high density clouds or origin at very high redshift and a connection with ancient , population iii stars . |
a number of authors have argued that , at sub - eddington accretion rates , the gravitational potential energy released by turbulent stresses in an accretion flow may be stored as thermal energy , rather than radiated as in thin accretion disks ( ichimaru 1977 ; rees et al . 1982 ; narayan & yi 1994 , 1995 ; abramowicz et al .
1995 ; see narayan , mahadevan , & quataert 1998 and kato , fukue , & mineshige 1998 for reviews ) . in this case , the gas heats up to nearly virial temperatures and is assumed to adopt a two - temperature configuration , with the protons significantly hotter than the radiating electrons .
there are two crucial microphysical issues relevant to these accretion models , which are called either ion tori ( rees et al .
1982 ) or advection - dominated accretion flows ( adafs ; narayan & yi 1994 ) .
the first is the value of @xmath0 , the fraction of the turbulent energy in the plasma which heats the electrons ( a fraction @xmath1 heats the protons ) .
the second is whether or not the dominant process for exchanging energy between electrons and protons is coulomb collisions .
adaf models typically assume that @xmath2 ( preferential proton heating ) and that coulomb collisions represent the only thermal coupling . in this case
, the plasma is unable to cool because it is nearly collisionless . the protons , which by assumption receive the turbulent energy , are ( 1 ) unable to radiate and ( 2 ) unable to transfer their thermal energy to the electrons , the more efficient radiators .
the purpose of this review is to present some rather simple , but general , theoretical considerations regarding particle heating in adafs , which aim to address the value of @xmath0 .
a fundamental assumption of my analysis is that the particle distribution functions , while not necessarily strictly thermal , are not pathologically nonthermal ; that is , i assume that the plasma does not reach an equilibrium configuration with , e.g. , highly anisotropic or bump - in - tail distribution functions . for the simple considerations presented here , only two properties of the plasma are important .
the first is the proton to electron temperature ratio , @xmath3 , taken to be @xmath4 .
note that it is consistent to assume @xmath5 while simultaneously investigating particle heating .
this is because the two temperature nature of the plasma is in large part due to the efficiency with which relativistic electrons cool ; it is rather insensitive to @xmath0 .
the second property of importance is the magnetic field strength in the plasma , parameterized by @xmath6 .
the quantity @xmath7 must be @xmath8 for the gravitationally confined plasmas of interest . like thin disks , adafs are turbulent magnetized plasmas .
on scales much larger than the larmor radius of the thermal protons ( @xmath9 ) , we can treat the turbulence using mhd , i.e. , as a superposition of the magnetosonic and alfvn waves .
particles are heated when the turbulence cascades to sufficiently small scales that it is dissipated .
let a wave in the plasma have wave vector @xmath10 and frequency @xmath11 and define @xmath12 and @xmath13 to be directions perpendicular and parallel to the local magnetic field . in collisionless plasmas ,
the wave dissipation mechanisms of interest are wave - particle resonances ( molecular viscosity , thermal conductivity , electrical resistivity , etc .
are entirely unimportant ) . in the next section
, i will argue that the turbulence always has frequencies well below the proton cyclotron frequency . in this case , resonance occurs when the wave s phase speed along the field line , @xmath14 , equals @xmath15 , the velocity of a particle along the field line .
a necessary ( but not sufficient ) condition for strong damping is that @xmath16 be comparable to the thermal speed of the particles , so that there are a large number of resonant particles .
this resonance actually corresponds to two physically distinct wave - particle interactions . in landau damping ( ld ) , particle acceleration is due to the wave s longitudinal electric field perturbation ( i.e. , the usual electrostatic force , @xmath17 ) . for a wave damped solely by ld , @xmath18 since the electrons are preferentially heated by a factor of @xmath19 , which is @xmath4 for @xmath20 . in transit - time
damping ( ttd ) , the magnetic analogue of ld ( and the collisionless analogue of fermi acceleration ; achterberg 1981 ) , the interaction is between the particle s effective magnetic moment ( @xmath21 ) and the wave s longitudinal magnetic field perturbation , @xmath22 ( stix 1992 ) . for a wave ( with @xmath23 ) damped solely by ttd , the protons are preferentially heated by a factor of @xmath24 ; the contribution to @xmath0 is therefore always @xmath25 .
this is because in plasmas with @xmath5 , the protons have the larger magnetic moment and so couple better to a wave s magnetic field perturbation .
in a plasma with @xmath26 , one can ( roughly speaking ) decompose mhd turbulence into three modes , the alfvn wave and the fast and slow magnetosonic modes .
the fast mode has the dispersion relation @xmath27 ( @xmath28 is the proton thermal speed ) and is essentially a sound wave .
the slow mode has @xmath29 , has a parallel magnetic field perturbation ( @xmath22 ) .
the alfvn wave also has @xmath30 in the limit that the wavelength is long compared to the larmor radius of thermal protons ( @xmath31 ) .
sound waves in an unmagnetized collisionless plasma are strongly damped by ld .
for @xmath5 , the dissipation time is formally @xmath32 the mode period , implying that the modes are non - propagating .
these modes are unlikely to be efficiently excited since , quite generally , strongly damped oscillators dissipate less energy than weakly damped oscillators ( consider , e.g. , the analogous problem of two resistors in parallel ; the resistor with the smallest resistance dissipates the most power ) . the same holds true for sound waves in a @xmath26 plasma , provided @xmath33 . for @xmath34 , i.e. , nearly perpendicular sound waves , the parallel phase speed is @xmath35 ( since @xmath36 by @xmath5 ) . since the wave is highly suprathermal , there are very few particles for the wave to resonate with and it is consequently undamped .
these waves likely dissipate by steepening and forming weak shock waves . in what follows
i will neglect quasi - perpendicular sound waves , assuming that they are energetically unimportant .
the reason is that i believe that the turbulence in adafs is intrinsically magnetic in origin , being initiated by some combination of balbus - hawley and convective instabilities .
the turbulent speed ( @xmath37 ) is therefore @xmath38 .
moreover , i will argue that @xmath39 is likely to be @xmath40 , in which case , if @xmath41 , excitation of sound waves will be strongly suppressed ( subsonic instabilities do not easily excite sound waves ) . in a @xmath42 collisionless plasma ,
slow modes are strongly damped and non - propagating for @xmath43 ( foote & kulsrud 1979 ) .
the reason is that , even in the limit of wavelengths long compared to the larmor radius of thermal protons ( @xmath44 ) , slow modes have a parallel magnetic field perturbation ( @xmath45 ) and are damped by ttd .
they are , as with sound waves , unlikely to be excited . for @xmath46 ,
these considerations fail , since @xmath47 .
quasi - parallel slow modes are , however , likely to cascade in @xmath48 , develop @xmath45 , and damp by ttd ( heating the protons ) .
the alfvn wave has a parallel phase speed @xmath49 ; in plasmas appropriate to adafs , @xmath39 is comparable to the electron and proton thermal speeds and so there are a large number of particles available to resonate with the wave . in the mhd limit
, however , the alfvn wave is undamped by linear collisionless effects .
this is because @xmath50 and @xmath51 , i.e. , there is no electric / magnetic field perturbation associated with the wave which can accelerate particles .
this lack of dissipation on large scales implies that understanding heating of particles by alfvnic turbulence requires understanding how the energy cascades to small scales , where dissipation finally becomes important .
the nature of alfvnic turbulence in a collisionless , @xmath26 , plasma is not fully understood .
a plausible scenario is due to goldreich and sridhar ( 1995 ; gs ) .
they argue that alfvnic turbulence naturally evolves into a critically balanced state in which the timescale for nonlinear effects to transfer energy from a wavevector @xmath52 to a wavevector @xmath53 ( @xmath54 the cascade time , @xmath55 ) is comparable to the linear wave period at that scale , @xmath56 ; this determines how rapid the dissipation must be to halt the cascade .
it also implies that the cascade is highly anisotropic , with the energy cascading primarily perpendicular to the local magnetic field ; the parallel and perpendicular sizes of a wave at any scale are correlated , with @xmath57 , where @xmath58 is the outer scale of the turbulence .
quasi - perpendicular alfvnic turbulence has the important property that its frequencies are always much less than the proton cyclotron frequency .
this is because the cascade occurs primarily in @xmath48 while the alfvn frequency is @xmath59 .
this justifies my claim in 3 that low frequency wave - particle resonances are of primary importance .
the character of an alfvn wave changes when @xmath60 ; at this point the wave develops a non - zero parallel magnetic field perturbation ( in contrast to long wavelength perturbations , for which @xmath51 ) .
the waves can therefore be dissipated by ttd ( quataert 1998 , gruzinov 1998 , quataert & gruzinov 1998 ; qg ) .
figure 1 shows kinetic theory calculations of the dissipation of nearly perpendicular alfvn waves . for @xmath61 ( where the waves can still be called alfvn waves ) , the dissipation is independent of @xmath3 , but is a strong function of @xmath7 .
the electron temperature is unimportant since the electrons do not participate significantly in the damping .
the dissipation is sensitive to @xmath62 , since this determines the number of resonant protons and the protons magnetic moment ( which is the effective coupling coefficient between the protons and the wave ) . for large @xmath42 ,
alfvn waves are strongly damped for @xmath63 .
most of the turbulent energy heats the protons , since they are responsible for ttd ( 3 ) . for @xmath64 , however , the maximal dissipation rate for an alfvn wave in the gs cascade ( obtained at @xmath65 ) is @xmath66 ( see figure 1 ) .
since the timescale for energy to cascade through the inertial range is @xmath67 , this suggests that , for plasmas with equipartition magnetic fields ( @xmath68 ) , very little of the turbulent energy is dissipated on scales comparable to or greater than @xmath69 .
alfvn waves only exist for @xmath61 . for @xmath70 , the same mode
is called the whistler . as emphasized by gruzinov ( 1998 ;
see also qg ) , whistlers are unlikely to heat the protons . for @xmath71 , but @xmath72 , whistlers have @xmath73 .
thus , in a mode period , a particle undergoes many larmor orbits . since the mode s perpendicular wavelength is smaller than the proton larmor radius ( @xmath71 ) , the protons ( but not the electrons ) sample a rapidly varying electro - magnetic field in the course of a larmor orbit . as a result , they are `` frozen out '' and become dynamically unimportant ; the protons effectively no longer `` see '' the wave .
they therefore can not contribute to damping the whistler energy , which cascades to smaller length scales until it is damped by the electrons ( this is the origin of the strong damping at @xmath74 in figure 1 ) .
the above arguments suggest that the crucial physics determining the electron heating rate in adafs is what fraction of the turbulent energy is dissipated as alfvn waves ( heating the protons ) and what fraction cascades to small scales , becoming whistlers and heating the electrons .
this depends on @xmath7 , but not on @xmath3 .
unfortunately , estimates of the fraction of the turbulent energy that becomes whistlers are highly uncertain , as they are exponentially sensitive to the details of turbulence on the scale of the proton larmor radius , where both nonlinear and kinetic effects are important . for example , we do not accurately know @xmath55 , the nonlinear timescale of the turbulence . nor is the calculation of the damping of the turbulence that secure ( because it is based on a theory of linear alfvn waves damped by thermal particles ; both linearity and thermal protons are crude approximations ) . parameterizing the uncertainty in the details of the turbulence , and using numerical simulations of mhd turbulence as a guide
, qg estimated that the critical @xmath7 above which proton heating dominates ( say @xmath75 ) is @xmath76 , with an uncertainty of @xmath76 !
the physics of particle heating in a collisionless plasma suggests that , in adafs and ion - tori , proton heating may be favored for relatively subthermal magnetic fields , but not for strictly equipartition ones ( @xmath77 ) . in this paper ,
i have focused primarily on heating of particles by small scale , incompressible , alfvnic turbulence . for @xmath78
, it should be quite accurate to neglect sound wave and slow mode excitation . as emphasized by blackman ( 1998 ) ,
however , one expects the importance of compressibility to be an increasing function of decreasing @xmath7 .
for example , at @xmath64 , slow mode excitation can not be neglected ( these modes are damped by ttd on relatively large scales , heating the protons ; e.g. , blackman 1998 ) .
compressibility ( and hence dissipation ) of large scale alfvn waves is also potentially important .
my belief is that , while these effects are undoubtedly present , it is unlikely that the majority of the turbulent energy is dissipated on large scales ; thus , to order of magnitude , the analysis presented in this review is appropriate .
more quantitative estimates are , however , clearly needed .
it is a pleasure to acknowledge numerous discussions with andrei gruzinov and ramesh narayan on the physics of adafs .
this work was supported by an nsf graduate research fellowship and by nsf grant ast 9423209 .
abramowicz , m. et al .
, 1995 , , 438 , l37 achterberg , a. 1981 , a&a , 97 , 259 bisnovatyi - kogan , g. s. , & lovelace , r. v. e. , 1997 , , 486 , l43 blackman , e. 1998 , mnras , in press ( astro - ph/9710137 ) foote , e. a. & kulsrud , r. m. 1979 , apj , 233 , 302 goldreich , p. & sridhar , s. 1995 , , 438 , 763 ( gs ) gruzinov , a. 1998 , , 501 , 787 ichimaru , s. 1977 , , 214 , 840 kato , s. , fukue , j. , mineshige , s. , 1998 , _ black - hole accretion disks _
( japan : kyoto university press ) narayan , r. , mahadevan , r. , & quataert , e. , 1998 , in _ the theory of black hole accretion discs _ , eds . m.a .
abramowicz , g. bjornsson , and j.e .
pringle ( cambridge : cambridge university press ) ( astro - ph/9803131 ) narayan , r. , & yi , i. , 1994 , , 428 , l13 narayan , r. , & yi , i. , 1995 , , 452 , 710 quataert , e. 1998 , , 500 , 978 quataert , e. , & gruzinov , a. , 1998 , apj submitted ( astro - ph/9803112 ) ( qg ) rees , m. j. , begelman , m. c. , blandford , r. d. , & phinney , e. s. , 1982 , nature , 295 , 17 stix , t.h .
1992 , waves in plasmas ( new york : aip ) | i review particle heating by mhd turbulence in collisionless plasmas appropriate to advection - dominated accretion flows .
these considerations suggest that the preferential turbulent heating of protons assumed by theoretical models is only achieved for relatively subthermal magnetic fields . |
an important outstanding issue in the general theory of relativity is the problem of finding an acceptable definition of energy - momentum localization .
h. bondi @xcite has argued that general relativity does not permit a non - localizible form of energy , so in principle , an acceptable definition of local energy - momentum density should exist .
the first attempts to identify such a density by einstein himself @xcite as well as landau and lifshitz @xcite , papapetrou @xcite , and weinberg @xcite , were all defined in terms of non - tensorial , coordinate - dependent pseudotensors .
unfortunately , the physical meaning of the pseudotensors has been and remains a serious point of contention .
( for example , one train of thought suggests that a pseudotensor approach could conflict with the equivalence principle ( @xcite ) ) .
attempts to find alternative definitions of a local energy - momentum density ( by mller @xcite , komar @xcite , penrose @xcite , and others ) , however , have not been substantially more successful ; for example , bergqvist @xcite showed that no two of the many coordinate - independent definitions of mass give the same result for the reissner - nordstrom and kerr spacetimes .
by contrast , aguirregabiria , chamorro , and virbhadra @xcite showed that for all spacetimes of kerr - schild class , the energy - momentum complexes defined by einstein , landau - lifshitz , papapetrou and weinberg all give the same answer . in @xcite virbhadra noticed that the abovementioned energy - momentum complexes coincide for a class of solutions more general than kerr - schild .
for example , rosen and virbhadra @xcite showed that several energy - momentum complexes give the same and a meaningful result for the einstein - rosen metric describing cylindrical gravitational waves ; note that this metric is not of kerr - schild class .
furthermore , chang , nester and chen @xcite , @xcite showed that these energy - momentum complexes can be redefined as the value of a hamiltonian , which has a surface and a boundary term .
the pseudotensors of einstein , landau - lifshitz , papapetrou , weinberg and mller can then be defined in terms of the boundary term of the hamiltonian .
thus the pseudotensors are in fact quasi - local , and under this formulation the energy - momentum complexes no longer conflict with the equivalence principle . as a result
there has been a renaissance of interest in and use of one or more energy - momentum complexes in a wide variety of contexts @xcite . in general the various energy - momentum complexes
give different answers @xcite , @xcite .
chang , nestor and chen @xcite pointed out that the physical significance of the boundary term is still unclear , and hence there are _ a priori _ many choices possible for an appropriate boundary term .
it is therefore natural to ask if any of the various definitions of quasi - local energy - momentum density is preferable to the others . in this paper
we calculate the energy and momentum distributions of a class of regular black holes defined by mars , martn - pratt , and senovilla @xcite using several of the complexes mentioned above . we will show that the einstein energy - momentum complex is one of the complexes that gives a meaningful answer for the energy and momentum distributions , supporting the hypothesis ( @xcite,@xcite ) that the einstein energy - momentum complex is a reasonable definition of energy - momentum density .
we use the convention that latin indices take values from 0 to 3 and greek indices take values from 1 to 3 , and we take units where @xmath0 and @xmath1 . the comma and semicolon ,
respectively , stand for partial and covariant derivatives .
although the singularity theorem of hawking and ellis @xcite forbids the existence of regular black holes with matter content satisfying the strong energy condition everywhere , it is still possible to construct regular black holes satisfying the so called weak energy condition everywhere ( the energy - momentum tensor @xmath2 must satisfy @xmath3 where @xmath4 is any timelike vector at any point of spacetime . )
a famous example of such a regular black hole was given by bardeen @xcite .
( also see sharifi @xcite for the energy computation ) . in this paper
we consider another class of regular black holes constructed by mars , martn - pratt and senovilla @xcite .
consider the following line element : @xmath5 where @xmath6 the functions @xmath7 and @xmath8 of @xmath9 are chosen so as to satisfy the following three constraints : 1 .
the spacetime must represent a schwarzchild black hole , ( i.e. @xmath10 where m is some positive constant which represents the total mass of the black hole , and prime denotes the derivative with respect to @xmath9 ) , 2 .
the model fulfills the weak energy conditions everywhere , and 3 .
the spacetime is regular . in the paper @xcite
the authors give two explicit examples of models that fulfill all these conditions ( hence models do exist ) . in this paper
we choose to leave @xmath7 and @xmath8 general in all of our computations ; the computations are all done for an arbitrary choice of functions @xmath7 and @xmath8 of @xmath9 that satisfy the above criteria . however , for the graphs we will use the following solution from @xcite : @xmath11 where @xmath12 is the adimensional variable @xmath13 , and the corresponding function @xmath8 can be integrated from @xmath14 note that the line element is a modification of the eddington - finkelstein form of a schwarzschild - like spacetime @xmath15 by @xmath16 in the `` @xmath17 '' direction , since the ( unorthodox ) transformation @xmath18 gets rid of this term .
hence , the energy of the interior of a black hole defined by the line element should be modified from the energy of a black hole defined by by the scalar field @xmath16 .
thus from physical considerations we expect the energy of the region enclosed in a 2-sphere to be given by the product @xmath19 of the mass function @xmath20 and the scalar field @xmath16 . in order to calculate the energy and momentum distributions in the next two sections we will need to rewrite in cartesian coordinates . transforming the line element to cartesian coordinates via @xmath21 yields @xmath22 ^ 2\\+e^{2\beta(r)}\big[dt^2-(1-e^{2\beta(r)}\chi(r))[dt+\frac{1}{r}(xdx+ydy+zdz)]^2\big]\end{gathered}\ ] ] note that ( equivalently ) is a static spherically symmetric metric , but is * not * a kerr - schild class metric .
in fact , falls outside the class of metrics for which the energy - momentum complexes of einstein , landau - lifshitz , papapetrou , and weinberg coincide .
we will demonstrate this explicitly in the following two sections .
the einstein energy - momentum complex is @xmath23 where @xmath24_{,m}\end{aligned}\ ] ] @xmath25 and @xmath26 denote energy and momentum density respectively .
the energy and momentum components are given by @xmath27 where we applied gauss s theorem to get the second equality ( @xmath28 and @xmath29 stand for the energy and momentum components respectively , and @xmath30 denotes the outward unit normal to the infinitesimal surface element @xmath31 ) .
using in , it is easy to compute the twelve superpotentials @xmath32 as follows : @xmath33 @xmath34 @xmath35 @xmath36 @xmath37 @xmath38 @xmath39 @xmath40 @xmath41 @xmath42 ( here prime denotes derivative with respect to @xmath9 ) . now applying one gets @xmath43 for the energy component .
note that is the value for @xmath44 we deduced in section 2 .
furthermore note that for @xmath45 , reduces to @xmath46 , the energy expression for a schwarzchild black hole . using and recalling our specific choice of model in section 2 , in figure 1 we plot @xmath47 ( on the @xmath48-axis ) versus @xmath49 .
similarly , the momentum components are : @xmath50 as expected for a static spherically symmetric black hole .
the landau - lifshitz energy - momentum complex is @xmath51 where @xmath52 @xmath53 and @xmath54 denote energy and momentum density respectively .
the energy and momentum components are given by @xmath55 where we applied gauss s theorem to get the second equality ( @xmath28 and @xmath29 stand for the energy and momentum components respectively , and @xmath30 denotes the outward unit normal to the infinitesimal surface element @xmath31 ) .
using in , it is easy to compute the twelve superpotentials @xmath56 as follows : @xmath57 @xmath58 @xmath59 @xmath60 @xmath61 @xmath62 @xmath63 @xmath64 @xmath65 @xmath66 now applying one gets @xmath67 for the energy component , and @xmath68 for the momentum components .
although formula reduces to the correct expression ( @xmath46 ) for the region @xmath69 , the energy component differs from for @xmath70 .
thus falls outside the class of metrics for which the energy - momentum complexes of einstein , landau - lifshitz , papapetrou , and weinberg coincide .
furthermore , negative energy is not physically meaningful for electrically neutral nonrotating massive objects , while the energy component is negative for small @xmath9 .
( see figure 2 , where we plot @xmath47 ( on the @xmath48-axis ) versus @xmath49 . ) therefore , based on physical considerations , appears to be incorrect for @xmath70 , which supports the contention that the einstein energy - momentum complex is a more reasonable computational tool .
the mller energy - momentum complex is @xmath71 where @xmath72 @xmath73 and @xmath74 denote energy and momentum density respectively .
the energy and momentum components are given by @xmath75 where we applied gauss s theorem to get the second equality ( @xmath28 and @xmath29 stand for the energy and momentum components respectively , and @xmath30 denotes the outward unit normal to the infinitesimal surface element @xmath31 ) .
using in , it is easy to compute the twelve superpotentials @xmath76 as follows : @xmath77 @xmath78 @xmath79 @xmath80 @xmath81 @xmath82 @xmath83 @xmath84 @xmath85 @xmath86 now applying one gets @xmath87 for the energy component ( here prime denotes derivative with respect to @xmath9 ) , and @xmath88 for the momentum components .
although again formula reduces to the correct expression ( @xmath46 ) for the region @xmath69 , the energy component differs from for @xmath70 .
( see figure 3 , where we plot @xmath47 ( on the @xmath48-axis ) versus @xmath49 . ) therefore , based on comparison with our expected answer , is less reasonable for @xmath70 , which supports the contention that the einstein energy - momentum complex is a more reasonable computational tool .
the papapetrou energy - momentum complex is @xmath89 where @xmath90 and @xmath91 @xmath92 and @xmath93 denote energy and momentum density respectively .
the energy and momentum components are given by @xmath94 where we applied gauss s theorem to get the second equality ( @xmath28 and @xmath29 stand for the energy and momentum components respectively , and @xmath30 denotes the outward unit normal to the infinitesimal surface element @xmath31 ) .
note that we can apply gauss theorem since the metric ( equivalently ) is time - independent .
using in , it is easy to compute the only three nonzero superpotentials @xmath95 as follows : @xmath96 since @xmath97 now applying one gets the correct expression @xmath98 for the energy component , and @xmath99 for the momentum components , which agree with the expected values .
thus the papapetrou energy - momentum complex also appears to be a reasonable definition for energy - momentum density .
the results of this paper support the cooperstock hypothesis . the cooperstock hypothesis @xcite states that energy is localized to the region where the energy - momentum tensor is non - vanishing .
this hypothesis would imply that there is no energy - momentum contribution from `` vacuum '' regions of space - time .
if true , this hypothesis would have broad implications .
for example , the hypothesis suggests that gravitational waves have no energy and that current attempts to detect these waves using bar detectors are doomed to failure .
in this paper we calculated the energy of a regular black hole of type defined by mars , martn - prats and senovilla ( a `` mmas - class '' black hole ) .
we showed that its energy and momentum could be computed by using the einstein and papapetrou energy - momentum complexes , and that these calculations agree with physical intuition .
furthermore , we showed that the energy and momentum of a mmas - class black hole could not be computed in general using the energy - momentum complex due to landau and lifshitz , that the energy - momentum complex due to mller gives a less reasonable answer , and that these two calculations give results inconsistent with each other and with the energy - momentum complexes of einstein and papapetrou .
the results of this paper support the cooperstock hypothesis .
chang , nestor and chen @xcite stress the importance of identifying appropriate quasi - local energy - momentum criteria in order to identify the right boundary conditions in the definition of energy - momentum density .
the results of this paper suggest that the einstein and papapetrou energy - momentum complexes may define reasonable boundary conditions . however , the situation is more complicated in general .
for example , the einstein energy - momentum complex is not symmetric in its indices and so can not be used to calculate angular momentum . furthermore , whereas both the landau - lifshitz and the papapetrou complexes ( as defined in this paper ) are symmetric in their indices , hence capable of defining conservation laws of angular momentum , the landau - lifshitz complex seems to give more reasonable answers in some cases than the papapetrou complex ( see for example @xcite ) .
additionally , the papapetrou complex seems much less amenable to calculation for time dependent metrics than its competitors . in any case the physical meaning of the einstein boundary term , its apparent dependence on cartesian coordinates , and whether or not the papapetrou boundary term is deserving of equal footing with einstein s , is in urgent need of further study .
i thank v. s. virbhadra for invaluable assistance , and the two anonymous referees for their insightful recommendations .
99 h. bondi , proc .
london * a427 * , 249 ( 1990 ) a. einstein , sitzungsber . preuss . akad
berlin ( math.phys ) 778 ( 1915 ) , addendum-_ibid . _ 799 ( 1915 ) l. d. landau and e. m. lifshitz , _ the classical theory of fields _ , ( addison - wesley press , reading ma ) 317 ( 1951 ) a. papapetrou , proc .
. acad . * a 52 * , 11 ( 1948 ) s. weinberg , _ gravitation and cosmology : principles and applications of general theory of relativity _ , ( wiley , new york ) 165 ( 1972 ) c. w. misner , k. thorne , and j. a. wheeler , _ gravitation _ , ( freeman , san fransisco , 1973 ) c. mller , ann .
( n.y . ) * 4 * , 347 ( 1958 ) a. komar , phys . rev . * 113 * , 934 ( 1959 ) r. penrose , proc .
london * a381 * , 53 ( 1982 ) g. bergqvist , class . quantum grav .
* 9 * , 1753 ( 1992 ) j. m. aguirregabiria , a. chamorro , and k. s. virbhadra , gen .
* 28 * , 1393 ( 1996 ) c. c. chang , j. m. nester , and c. m. chen , arxiv : gr - qc/9912058 v1 15 dec 1999 c. c. chang , j. m. nester , and c. m. chen , phys .
lett . * 83 * , 1897 ( 1999 ) k. s. virbhadra , phys .
d * 60 * , 104041 ( 1999 ) n. rosen , and k. s. virbhadra , gen .
. gravit . * 25 * , 42 ( 1993 ) ; k. s. virbhadra , pramana - j . phys .
* 45 * , 215 ( 1995 ) s.s.xulu , int .
phys * 37 * , 1773 ( 1998 ) ; s.s.xulu,int . j. mod .
d * 7 * , 773 ( 1998 ) ; i.radinschi , acta phys . slov . *
49 * , 789 ( 1999 ) ; i.c.yang , chin . j. phys . * 38 * , 1040 ( 2000 ) ; s.s.xulu,int . j. mod .
a * 15 * , 2979 ( 2000 ) ; s.s.xulu,int .
a * 15 * , 4849 ( 2000 ) ; i.radinschi , mod . phys .
a * 15 * , 803 ( 2000 ) ; s.s.xulu,mod .
a * 15 * , 1511 ( 2000 ) ; i.radinschi , mod . phys .
a * 15 * , 2171 ( 2000 ) ; i.radinschi , mod .
a * 16 * , 673 ( 2001 ) ; m.sharif , int .
a * 17 * , 1159 ( 2002 ) ; i.c .
yang and i.radinschi , mod .
a * 17 * , 1175 ( 2002 ) ; t. bringly , mod .
a * 17 * , 157 ( 2002 ) ; i.c .
yang and i.radinschi , chin .
* 41 * , 326 ( 2003 ) ; s.s.xulu , astrophys .
space sci . * 283 * , 23 ( 2003 ) ; e.c.vagenas , int .
a * 18 * , 5781 ( 2003 ) ; e.c .
vagenas , int . j. mod .
a * 18 * , 5949 ( 2003 ) ; e.c.vagenas , mod .
a * 19 * , 213 ( 2004 ) ; i.c.yang and i.radinschi , chin .
* 42 * , 40 ( 2004 ) ; i. radinschi , arxiv : gr - qc/0103110 ; i. radinschi , arxiv : gr - qc/0302047 ; r.m.gad , arxiv : gr - qc/0306101 ; r.m.gad , arxiv : gr - qc/0307010 ; s.s.xulu , arxiv : hep - th/0308077 ; m.p.dabrowski and j.garecki , arxiv : gr - qc/0309064 ; m.sharif , arxiv : gr - qc/0310018 ; r.m.gad , arxiv : gr - qc/0401039 ; r.m.gad , arxiv : gr - qc/0403097 ; r.m.gad , arxiv : gr - qc/0404108 ; e.c.vagenas , arxiv : gr - qc/0405047 ; i.c.yang and i.radinschi , arxiv : gr - qc/0405118 s. s. xulu , arxiv : hep - th/0308070 v1 11 aug 2003 m. mars , m. m. martn - prats , and j. m. m. senovilla , class .
quantum grav .
* 13 * , l51 ( 1996 ) s. w. hawking , g. f. r. ellis , _ the large scale structure of space - time _ ( cambridge university press , cambridge , england , 1973 ) j. bardeen , in proceedings of gr5 , tiflis , ussr , 1968 ( unpublished ) , also a. borde , phys .
d , * 55 * number 12 , 7615 , ( 1997 ) m. sharif , arxiv : hep - th/0310151 v1 16 oct 2003 f. i. cooperstock , mod .
. lett . * a 14
* , 1531 ( 1999 ) f. i. cooperstock and v. faraoni , class quantum grav * 10 * , 1189 , ( 1993 ) d. bak , d. cangemi , and r. jackw , phys .
d * 42 * , 5173 ( 1994 ) | we compute the energy and momentum of a regular black hole of type defined by mars , martn - prats , and senovilla using the einstein and papapetrou definitions for energy - momentum density .
some other definitions of energy - momentum density are shown to give mutually contradictory and less reasonable results .
results support the cooperstock hypothesis .
[ section ] [ thm]lemma [ thm]corollary [ thm]proposition [ thm]remark [ thm]definition [ thm]example [ thm]conjecture [ thm]proposition - conjecture [ thm]proposition - definition3 [ thm]proposition - definition [ thm]supplementary definition [ thm]definition2 # 1#2 .. |
the _ diffeotopy group _ @xmath4 of a smooth manifold @xmath5 is the quotient of the diffeomorphism group @xmath6 by its normal subgroup @xmath7 of diffeomorphisms isotopic to the identity . alternatively , one may think of the diffeotopy group as the group @xmath8 of path components of @xmath6 , since any continuous path in @xmath6 can be approximated by a smooth one , i.e. an isotopy .
we use this terminology to emphasise that we work in the differentiable category throughout . in the topological realm ,
with diffeomorphisms replaced by homeomorphisms , one speaks of the _ homeotopy group_. in either situation , the more popular term is _ mapping class group _ , sometimes with the attribute ` extended ' in order to indicate that orientation - reversing maps are allowed .
quite a bit is known about the diffeotopy groups of @xmath9-manifolds .
the theorem of cerf @xcite says that @xmath10 .
the diffeotopy groups of lens spaces were computed independently by bonahon @xcite and hodgson
rubinstein @xcite . for other known results , open questions , and an extensive bibliography , see kirby s problem list @xcite , especially problems 3.34 to 3.36 .
the diffeotopy group of @xmath0 was determined by gluck @xcite .
he showed that @xmath11 .
( actually , gluck dealt with the homeotopy group , but in dimension @xmath9 this amounts to the same ; see the discussion in section 5.8 of @xcite . )
our aim in this note is to derive that result by contact topological means .
the main ingredients are the classification of contact structures on @xmath0 up to isotopy , a result of colin about isotopies of @xmath12-spheres in contact @xmath9-manifolds , and a theorem of giroux concerning the space of contact elements on @xmath13 and its contactomorphism group .
this may indicate to what extent one might hope to generalise our method to other @xmath9-manifolds .
the key point in the determination of the diffeotopy group of @xmath0 is to show that any diffeomorphism acting trivially on homology is isotopic to either the identity or a diffeomorphism @xmath14 of order @xmath12 ( up to isotopy ) that will be described in the next section .
this argument will take up sections [ section : contacto ] to [ section : reduction ] .
we then put this result to use in contact topology . in section [ section : space ] we show that the fundamental group of the space of contact structures on @xmath0 , based at the standard tight contact structure @xmath15 , is isomorphic to @xmath2 .
this follows essentially from the observation that the mentioned diffeomorphism @xmath14 is isotopic to a contactomorphism @xmath16 of infinite order in the contactomorphism group ( as was noticed previously by gompf @xcite ) .
in section [ section : knots ] we give an explicit description of an infinite family of homologically trivial legendrian knots in @xmath17 , all of which have the same topological knot type , thurston
bennequin invariant , and rotation number , but which are pairwise not legendrian isotopic .
this family has previously been described by fraser , albeit in an implicit fashion only .
moreover , we shall explain why we regard her argument as incomplete .
given a manifold @xmath5 , write @xmath18 for the group of automorphisms of the homology group @xmath19 .
we consider the homomorphism @xmath20 & \longmapsto & ( f_*|_{h_1 } & , & f_*|_{h_2 } ) .
\end{array}\ ] ] since @xmath21 for @xmath22 , this gives a homomorphism @xmath23 .
for the interpretation of @xmath24 as the automorphism group of @xmath2 it is convenient to write @xmath24 multiplicatively with elements @xmath25 . in order to study the properties of @xmath26 , we introduce the following diffeomorphisms .
write @xmath27 for the rotation of @xmath28 about the @xmath29-axis through an angle @xmath30 .
we think of @xmath31 as @xmath32 .
define diffeomorphisms @xmath33 of @xmath0 by @xmath34 then @xmath35)=(-1,1)$ ] , @xmath36)=(1,-1)$ ] , and @xmath37)=(1,1)$ ] .
so @xmath26 is surjective , and since @xmath38 and @xmath39 commute with each other a splitting of @xmath26 can be defined by sending @xmath40 to @xmath41 $ ] , the element @xmath42 to @xmath43 $ ] , and @xmath44 to @xmath45 $ ] .
we therefore have a split short exact sequence @xmath46 the class @xmath47 $ ] has order @xmath12 in @xmath48 . the fact that the order of @xmath47 $ ] is at most @xmath12 follows from @xmath49 and @xmath50
actually , this shows that @xmath51 is isotopic to the identity via an isotopy preserving the @xmath52-leaves in the product foliation of @xmath0 . in order to show that @xmath14 is not isotopic to the identity
, we choose a trivialisation of the tangent bundle @xmath53 by an oriented frame .
we may assume that along @xmath54 this frame is @xmath55 .
then any orientation - preserving diffeomorphism @xmath56 of @xmath0 induces an element @xmath57\in\pi_1({\mbox{\rm gl}}_3^+)$ ] .
isotopic diffeomorphisms induce the same element .
the identity on @xmath0 induces the trivial element ; the diffeomorphism @xmath14 , the non - trivial element in @xmath58 .
our main goal will be to prove the following statement .
[ prop : kernel ] the subgroup @xmath59 is generated by @xmath47 $ ] , and hence isomorphic to @xmath24 .
in other words , any diffeomorphism of @xmath0 acting trivially on homology is isotopic to either i d or @xmath14 .
the result @xmath60 is an immediate consequence : from the split short exact sequence above we know that @xmath48 is the semidirect product of the normal subgroup @xmath24 and the quotient @xmath61 ; but a normal subgroup of order @xmath12 is central , so the action of the quotient by conjugation is trivial .
thus , let @xmath56 be a diffeomorphism of @xmath0 acting trivially on homology .
( in particular , @xmath56 preserves the orientation . )
the strategy will be to isotope @xmath56 step by step to a diffeomorphism satisfying a number of additional properties , until we arrive at i d or @xmath14 .
after each step , we continue to write @xmath56 for the new diffeomorphism .
we shall rely freely on some fundamental notions and results from contact topology , all of which can be found in @xcite .
the standard tight contact structure @xmath15 on @xmath62 is given by @xmath63 this is the unique positive tight contact structure on @xmath0 up to isotopy , see ( * ? ? ?
4.10.1 ) .
the diffeomorphism @xmath56 is isotopic to a contactomorphism of @xmath15 .
the contact structure @xmath64 , which is again positive and tight , is isotopic to @xmath15 .
gray stability ( * ? ? ?
2.2.2 ) then gives the desired isotopy .
later on we shall need a contactomorphism @xmath16 representing the class @xmath47 $ ] ( there should be no confusion with the notation @xmath27 used earlier ) .
there are two ways of exhibiting such a contactomorphism : the first one uses the above description of @xmath65 ; the second one is better adapted to describing the effect on legendrian curves in the front projection picture . a straightforward computation yields
@xmath66 we claim that @xmath14 can be isotoped to a contactomorphism @xmath16 by an isotopy that shifts each @xmath12-sphere @xmath67 along its characteristic foliation induced by @xmath15 . indeed , that foliation is given by the vector field @xmath68 with singular points at @xmath69 .
one computes @xmath70 this shows that the flow of @xmath71 has the desired effect of decreasing the @xmath72-component relative to the @xmath73- and @xmath74-components ; thus , a suitable rescaling of @xmath71 by a function that depends only on @xmath29 will give us a flow that moves the contact structure @xmath75 back to @xmath76 . an alternative picture , due to gompf @xcite ,
is based on a contactomorphism @xmath77 the @xmath12-spheres @xmath78 correspond to the annuli @xmath79 , each compactified by two points at @xmath80 . now a simple description of a contactomorphism @xmath16 in the class @xmath47 $ ] is given by a dehn twist along a circle @xmath79 in the torus @xmath81 ( plus a shift in the @xmath82-direction to make it a contactomorphism ) : @xmath83 figure [ figure : contact - r ] shows the effect of that dehn twist on the legendrian circle @xmath84 in @xmath85 , followed by a legendrian isotopy corresponding to gompf s ` move 6 ' , of which we shall see more in the next section .
( the figure shows the front projection to the @xmath86-torus . )
@xmath87 at 198 74 @xmath16 at 198 50 , followed by ` move 6'.,title="fig : " ]
as described in ( * ? ? ?
* section 2 ) or ( * ? ? ? * section 11.1 ) , cf .
@xcite , one can represent the contact manifold @xmath65 by the kirby diagram with one @xmath88-handle only ( figure [ figure:1-handle ] ) in the standard contact structure on @xmath89 .
the attaching balls for the @xmath88-handle are drawn as round balls , but it is understood that these balls are in fact chosen in such a way that the characteristic foliation on their boundary induced by the standard contact structure on @xmath89 is the same as the characteristic foliation on @xmath67 induced by @xmath15 . with its standard tight contact structure . ]
the contactomorphism @xmath56 is contact isotopic to a contactomorphism fixing a sphere @xmath90 , which we think of as the boundary of the attaching balls in figure [ figure:1-handle ] .
since @xmath56 is a contactomorphism , @xmath15 induces the same characteristic foliation on @xmath91 as on @xmath92 . as shown by colin @xcite , with @xmath15 being tight this implies that @xmath91 and @xmath92 are contact isotopic and hence @xmath56 contact isotopic to a contactomorphism fixing @xmath92 , provided the two @xmath12-spheres are topologically isotopic . for showing the existence of such a topological isotopy , we essentially rely on stage one of gluck s proof @xcite ;
the argument is included here for the reader s convenience .
if @xmath93 is disjoint from @xmath92 , then those two spheres bound a compact manifold that constitutes an @xmath94-cobordism @xmath95 between them .
this follows from the fact that the two spheres are homotopic ; recall that @xmath56 acts trivially on @xmath96 .
this @xmath94-cobordism @xmath95 is contained inside @xmath0 and therefore does not contain any fake @xmath9-cells without appeal to perelman s positive answer to the poincar conjecture .
it follows that @xmath95 is diffeomorphic to @xmath97 $ ] , and hence @xmath92 isotopic to @xmath98 .
( this argument is due to laudenbach @xcite . ) in the general case , we first use an isotopy to bring @xmath92 and @xmath98 into general position , such that they intersect transversely in a finite number of circles .
we want to isotope @xmath98 further to a sphere disjoint from @xmath92 ; as just explained this will conclude the argument .
let @xmath99 be one of the circles of intersection , chosen in such a way that it bounds a @xmath12-disc @xmath100 in @xmath98 not containing any other circles of intersection . in @xmath92 , the circle
@xmath99 bounds two @xmath12-discs @xmath101 and @xmath102 .
one of the @xmath12-spheres @xmath103 and @xmath104 , say the former , bounds a @xmath9-ball , as can be seen by considering the situation in the universal cover of @xmath0 .
this allows us to isotope @xmath98 across this @xmath9-ball in order to remove the circle @xmath99 of intersection . in the process ,
all circles of intersection contained in @xmath101 will be removed as well .
see figure [ figure : remove ] for a schematic picture .
iterating this procedure , we separate @xmath92 and @xmath98 .
@xmath101 [ bl ] at 170 93 @xmath100 [ bl ] at 228 100 @xmath98 [ br ] at 86 90 @xmath92 [ b ] at 145 145 [ b ] at 248 145 and @xmath98.,title="fig : " ]
now consider the oriented legendrian circle @xmath105 representing what we shall call the _ positive _ generator of @xmath106 , as shown in the front projection picture in figure [ figure:1-handle - k ] .
( it is understood that @xmath107 be equipped with the standard contact structure @xmath108 ; legendrian knots are illustrated in the front projection to the @xmath86-plane . )
this corresponds to the legendrian circle on the left - hand side in figure [ figure : contact - r ] . in particular
, that figure shows that the legendrian knot @xmath109 is legendrian isotopic to the positive stabilisation @xmath110 of @xmath111 . .
] for some @xmath112 , the contactomorphism @xmath113 is contact isotopic to a contactomorphism fixing @xmath111 .
the image @xmath114 will be some legendrian knot representing the positive generator of @xmath106 and , since @xmath56 fixes @xmath92 , going exactly once over the @xmath88-handle . with the help of `
move 6 ' from @xcite , or what is also called the light bulb trick ( cf .
@xcite ) , one can unknot @xmath114 via a legendrian isotopy ( which extends to a contact isotopy by ( * ? ? ?
2.6.2 ) ) .
an example is shown in figure [ figure : lightbulb1 ] , where the final result of the isotopy is actually @xmath111 .
in general , the result will be some ( multiple ) stabilisation of @xmath111 . here is a more ` algorithmic ' description of this unknotting procedure .
first of all , by ( * ? ? ?
2.2 ) we may assume that , after a legendrian isotopy , @xmath114 is in standard form , i.e.its front projection is contained entirely between the two attaching balls for the @xmath88-handle .
given a knotted piece of string with loose ends , one can clearly unknot it by contracting the string from one of its ends . if we imagine the attaching balls as the ends of such a piece of string , this contraction can be regarded as a motion of the right - hand ball , say .
we thus remove all crossings in the front projection , while preserving the cusps ; we need legendrian reidemeister moves of the second kind to slide all the cusps adjacent to the right - hand attaching ball over or under another strand in order to remove the crossing with that strand . with the light bulb trick
this translates into a legendrian isotopy with the attaching balls fixed .
the final result of this legendrian isotopy will be a legendrian knot whose front projection has no crossings , but which now winds several times around the right - hand attaching ball in the @xmath86-plane .
one can bring the knot back into standard form ( and still no crossings in the front projection ) as follows : perform a move of type 6 to introduce a single kink in the front projection ; then remove the kink with a legendrian reidemeister move of the first kind ( see figure [ figure : kink ] ) ; each such move reduces the ( absolute ) winding number of the front projection of @xmath114 around the right - hand ball .
positive and negative stabilisations can then be removed in pairs by a further application of the light bulb trick , as shown in figure [ figure : lightbulb2 ] .
thus , @xmath56 is contact isotopic to a contactomorphism that maps @xmath111 to a stabilisation @xmath115 for some @xmath116 .
then @xmath117 is contact isotopic to a contactomorphism that fixes @xmath111 .
even powers of @xmath16 are isotopic to the identity , but not contact isotopic to the identity .
this follows from the observation that the application of @xmath16 to @xmath111 increases its rotation number by @xmath88 .
notice that @xmath15 is trivial as a @xmath12-plane bundle ; a global non - vanishing section of @xmath15 is given by @xmath118 so the rotation number is well defined for arbitrary legendrian knots in the contact manifold @xmath65 .
we now want to show that after a further contact isotopy we may assume that @xmath113 fixes a whole neighbourhood of @xmath111 . we formulate this as a general statement .
[ lem : nbhd ] let @xmath119 be a legendrian knot in a contact @xmath9-manifold @xmath120 , and let @xmath121 be a contactomorphism of @xmath120 that fixes @xmath119 . then @xmath121 is contact isotopic to a contactomorphism that fixes a neighbourhood of @xmath119 . by the tubular neighbourhood theorem for legendrian knots (
2.5.9 ) , we may identify a tubular neighbourhood @xmath122 of @xmath119 in @xmath120 with a tubular neighbourhood of @xmath123 in @xmath124 with contact structure @xmath125 ; the knot @xmath119 is identified with @xmath123 .
the dilatation @xmath126 is a contactomorphism of @xmath127 for each @xmath128 .
this allows us to assume that the contactomorphic image of @xmath122 in @xmath127 has been chosen so large that when we restrict the contactomorphism induced by @xmath121 to @xmath129 , its image will stay inside this image of @xmath122 ( and that the same holds for the @xmath88-parameter family of contact embeddings considered below ) .
in fact , the argument in the proof of ( * ? ? ? * prop .
3.1 ) can be used to show that we may identify @xmath122 contactomorphically with all of @xmath127 .
although this is not essential , we shall assume it for ease of notation .
thus , we think of ( the restriction of ) @xmath121 as a contact embedding @xmath130 it will suffice to show that @xmath121 is contact isotopic to the inclusion ; the lemma then follows from the isotopy extension theorem for contact isotopies , cf .
* remark 2.6.8 ) .
we now mimic the proof of the contact disc theorem ( * ? ? ?
write @xmath121 in the form @xmath131 the condition for this to be a _ contact _ embedding is @xmath132 with some smooth function @xmath133 .
this can be rewritten as the following system of differential equations : @xmath134 the assumption that @xmath121 fixes @xmath119 translates into @xmath135 now , for @xmath136 $ ] , consider the contact embedding @xmath137 for @xmath138 , this converges to the map @xmath139 from the above system of differential equations we deduce @xmath140 the first differential equation in the above system gives @xmath141 when we evaluate this at @xmath142 , we find with the previous equations : @xmath143 finally , the first differential equation also yields @xmath144 since @xmath145 , we also have @xmath146 .
thus , evaluating the foregoing equation at @xmath142 gives @xmath147 in conclusion , we see that the map @xmath148 takes the form @xmath149 it is easy to check that any map of this form ( with @xmath150 ) is a contact embedding of @xmath129 into @xmath127 .
our initial embedding @xmath121 is thus seen to be contact isotopic to @xmath148 , and the convex linear interpolation between @xmath151 and the constant function @xmath88 defines a contact isotopy between @xmath148 and the inclusion map .
this finishes the proof of the lemma . as explained in (
* example 2.5.11 ) , a universal model for the tubular neighbourhood of a legendrian submanifold @xmath152 in a higher - dimensional contact manifold is provided by a neighbourhood of the zero section @xmath153 in the @xmath88-jet bundle of @xmath152 with its canonical contact structure @xmath154 , where @xmath155 is the canonical @xmath88-form on @xmath156 , written in local coordinates @xmath157 on @xmath152 and dual coordinates @xmath158 on the fibres of @xmath156 as @xmath159 .
the above proof carries over , _ mutatis mutandis _ , to show that the tubular neighbourhood of a legendrian submanifold is unique not only up to contactomorphism , but up to contact isotopy .
here is an alternative proof of lemma [ lem : nbhd ] .
admittedly , the methods used in it amount to cracking nuts with a sledgehammer , but they may be of some independent interest .
we define a new contactomorphism @xmath94 of @xmath120 as follows , cf .
* remark 4.1 ) . choose a standard tubular neighbourhood @xmath122 of @xmath119 , where the contact structure is given by @xmath160 under the identification of @xmath122 with @xmath129 ( and @xmath119 with @xmath123 ) .
observe that @xmath122 may be regarded as the space of ( cooriented ) contact elements of @xmath161 .
set @xmath162 on the closure of @xmath163 . on a smaller tubular neighbourhood @xmath164 , set @xmath165 . by the uniqueness up to contactomorphism of the non - rotative tight contact structure on @xmath166 $ ] with two dividing curves on each boundary component ,
see @xcite , this @xmath94 extends to a contactomorphism on all of @xmath120 .
then @xmath167 is a contactomorphism equal to the identity on @xmath163 .
so the restriction of @xmath167 to @xmath122 may be regarded as a contactomorphism , equal to the identity near the boundary , of the space of contact elements of @xmath161 . according to a result of giroux @xcite ,
the group of those contactomorphisms is connected .
this gives a contact isotopy from @xmath167 to the identity on @xmath5 .
it follows that @xmath121 is contact isotopic to @xmath94 , which has the desired properties .
giroux s paper @xcite has to be read with a certain amount of caution .
proposition 10 and the proofs of the main results ( though not the results as such ) are incorrect .
the proofs can be fixed using the methods of @xcite .
in a final step , we want to appeal once more to the result of giroux @xcite about contactomorphism groups of spaces of contact elements .
the complement of @xmath111 in @xmath65 is contactomorphic to the space of contact elements of @xmath168 . in @xcite
we described an explicit contactomorphism between the space of contact elements of @xmath13 and the complement of a legendrian unknot in @xmath89 with its standard contact structure ( which we shall also write as @xmath15 ) . that complement is seen to be contactomorphic to @xmath169 as follows . an alternative surgery picture for @xmath65 is given by a single contact @xmath170-surgery along a legendrian unknot in @xmath171 .
in this picture , @xmath111 becomes a legendrian push - off of the surgery curve , see @xcite .
the cancellation lemma from @xcite , cf .
6.4.5 ) , says that contact @xmath172-surgery along @xmath111 brings us back to @xmath171 . more specifically ( as the proof of the cancellation lemma shows ) , @xmath111 may be regarded as the belt sphere of the surgery along the legendrian unknot in @xmath171 , and the complement of that belt sphere in the surgered manifold is indeed contactomorphic to the complement of the surgery curve in the initial manifold . in the preceding section
we had found an integer @xmath173 such that ( after a contact isotopy ) @xmath113 fixes a neighbourhood of @xmath111 .
so we may interpret this map as a contactomorphism of the space of contact elements of @xmath161 , equal to the identity near the boundary . by giroux
@xcite , this contactomorphism is contact isotopic ( rel boundary ) to the identity .
thus , in total , our initial diffeomorphism @xmath56 of @xmath0 ( acting trivially on homology ) has been shown to be isotopic to either i d or @xmath14 , as was claimed in proposition [ prop : kernel ] .
\(1 ) the result of giroux about the contactomorphism group of the space of contact elements of @xmath161 uses cerf s theorem @xmath174 in its proof .
so the described methods can not , as yet , be used to give a contact geometric proof of cerf s theorem . however , there is in fact a contact geometric proof of the slightly weaker form of cerf s theorem , saying that every diffeomorphism of @xmath89 extends to a diffeomorphism of the @xmath175-ball a theorem popularly known as @xmath176 .
that proof is due to eliashberg @xcite ; for an exposition see @xcite .
\(2 ) observe that our argument has shown the following : any contactomorphism of @xmath65 acting trivially on homology is contact isotopic to a uniquely determined integer power of @xmath16 ; any contactomorphism that is topologically isotopic to the identity is contact isotopic to an even power of @xmath16 .
gonzalo and the second author have shown in @xcite that there are essential loops in the space of contact structures on torus bundles over the circle . the main ingredient in that proof was the classification of contact structures on the @xmath9-torus .
bourgeois @xcite reproved their result with the help of contact homology and used that technique to detect higher non - trivial homotopy groups of the space of contact structures on a number of higher - dimensional manifolds . here
we formulate such a statement for the fundamental group of the space of contact structures on @xmath0 .
write @xmath177 for the component of the space of contact structures on @xmath0 containing @xmath15 , and @xmath178 for the subgroup of @xmath179 consisting of contactomorphisms of @xmath15 . by gray stability
, we have a surjection @xmath180 with @xmath181 .
as shown in @xcite , this gives rise to a long exact sequence @xmath182 where we write @xmath183 for the inclusion @xmath184 ; this is essentially the homotopy long exact sequence of a serre fibration . by the second remark at the end of the preceding section
, we have @xmath185 , generated by the contact isotopy class of @xmath186 .
since this lies in the kernel of @xmath187 , there must be a subgroup isomorphic to @xmath2 in @xmath188 .
if we permit ourselves to rely on some additional information about the homotopy type of @xmath189 , we can actually show this to be the full fundamental group of @xmath177 .
the component @xmath177 of the space of contact structures on @xmath0 containing @xmath15 has fundamental group isomorphic to @xmath2 .
the homotopy type of the group of homeomorphisms of @xmath0 was determined , modulo the smale conjecture , by csar de s and rourke @xcite .
hatcher s proof @xcite of the smale conjecture not only completes their work , it also implies as shown by cerf @xcite that the space of diffeomorphisms of any @xmath9-manifold is homotopy equivalent to its space of homeomorphisms .
thus , @xmath190 where @xmath191 stands for the component of the contractible loop in the loop space of @xmath192 .
now , @xmath193 , and the generators of @xmath194 and @xmath195 in the above factorisation of @xmath189 can be realised as loops of contactomorphisms @xmath196,\ ] ] and @xmath197,\ ] ] respectively .
thus , the homotopy exact sequence becomes @xmath198 the proposition follows .
in @xcite , fraser described an infinite family of legendrian knots in the contact manifold @xmath199 all of which have the same topological knot type and the same classical invariants @xmath200 and @xmath201 , but which are nonetheless pairwise not legendrian isotopic .
the idea for distinguishing these knots is to perform legendrian surgery on them ( or contact @xmath172-surgery in the language of @xcite ) , and then to observe that the contact structures on the surgered manifold ( which happens to be the @xmath9-torus @xmath202 ) are pairwise not isotopic .
this argument , in our view , is incomplete because it hinges on the statement `` legendrian surgery on legendrian isotopic knots produces isotopic contact structures on the surgered manifold '' which is meaningless , as we want to explain .
suppose you have two legendrian isotopic knots @xmath203 , @xmath204 in a contact @xmath9-manifold @xmath120 .
the legendrian isotopy extends to a contact isotopy @xmath205 , @xmath206 $ ] , of @xmath120 with @xmath207 . for each @xmath206
$ ] , the contactomorphism @xmath205 of @xmath120 induces a contactomorphism between the contact manifold @xmath208 obtained by legendrian surgery along @xmath203 and the contact manifold @xmath209 obtained by legendrian surgery along @xmath210 .
but there is no way , in general , to identify @xmath208 with @xmath209 ( even as mere differential manifolds ) other than with the diffeomorphism induced by @xmath205 .
so we obtain a parametric family of contact manifolds , all of which are contactomorphic , but not an isotopy of contact structures on a fixed differential manifold . in fact , in situations where there is a canonical way of identifying the surgered manifolds ,
the statement in question is false , in general .
this is illustrated by the following example from ( * ? ? ?
* exercise 11.3.12 ( c ) ) , see figure [ figure : twosharks ] .
contact @xmath172-surgery on the ` shark ' in @xmath171 with its mouth on the left or on the right corresponds topologically to a surgery on the unknot with surgery coefficient @xmath211 relative to the surface framing .
if we take the obvious topological identification of the shark with the unknot , this allows us to identify the surgered manifold in both cases with the lens space @xmath212 . with respect to this identification
, the two resulting contact structures on the surgered manifold @xmath212 can be distinguished via their induced spin@xmath213 structure , so they are not isotopic .
( under the identification in question , which gives the two sharks the same , say the counter - clockwise orientation , the shark on the left has @xmath214 , the one on the right , @xmath215 .
this implies that the corresponding spin@xmath213 structures have first chern class @xmath216 ; see ( * ? ? ?
however , there is a legendrian isotopy from one shark to the other ( reversing its orientation ) , and this induces a contactomorphism of the surgered contact manifolds it is simply the contactomorphism induced by the contactomorphism @xmath217 relating the two contact surgery diagrams .
thus , if one wants to show with the help of legendrian surgery that two legendrian knots @xmath219 can not be legendrian isotopic , one has to require , in general , that the surgered contact manifolds are not contactomorphic . in fraser s set - up , unfortunately , the surgered manifolds happen to be contactomorphic by construction .
nonetheless , we now want to show that fraser s idea can be made to work .
in fact , the examples we are going to discuss presently are explicit realisations of the knots described only implicitly by fraser .
figure [ figure : fraser ] shows a family @xmath220 , @xmath112 , of legendrian knots in @xmath221 ; for @xmath222 , the zig - zags are to be interpreted as @xmath223 pairs of zig - zags in the opposite direction .
observe that @xmath224 , where @xmath16 is regarded as a contactomorphism acting only on the upper ( in the picture ) summand @xmath0 there is a realisation of @xmath16 that fixes a disc and hence is compatible with taking the connected sum .
all these knots have the same topological knot type , as can be shown by applying the topological light bulb trick .
moreover , the well - known formul for computing the classical invariants which take the same form for a legendrian knot in ` standard form ' in @xmath221 as in @xmath171 , see @xcite give @xmath225 and @xmath226 for all @xmath112 ( for either orientation of those knots ) .
arguing by contradiction , let us assume that there are two legendrian isotopic knots @xmath230 and @xmath231 , with @xmath232 .
then @xmath233 and @xmath234 , where @xmath235 , will be legendrian isotopic . since @xmath224 , this implies that @xmath236 is contact isotopic to a contactomorphism of @xmath221 that fixes @xmath203 . by lemma [ lem : nbhd ] , @xmath236 is then contact isotopic to a contactomorphism @xmath237 that fixes a neighbourhood @xmath238 of @xmath203 .
the stein fillable and hence tight contact manifold obtained by contact @xmath172-surgery on @xmath203 is @xmath202 , see ( * ? ? ?
* example 11.2.4 ) , with its standard contact structure @xmath239 , cf .
@xcite . interpreted as a kirby diagram , figure [ figure : fraser ] ( with @xmath240 and framing for the handle attachment equal to @xmath218 relative to the contact framing ) describes @xmath241 .
the @xmath161-fibre is represented by the cocore of the @xmath12-handle , so the belt sphere of the surgery on @xmath203 is an @xmath31-fibre of @xmath202 , see ( * ? ? ?
* example 4.6.5 ) .
this @xmath31-fibre corresponds ( up to isotopy ) to the @xmath30-coordinate in the description of @xmath242 , for the @xmath30-circles are uniquely characterised by the fact that they become homotopically trivial in any stein filling @xmath95 of @xmath202 , see the proof of ( * ? ? ?
* lemma 4.3 ) .
that fact rests on two observations .
first of all , the homomorphism @xmath243 induced by inclusion is surjective ; this follows from the cell structure of stein manifolds .
secondly , the @xmath30-fibres must lie in the kernel of this homomorphism , otherwise one could pass to a cover and obtain a stein filling of the contact structure @xmath244 for some @xmath245 , which is impossible by a result of eliashberg @xcite . in the proof of the cancellation lemma given in @xcite it is shown explicitly that the belt sphere of the surgery is legendrian isotopic , in the surgered manifold , to a legendrian push - off of @xmath203 .
alternatively , we can isotope it to a standard legendrian meridian of @xmath203 , as shown in ( * ? ? ?
* prop . 2 ) .
we now want to show that this standard legendrian meridian is in fact _
isotopic to the @xmath30-fibre @xmath246 in @xmath247 . for this
we appeal to the classification of linear legendrian curves in @xmath247 by ghiggini @xcite .
the thurston
bennequin invariant @xmath248 can be defined for such linear legendrian curves @xmath152 as the twisting of the contact structure relative to an incompressible torus containing @xmath152 .
this means that @xmath249 , since the contact structure @xmath242 makes one negative twist along a @xmath30-fibre relative to the framing given by the product structure @xmath250 ( and the orientation @xmath251 induced by @xmath242 ) .
this is the maximal @xmath200 in the topological knot type of @xmath246 ; see ( * ? ? ?
the definition of the rotation number @xmath201 for linear legendrian curves in @xmath247 depends on the choice of trivialisation of @xmath242 , but it can be normalised so that @xmath252 for curves realising the maximal @xmath200 . according to ( * ? ? ?
2.5 ) , the classical invariants suffice to classify legendrian realisations of the topological knot type of @xmath246 .
so all we have to show is that the standard legendrian meridian @xmath253 of @xmath203 has @xmath254 in the surgered manifold , i.e. in @xmath247 .
now , @xmath253 bounds a disc in @xmath89 . with respect to
the framing given by that disc in @xmath89 , the contact structure makes one negative twist along @xmath253 ( since @xmath253 is a standard legendrian unknot in @xmath89 with @xmath254 ) . a close inspection of ( * ? ? ?
* example 4.6.5 ) shows that the framing which @xmath253 inherits from the meridional disc , now regarded as a framing of @xmath253 in @xmath202 , is the same it inherits from an incompressible torus , whence it follows that @xmath255 also in the surgered manifold @xmath247 .
( for that last statement about framings , imagine @xmath89 being cut in a plane passing through the attaching balls of one of the @xmath88-handles , and with the attaching balls for the second @xmath88-handle symmetric to this plane .
then , in the @xmath12-sphere cut out by this plane , a circle around one of the attaching balls that ball being seen as a disc in this @xmath12-sphere defines @xmath253 up to isotopy .
the @xmath12-sphere with two discs removed , together with a cylinder contained in the boundary of the @xmath88-handle , is an incompressible torus in the surgered manifold , containing @xmath253 . ) as a check for consistency , we observe that because of @xmath249 in @xmath247 , contact @xmath170-surgery along such a fibre , which brings us back to @xmath221 by the cancellation lemma , is topologically a surgery with framing given by the product structure of @xmath202 , and that does indeed produce @xmath256 .
if we perform the surgery along @xmath203 inside the neighbourhood @xmath238 , the fact that the belt sphere of the surgery is @xmath246 ( up to legendrian isotopy ) implies that we have a contactomorphism between @xmath257 and @xmath258 for some neighbourhood @xmath259 of @xmath246 .
it follows that the contactomorphism @xmath237 of @xmath221 , which fixes @xmath238 , induces a contactomorphism of @xmath247 that fixes @xmath259 .
this may be interpreted as a contactomorphism of the space of contact elements of @xmath260 with a disc @xmath161 removed , equal to the identity near the boundary . by giroux s
theorem @xcite , this contactomorphism is contact isotopic ( rel boundary ) to one that is lifted from a diffeomorphism of the base @xmath261 .
( recall that the differential of a diffeomorphism of any given manifold induces a contactomorphism of the space of contact elements of that manifold . )
we continue to write @xmath237 for this contactomorphism and its extension to @xmath221 . using the action of the diffeotopy group of @xmath261 by contactomorphisms on @xmath258
, we may assume that the identification of @xmath258 with @xmath257 has been chosen in such a way that one of the standard generators of @xmath262 corresponds to a loop in @xmath263 going once ( homologically , or geometrically counted with sign ) over the upper @xmath88-handle in figure [ figure : fraser ] .
a concrete legendrian realisation @xmath264 of such a loop is shown in figure [ figure:3torus ] , where @xmath246 is taken to be the fibre over @xmath265 .
we take @xmath266 ( oriented by @xmath267 , cooriented by @xmath268 ) ; its legendrian lift @xmath264 coincides with @xmath269 .
transverse to @xmath264 we see an annulus @xmath270 in @xmath271 .
each of the two boundary components of that annulus bounds a disc in @xmath256 , so there we have a @xmath12-sphere transverse to @xmath264 , corresponding to the @xmath52-factor in the upper summand @xmath0 . we also write @xmath264 for the corresponding legendrian loop in @xmath257 .
the contactomorphism @xmath237 of @xmath221 , being isotopic to @xmath236 , sends @xmath264 to a legendrian knot @xmath277 that is smoothly homotopic in @xmath256 to @xmath264 .
this translates into a homotopy in @xmath263 at the price of adding a meridional loop every time the original homotopy crosses @xmath203 . in @xmath278
this becomes a homotopy between @xmath264 and @xmath279 , modulo adding a @xmath30-fibre for each of the meridional crossings .
when projected to @xmath261 , this defines a homotopy between @xmath269 and the projection @xmath280 of @xmath279 . if we write @xmath281 for the diffeomorphism of @xmath261 whose lift is @xmath237 , then @xmath282 , so @xmath280 is a simple closed curve in @xmath261 homotopic to @xmath269 . by baer
s theorem , see @xcite , there is an isotopy of @xmath261 , identical near the boundary , that moves @xmath280 back to @xmath269 ( with the original orientation , for homological reasons ) .
the lift of this isotopy is a contact isotopy of @xmath258 , fixed near the boundary , that moves @xmath279 back to @xmath264 .
thus , @xmath236 is contact isotopic to a contactomorphism of @xmath221 that fixes @xmath264 , which means that @xmath236 sends @xmath264 to a legendrian isotopic copy of @xmath264 , contradicting the fact that @xmath236 changes the rotation number of any oriented legendrian circle that passes once ( in positive direction , passings counted with sign ) over the upper @xmath88-handle by @xmath173 .
prior to fraser s work , no examples were known of legendrian knots that could not be distinguished by the classical invariants .
the first examples of this type in @xmath283 were found by chekanov @xcite , who used legendrian contact homology to distinguish the knots .
various other non - classical invariants have been developed in the meantime , such as normal rulings @xcite or knot floer homology invariants @xcite .
99 , diffotopies des espaces lenticulaires , _ topology _ * 22 * ( 1983 ) , 305314 . , contact homology and homotopy groups of the space of contact structures , _ math .
* 13 * ( 2006 ) , 7185 .
, groupes dautomorphismes et groupes de diffomorphismes des varits compactes de dimension @xmath9 , _ bull .
france _ * 87 * ( 1959 ) , 319329 .
, _ sur les diffomorphismes de la sphre de dimension trois _ @xmath284 , lecture notes in math . * 53 * , springer - verlag , berlin ( 1968 ) . , the homotopy type of homeomorphisms of @xmath9-manifolds , _ bull .
_ * 1 * ( 1979 ) , 251254 .
, differential algebra of legendrian links , _ invent .
_ * 150 * ( 2002 ) , 441483 .
, minimal atlases of closed contact manifolds , proceedings of the yashafest , to appear . , combinatorics of fronts of legendrian links , and arnold s @xmath175-conjectures , _ russian math
. surveys _ * 60 * ( 2005 ) , 95149 . , chirurgies dindice un et isotopies de sphres dans les varits de contact tendues , _ c. r. acad .
paris sr .
_ * 324 * ( 1997 ) , 659663 . , a legendrian surgery presentation of contact @xmath9-manifolds , _ math .
cambridge philos .
soc . _ * 136 * ( 2004 ) , 583598 . , legendrian knots and links classified by classical invariants , _ commun
. contemp .
* 9 * ( 2007 ) , 135162 .
, handle moves in contact surgery diagrams , _
_ * 2 * ( 2009 ) , 105122 . , legendrian helix and cable links , arxiv : math/0611080v2 . , contact @xmath9-manifolds twenty years since j. martinet s work , _ ann .
fourier ( grenoble ) _ * 42 * ( 1992 ) , 165192 . ,
unique holomorphically fillable contact structure on the @xmath9-torus , _ internat .
. notices _ * 1996 * , 7782 .
, example of nonisotopic legendrian curves not distinguished by the invariants @xmath200 and @xmath285 , _ internat .
. notices _ * 1996 * , 923928 . , chekanov eliashberg invariant of legendrian knots : existence of augmentations , _ j. geom
* 47 * ( 2003 ) , 4365 .
, _ an introduction to contact topology _ , cambridge stud .
adv . math . * 109 * , cambridge university press , cambridge ( 2008 ) .
, on the topology of the space of contact structures on torus bundles , _ bull .
london math .
* 36 * ( 2004 ) , 640646 .
, linear legendrian curves in @xmath202 , _ math . proc . cambridge philos . soc . _
* 140 * ( 2006 ) , 451473 .
, sur les transformations de contact au - dessus des surfaces , in _ essays on geometry and related topics _ , eds .
. ghys et al . , vol . 2 , monogr .
enseign . math .
* 38 * ( genve , 2001 ) , 329350 . ,
the embedding of two - spheres in the four - sphere , _ trans .
* 104 * ( 1962 ) , 308333 . , handlebody construction of stein surfaces , _ ann . of math .
( 2 ) _ * 148 * ( 1998 ) , 619693 . , _ @xmath175-manifolds and kirby calculus _ , grad .
stud . math . * 20 * , american mathematical society , providence , ri ( 1999 ) .
, a proof of the smale conjecture , @xmath286 , _ ann . of math .
( 2 ) _ * 117 * ( 1983 ) , 553607 . , involutions and isotopies of lens spaces , in _ knot theory and manifolds ( vancouver , b.c . , 1983 ) _ , lecture notes in math . *
1144 * , springer - verlag , berlin ( 1985 ) , pp .
, on the classification of tight contact structures i , _ geom .
_ * 4 * ( 2000 ) , 309368 ; erratum : factoring nonrotative @xmath287 layers , _ geom .
_ * 5 * ( 2001 ) , 925938 . , the classification of tight contact structures on the @xmath9-torus , _ comm .
geom . _ * 5 * ( 1997 ) , 413438 . , problems in low - dimensional topology , in _ geometric topology ( athens , ga , 1993 ) _ , ams / ip stud .
adv . math .
* 2 * ( part 2 ) , american mathematical society , providence , ri ; international press , cambridge , ma ( 1997 ) , pp .
35473 . , sur les @xmath12-sphres dune varit de dimension @xmath9 , _ ann . of math .
( 2 ) _ * 97 * ( 1973 ) , 5781 . ,
heegaard floer invariants of legendrian knots in contact three - manifolds , arxiv:0802.0628 . ,
_ sur quelques proprits riemanniennes des structures de contact en dimension trois _ , thse de doctorat ( ens lyon , 2008 ) .
, _ surgery on contact @xmath9-manifolds and stein surfaces _ , bolyai soc .
math . stud . * 13 * , springer - verlag , berlin ; jnos bolyai mathematical society , budapest ( 2004 ) .
, legendrian knots , transverse knots and combinatorial floer homology , _ geom .
_ * 12 * ( 2008 ) , 941980 . , _ classical topology and combinatorial group theory _
, 2nd edition , grad .
texts in math . *
72 * , springer - verlag , berlin ( 1993 ) .
, gauge theory and stein fillings of certain @xmath9-manifolds , _ turkish j. math . _
* 26 * ( 2002 ) , 115130 . | as shown by h. gluck in 1962 , the diffeotopy group of @xmath0 is isomorphic to @xmath1 . here
an alternative proof of this result is given , relying on contact topology .
we then discuss two applications to contact topology : ( i ) it is shown that the fundamental group of the space of contact structures on @xmath0 , based at the standard tight contact structure , is isomorphic to @xmath2 ; ( ii ) inspired by previous work of m. fraser , an example is given of an integer family of legendrian knots in @xmath3 ( with its standard tight contact structure ) that can be distinguished with the help of contact surgery , but not by the classical invariants ( topological knot type , thurston
bennequin invariant , and rotation number ) . |
extremely dense strange quark matter with a strangeness fraction of order one and charge neutrality has been suggested to be the absolutely stable form of matter at high densities ( or at least to be metastable because of the weak interaction ) @xcite .
finite chunks of strange quark matter with large strangeness fractions , the so - called strangelets , are predicted to be more stable than normal nuclei @xcite .
recent theoretical and experimental searches for strange quark matter can be found in refs.@xcite on the other hand , metastable strange systems with strangeness fractions of order one and charge neutrality might also exist in the hadronic phase at moderate values of density , between two and three times nuclear matter density .
the properties of metastable exotic multihypernuclear objects with @xmath6 and @xmath7 hyperons reveal quite similar features as the strangelets proposed as a unique signature for quark - gluon plasma formation in heavy ion collisions @xcite .
the equilibrium between the quark and hadronic phases has been studied in @xcite .
strange hadronic matter in bulk has been discussed by glendenning@xcite .
it is found that in an extended mean field theory , a large class of bound multistrange objects formed from combinations of @xmath8 baryons are stable against strong decay .
the presence of filled @xmath6 orbitals blocks the strong decay @xmath9@xcite .
the maximal binding energy per baryon of @xmath10 occurs at a strangeness fraction or strangeness per baryon @xmath11 1.0 - 1.2 , charge per baryon @xmath12 -0.1 - 0.0 and baryon density 2.5 - 3 times that of ordinary nuclei@xcite .
it is comparable to that of hypothetically stable strange quark matter ( `` strangelets '' ) , which has a binding energy per particle @xmath13 -10 to -20 mev .
the predicted phenomenon of metastability of strange hadronic matter and the actual values of the binding energy depend specifically on the partly unknown hyperon potentials assumed in dense matter .
some studies used basically brueckner - hartree - fock ( bhf ) calculations with different nijmegen soft core potentials @xcite . other studies@xcite extend relativistic mean field theory ( rmf ) from ordinary nuclei ( @xmath14 ) to multistrange nuclei with ( @xmath15 ) with the attractive @xmath0 interaction of the nijmegen potential model d @xcite .
recently , stoks and lee @xcite have studied strange hadronic matter using bhf theory and @xmath16 matrices for coupled baryon channels using an @xmath1 extension@xcite of the nijmegen soft core nsc97 potential from the @xmath170,-1 sector into the unexplored @xmath17 - 2,-3,-4 sector .
they have shown that @xmath3 systems are only loosely bound and that charge neutral strangeness - rich hadronic systems are unlikely to exist in nature .
unfortunately their procedure is not self - consistent due to the constraint of equal hyperon fractions that they impose and hence does not give the minimum energy @xcite .
schaffner - bielich and gal @xcite have carried out unconstrained rmf calculations and have found larger binding energies . for small strangeness fractions @xmath18
, strange hadronic matter is mainly composed of @xmath3 and the calculated binding energy closely follows that calculated by using the hyperon potentials of the earlier versions of the soft core nijmegen potential . for larger strangeness fractions @xmath19 ,
the calculated binding energy increases substantially due to a phase transition into @xmath4 dominated matter . in this work
we use the quark - meson coupling model ( qmc ) to study strange hadronic matter .
the qmc model uses the quark degrees of freedom explicitly by coupling the scalar @xmath20 and vector @xmath21 mesonic fields directly to the up and down quarks inside the hadrons which are treated as non - overlapping mit bags .
it was first proposed by guichon@xcite and has been used in various nuclear calculations@xcite .
more recently it has been modified to include a medium - dependent bag parameter coupled to the scalar @xmath20 field@xcite .
this coupling is motivated by invoking the nontopological soliton model for the nucleon@xcite .
quark deconfinement in hot nuclear matter and the phase transition from the hadronic phase to the quark - gluon plasma have been studied in this model@xcite .
this modified quark - meson coupling model ( mqmc)@xcite has been extended to the study of strange hadronic matter by introducing additional scalar @xmath22 and vector @xmath23 mesonic fields which couple only to the strange quark . this extended model has been used to study the properties of hot hypernuclear matter@xcite and neutron star matter@xcite .
we assume strange hadronic matter to consist of the baryon members of the su(3 ) octet and decuplet .
the octet is comprised of the nucleons and the @xmath6 , @xmath2 and @xmath7 hyperons while the decuplet contains the @xmath24 and @xmath25 baryons .
the decuplet baryons are however not found to have a significant contribution in the current calculations at zero temperature and are henceforth dropped from our formalism .
we studied this system with two sets of hyperon - hyperon ( @xmath0 ) potentials , the first set is determined from the nijmegen hard - core potential model d and the second set corresponds to the potentials obtained in a recent su(3 ) extension of the nijmegen soft - core potential model nsc97 .
the differences between these sets are essentially attributable to the extremely attractive @xmath26 and @xmath27 interactions in the second set which allow the possibility of deeply bound nuclear matter with hyperons .
we also consider these two potential models in the context of the qmc model with medium independent bag constant .
the current paper is organized as follows . in section
ii we describe the details of the mqmc model , in section iii we show how we fit the various parameters in the model and in section iv we present and discuss our results .
finally we summarize our conclusions in section v.
in the qmc model , the quark field @xmath28 inside a bag of radius @xmath29 representing a baryon of species @xmath30 satisfies the dirac equation @xmath31\psi_{q}(\vec{r},t)=0 , \label{dirac } \ ] ] where the quark is assumed to couple directly to the scalar and vector meson fields and @xmath32 is the current mass of a quark of flavor @xmath33 .
the current quark masses are taken , for the up and down flavor quarks , to be @xmath34 while for the strange flavor @xmath35 inclusion of small current quark masses for the nonstrange flavors or other values for the strange flavor leads only to small numerical refinements of the present results . in the mean field approximation the meson fields
are treated classically and the space like components of the vector fields vanish for infinite systems due to rotational invariance . as a result @xmath36 and @xmath37 .
the nonstrange ( up and down ) flavor quarks are coupled to the scalar @xmath38 and vector @xmath39 mesons while the strange flavor quarks are coupled to @xmath40 and @xmath41 . for a given value of the bag radius @xmath29 for baryon species @xmath30 and the scalar fields @xmath20 and @xmath42 , the quark momentum @xmath43 is determined by the boundary condition of confinement which , for quarks of flavor @xmath33 in a spherical bag , reduces to @xmath44 , where @xmath45 we have defined the effective quark mass inside the bag as @xmath46 and the effective quark energy is given by @xmath47 the bag energy for baryon species @xmath30 is given by @xmath48 where @xmath49 term is the zero - point energy of the quarks and @xmath50 is the medium - dependent bag parameter . in the simple qmc model
, the bag parameter is taken as @xmath51 corresponding to its value for a free baryon .
the medium effects are taken into account in the mqmc model @xcite by coupling the bag parameter to the scalar meson fields .
in the present work we use the following generalized ansatz for the coupling of the bag parameter to the scalar fields @xmath52 \label{bagcon } \ ] ] with @xmath53 and @xmath54 as additional parameters . here it may be worth mentioning that in ref .
@xcite the bag constant is coupled to the nonstrange @xmath20 scalar field but not to @xmath55 . the spurious center - of - mass energy
is subtracted to obtain the effective baryon mass@xcite @xmath56 where @xmath57 the bag radius @xmath58 for baryon species @xmath30 is obtained through the minimization of the baryon mass with respect to the bag radius@xcite @xmath59 the zero - point energy parameters @xmath60 of eq.(4 ) are used to fit the actual masses of the free baryons @xmath61 and @xmath62 and are found to be @xmath60=2.03 , 1.814 , 1.629 and 1.505 for the @xmath63 and
@xmath7 hyperons respectively , corresponding to a free baryon bag parameter @xmath64 and a free nucleon bag radius @xmath65 .
the total energy density of cold infinite strange hadronic matter at finite baryon density @xmath66 is given by @xmath67 where the summation @xmath30 runs over the 8 members of the baryon octet which reduces for symmetric hypernuclear matter to 4 species with the spin - isospin degeneracy factors @xmath68 and @xmath69 for @xmath70 , @xmath6 , @xmath2 and @xmath7 , respectively and where @xmath71 is the step function at the fermi momentum @xmath72 .
the effective fermi energy of baryon species @xmath30 is given by @xmath73 which is equal to the chemical potential @xmath74 @xmath75 where @xmath76 and @xmath77 are the baryon and strangeness quantum numbers , respectively .
the two independent chemical potentials @xmath78 and @xmath79 are obtained from the conservation of the total baryon density @xmath80 and the total strangeness density @xmath81 the vector mean fields are determined by @xmath82 and @xmath83 where @xmath84 and @xmath85 are the meson - baryon coupling constants defined in eqs . ( 18 ) and ( 19 ) below .
the pressure is the negative of the grand thermodynamic potential density and is given by @xmath86
we assume that the @xmath20 and @xmath21 mesons couple only to the up and down quarks while @xmath42 and @xmath87 couple to the strange quark .
we thus set @xmath88 where @xmath89 refers to the up and down flavors while @xmath90 denotes the strange flavor . by assuming the @xmath91 symmetry of the simple quark model we have the relations @xmath92 and @xmath93 .
the coupling of each baryon species with the vector mesons is calculated by counting the constituent quarks @xmath94 and @xmath95 with these assumptions the only free parameters left at our disposal are the quark - meson coupling constants @xmath96 and @xmath97 and the bag coupling constants @xmath98 for the 4 baryon species and these parameters are adjusted to fit nuclear properties as well as the extrapolated properties of hypernuclear matter .
the coupling constants of the scalar and vector mesons to the nonstrange quarks are taken as @xmath99 and @xmath100 which together with a bag coupling constant @xmath101 yield a binding energy of 16 mev and a compressibility @xmath102 of 289 mev at the normal saturation density @xmath103 of nuclear matter@xcite .
table i summarizes the values used in the current work for the basic quark - meson coupling constants as well as the two sets of the coupling constants @xmath53 and @xmath104 in the bag parameter @xmath105 .
these sets are chosen to fit nuclear and hypernuclear properties .
the parameters @xmath53 are taken to fit the hyperon potentials in nuclear matter : @xmath106 where the hyperon potentials are defined by @xmath107 however , we make two different choices for the constants @xmath104 . in the first set , referred to hereafter as mqmc - i , the medium constants @xmath104 are adjusted so that the potential of a single hyperon embedded in a bath of @xmath7 matter becomes @xmath108 in accordance with the attractive hyperon - hyperon interaction of the nijmegen potential model d@xcite .
furthermore , we adopt the approximation @xmath109 to fit the medium constants .
the resulting @xmath110 is about -20 mev . in mqmc - ii
, we adopt the @xmath0 interactions which occur in the most recent @xmath1 extension of the nijmegen soft - core potential model nsc97@xcite .
the phenomenology in this model departs substantially from that in mqmc - i . in particular , the @xmath26 and @xmath27 interactions are predicted to be highly attractive in some channels , leading to bound states .
we have adjusted the bag constants @xmath54 where @xmath111 to reproduce qualitatively the same binding energy curves of each hyperon species in its own hyperonic matter @xmath112 as those produced by the model nsc97f@xcite .
the resulting binding energy curves are shown in fig.[fig_pot ] .
no binding occurs for @xmath6 hyperons which are already unbound by 8 - 10 mev at the rather low density @xmath113 . on the other hand ,
@xmath2 matter is deeply bound at -33 mev per baryon at @xmath114 which is twice as deep as ordinary nuclear matter , and @xmath7 matter has an energy of -23 mev per baryon at @xmath115 .
furthermore , we have also considered the original ( unmodified ) qmc model where the bag constant is considered to be medium - independent and simply takes its free space value @xmath116 . in analogy with the mqmc calculations we consider two sets of parameters .
the qmc - i set reproduces the @xmath0 potentials corresponding to the nijmegen hard core potential model d as in mqmc - i .
the scalar coupling constants are taken as @xmath117 and @xmath118 while the vector ones are taken as @xmath119=1.471 , 1.868 , 5.148 and 1.662 and @xmath120=0.0 , 2.289 , 0.0 and 3.017 for @xmath121 and @xmath7 , respectively .
the qmc - ii set reproduces the @xmath0 potentials corresponding to the nijmegen soft - core potential nsc97 as in mqmc - ii .
the scalar coupling constants are taken to be the same as in qmc - i while the vector ones are taken as @xmath119=1.471 , 2.942 , 1.103 and 1.721 and @xmath120=0.0 , 4.160 , 1.560 and 2.434 for @xmath121 and @xmath7 , respectively .
the resulting energy per particle curves of each baryon species @xmath122 in its own matter @xmath123 are depicted in fig .
[ fig_pot ] as a function of density for models mqmc - i and ii . for nucleons ,
the curves are the same in both models since they fit the same nuclear matter properties . in mqmc - i ,
the interactions for @xmath6 and @xmath2 are repulsive with @xmath124=+10 mev and @xmath125=+5 mev at @xmath126 for @xmath6 and @xmath2 hyperons respectively .
the @xmath6 hyperon has a shallow local minimum at @xmath127 while @xmath128 has an absolute minimum of @xmath129 per baryon at @xmath130 . on the other hand , for mqmc - ii
, the energy @xmath124 reaches @xmath131 mev already at a density @xmath132 .
this more repulsive potential can be attributed to the very weak underlying @xmath133 interaction in the extended nsc97 model .
in contrast , the @xmath2 hyperon is deeply bound with a minimum of @xmath134 mev per baryon at @xmath135 .
also @xmath7 hyperon matter has a strong attractive potential with a minimum of -23 mev per baryon at @xmath136 .
this shows that mqmc - ii predicts that the @xmath26 and @xmath27 interactions are highly attractive in some channels , leading to bound states .
it is also clear that a mixture of @xmath2 and @xmath7 matter must be very deeply bound unless there is an overwhelmingly repulsive interaction between the @xmath2 and @xmath7 hyperons .
[ fig_mas_a ] and [ fig_mas_b ] display the effective masses of the baryons @xmath137 vs @xmath138 for several strangeness fractions @xmath139 using mqmc - i and ii , respectively . in mqmc - i ,
the effective mass for each baryon species decreases monotonically with density .
it is seen that as @xmath139 increases , @xmath140 increases while @xmath141 and @xmath142 decrease .
the @xmath6 hyperon effective mass however is only weakly sensitive to variations in @xmath139 . in mqmc - ii ,
the situation is rather different , as can be seen from fig .
[ fig_mas_b ] . at first the effective masses of the nucleon and @xmath6 hyperon
decrease smoothly with @xmath138 for low strangeness fractions .
as @xmath139 exceeds 1.4 , these effective masses jump suddenly to higher values as @xmath138 reaches a critical value and then start to decrease again monotonically for higher values of @xmath138 .
the jump or discontinuity in the effective mass of the nucleon can be as high as @xmath143 mev for @xmath144 at a critical density of @xmath145 .
the @xmath6 hyperon has a jump of around 40 mev for @xmath144 at a critical density @xmath146 .
the effective masses of the @xmath2 and @xmath7 hyperons have the opposite behavior with @xmath141 and @xmath142 decreasing when the net strangeness fraction @xmath139 of the system increases .
furthermore they suddenly jump to lower values as @xmath138 exceeds a critical density for @xmath147 .
it is interesting to note that in mqmc - ii the various effective baryon masses become independent of @xmath139 for @xmath148 and @xmath149 .
this behavior indicates that a phase transition takes place in mqmc - ii at the critical densities where the discontinuities in the effective mass occur .
the nature of this transition from the @xmath3 phase to the @xmath4 phase will become apparent in the following figures
. figs .
[ fig_rad_a ] and [ fig_rad_b ] display the density dependence of the bag radius @xmath29 for the @xmath121 and @xmath7 baryons for several strangeness fractions @xmath139 using models mqmc - i and ii , respectively .
the thick solid line indicates the limiting bag radius value @xmath150 where the excluded volume occupies all space and the assumption of nonoverlapping bags definitely breaks down .
it is interesting to note here that the bag radius in mqmc tends to increase with the increasing baryon density in contrary to the ordinary qmc models where the bag radius tends to decrease slowly or saturate at a constant value .
this is because in mqmc the bag parameter decreases with density which allows the bag to expand .
this problem can be cured by introducing quark - quark correlations @xcite which however are not included in the present calculations .
such correlations tend to make the bags shrink at higher densities so that the overlap assumption is not violated and @xmath151 will have values much larger than those given by the thick line in fig .
[ fig_rad_a ] .
finally it is observed that in mqmc - i the radius of the nucleon decreases while @xmath152 and @xmath153 increase with increasing strangeness fractions while @xmath154 is almost unaffected by variations in @xmath139 .
in contrast to mqmc - i where the bag radius increases monotonically with baryon density @xmath138 we notice from fig .
[ fig_rad_b ] that in mqmc - ii this variation has sudden jumps or discontinuities for @xmath155 .
for example , for a strangeness fraction @xmath144 , @xmath156 drops from @xmath157 to @xmath158 when the baryon density reaches @xmath159 .
the shrinking in the nucleon bag size for @xmath160 takes place before the breakdown in the nonoverlap assumption even in the current calculations where the quark - quark correlations are not included .
furthermore , the phase transition tends to occur at lower and lower densities as @xmath139 increases and the bag radius becomes independent of @xmath139 .
the same behavior is found for the radius of the @xmath6 hyperons but the situation is a quite different for the @xmath2 and @xmath7 hyperons .
@xmath161 ( @xmath162 ) both increase with respect to @xmath139 and both jump to a higher value at the critical baryon density when the phase transition takes place for @xmath139 exceeding 1.4 . with @xmath144 , @xmath151
is reached at @xmath163 for @xmath2 and @xmath7 hyperons indicating that , even with the neglect of quark - quark correlations , mqmc - ii is valid up to @xmath163 for such large finite strangeness fractions .
[ fig_bind ] displays the binding energy per baryon for strange hadronic matter vs the baryon density @xmath138 for various strangeness fractions @xmath139 .
it is noted in fig.[fig_bind](a ) that the energy curves for mqmc - i have only one global minimum for every value of the strangeness fraction @xmath139 . at first
, the depth of the minimum increases as @xmath139 increases until it reaches a strangeness fraction of about 1.3 after which the depth of the minimum decreases again for higher values of @xmath139 and approaches that of purely @xmath7 matter for @xmath164 .
it is found that in mqmc - i strange hadronic matter is built out always from a mixture of @xmath165 and @xmath7 baryons .
hence each energy curve has only one global minimum .
[ fig_bind](b ) indicates that the situation for mqmc - ii is drastically different .
for low values of @xmath139 each curve still has a single minimum whose depth increases with the increase in the strangeness fraction
. however , when @xmath139 reaches about @xmath166 , a second minimum appears at a rather high density @xmath167 .
moreover , this second energy minimum quickly deepens and reaches @xmath168 and @xmath169 mev for @xmath170 and 1.5 , respectively .
the appearance of this second minimum can be attributed to the increasingly dominant presence of the @xmath2 hyperons in the system which seems to undergo a first order phase transition from @xmath3 and @xmath171 to @xmath4 strange hadronic matter but at a rather higher baryonic density . for values of the strangeness fraction @xmath139 greater than 1.5 the potential well depth starts to decrease again as the @xmath7 starts to dominate and the energy curves gradually approach that for pure @xmath7 matter for @xmath164 .
[ fig_min ] displays the depth of the energy minimum and the corresponding baryonic density as functions of the strangeness fraction @xmath139 .
it is evident that the depth of the minimum in mqmc - i increases with the strangeness fraction until the curve has a maximum binding energy of @xmath172 at @xmath173 .
the binding energy magnitude decreases again as @xmath139 increases to form pure @xmath7 matter at @xmath174 with a binding energy of 10 mev .
as can be seen from fig.[fig_min](b ) the locations of these minima appear at higher densities as @xmath139 increases until it reaches @xmath173 where it hovers around @xmath175 for @xmath176 . beyond @xmath177 , the location of the energy minimum moves to lower densities .
the situation for mqmc - ii is rather similar to that of mqmc - i for @xmath178 , but it is essentially different for @xmath179 .
when the strangeness fraction reaches @xmath180 , a second minimum in the binding energy appears at @xmath167 .
furthermore , the first minimum disappears at @xmath144 and a transition to the second minimum ( the @xmath4 phase ) occurs due to the reaction @xmath181 .
the upper and lower branches of the mqmc - ii curve in fig.[fig_min](b ) thus refer to two different phases with the discontinuity in the density of the system indicating it is a first order transition .
the binding energy reaches a maximum of 56 mev for @xmath144 and @xmath182 and then decreases to 25.7 mev as @xmath139 increases to @xmath164 where the second minimum disappears since the system is now composed solely of @xmath7 hyperons .
we also display in fig .
[ fig_min ] the results for the qmc - i and qmc - ii models which are seen to be very similar to those found in mqmc - i with a single stable @xmath3 phase .
in particular , unlike mqmc - ii , qmc - ii does not lead to a stable @xmath4 system .
finally it is worth mentioning that , due to the shrinking of the bag radius with increasing baryon density , the nonoverlap assumption survives up to @xmath183
. figs .
[ fig_dens_a ] and [ fig_dens_b ] display the fractional density of each baryon species @xmath184 in strange hadronic matter versus the total baryon density @xmath138 for several values of @xmath139 for models i and ii , respectively .
it is seen that in mqmc - i the @xmath2 hyperon does not appear at all while the @xmath6 has a substantial contribution for low densities @xmath185 .
the @xmath6 contribution , however , drops substantially for higher densities to saturate at a constant value around @xmath186 for @xmath187 and @xmath188 .
it also disappears for @xmath14 ( nuclear matter ) and @xmath174 ( pure @xmath7 matter ) .
the contribution of the nucleons @xmath189 decreases monotonically from one to zero as @xmath139 increases from zero to two . for @xmath19
the nucleons comprise a surprisingly small fraction of the total number of baryons for small densities but their contribution increases rapidly and saturates at higher densities .
the variation of the contribution of the @xmath7 hyperons with density is similar to that of the nucleons but with @xmath190 increasing from zero to one as @xmath139 increases from zero to two .
[ fig_dens_b ] shows however that the situation is significantly different in mqmc - ii , with the @xmath2 hyperons appearing above some critical density and their contribution rapidly saturating at higher densities .
the fractional density @xmath191 takes its maximum value of about @xmath192 at @xmath144 .
the @xmath6 hyperon contribution is large at low baryon densities @xmath193 but practically disappears for higher densities for @xmath194 .
the fractional densities for the @xmath70 s and @xmath7 s in mqmc - ii are similar to those in mqmc - i except for @xmath195 where there is a depletion at high densities caused by the appearance of the @xmath2 hyperons in the system .
the nucleons even disappear completely at high densities for @xmath196
. figs .
[ fig_thres_a ] and [ fig_thres_b ] display @xmath197 vs @xmath138 @xmath198 is the fermi energy and @xmath74 is the chemical potential , eqs .
( 11 ) and ( 12 ) .
if the ratio @xmath199 baryons of species @xmath30 can not exist in the system .
the baryon species first appears with zero fermi momentum @xmath200 when this ratio reaches one and then takes a finite fermi momentum as the ratio exceeds one .
> from fig .
[ fig_thres_a ] it is evident that in mqmc - i the @xmath2 hyperons can not be produced at all in strange hadronic matter .
furthermore , @xmath70 production is increasingly suppressed while @xmath7 production is enhanced as @xmath139 increases .
the situation is quite different in mqmc - ii , fig .
[ fig_thres_b ] , where the @xmath2 hyperons appear in the system for strangeness fractions @xmath201 .
they first appear at @xmath202 for @xmath203 and as @xmath139 increases they appear at lower baryon densities . however , the situation is opposite for @xmath6 which is produced abundantly for low strangeness fractions @xmath204 and then disappears when @xmath139 exceeds @xmath205 .
the contribution of the nucleons decreases significantly for @xmath206 and they disappear completely from the system at @xmath144 when the baryon density exceeds @xmath207 .
furthermore , the nucleons disappear at lower baryon densities as the strangness fraction @xmath139 increases .
it is thus seen that @xmath70 production in mqmc - ii is influenced drastically by the transition from the @xmath3 phase to the @xmath4 phase . on the other hand ,
@xmath7 hyperon production in mqmc - ii is not much different from that in mqmc - i .
finally , fig.[fig_press ] displays the pressure vs the baryon density @xmath138 for models i and ii .
mqmc - i has the normal van der waal - type loop at small densities reflecting the gas to liquid phase transition . in mqmc -
ii the pressure has an additional sudden and significant drop at high densities @xmath208 for strangeness fractions @xmath209 .
this dramatic drop in the pressure is due to the transition from the @xmath3 phase to the @xmath4 phase in strange hadronic matter and probably provides an experimental clue for it .
in the present work , we have studied , within the context of the mqmc model , the properties of strange hadronic matter composed of the su(3 ) octet baryons @xmath121 and @xmath7 using two different models for the hyperon - hyperon interactions . it is interesting to note here that the medium dependent bag constant for each baryon species reflect more realistic self interaction with the scalar mean fields which are corresponding to the nonlinear terms of the scalar mean fields in the normal rmf calculations .
mqmc - i is designed to mimic the consequences of the nijmegen hard - core potential model d @xcite .
it is nevertheless constrained by @xmath6 and @xmath7 nuclear phenomenology , and by a few @xmath133 hypernuclear species reported to date .
it accounts more realistically for the attractive @xmath133 and @xmath171 interactions , but ignores altogether @xmath2 hyperons . on the other hand , mqmc - ii
is designed to generate qualitatively similar baryon potentials to those obtained in the bhf approximation from the su(3 ) extensions of the nijmegen soft - core nsc97 potential models .
the nsc97 model has been tuned up to reproduce certain characteristics of @xmath6 hypernuclei , particularly its version nsc97f .
it yields particularly attractive @xmath27 , @xmath26 and @xmath210 interactions . in mqmc - i
, the @xmath2 hyperons do not appear at all and the @xmath6 hyperons have a substantial contribution at low densities for medium strangeness fractions @xmath188 .
however this contribution drops quickly and saturates for @xmath211 at a constant fraction density slightly larger than @xmath212 .
this model produces smooth binding energy curves over the entire range of @xmath139 and @xmath138 and forms ( meta-)stable strange hadronic matter in the bulk . due to the exceptionally strong attractive interactions among the @xmath2 and @xmath7 hyperons in mqmc - ii , the @xmath3- and @xmath171-dominated strange hadronic matter that exists for @xmath213
is replaced by @xmath4-dominated matter for @xmath214 , with binding energies per baryon reaching as much as 56 mev for @xmath144 .
the equation of state becomes much softer and seems to allow deeply bound hyperonic matter .
the system built out of @xmath4 forms a particularly ( meta-)stable state of strange hadronic matter in the bulk .
when the first order phase transition takes place above some critical baryon density and strangeness fraction , the @xmath2 hyperons appear and their contribution saturates rapidly to a constant fraction density that takes its maximum value of about @xmath192 at @xmath144 .
it is expected that such a phase transition may drastically change the global features of neutron stars@xcite .
our results for mqmc - ii are qualitatively similar to those of schaffner - bielich and gal @xcite where the same @xmath0 potentials of mqmc - ii ( what they call model n ) are employed within rmf theory but with only hadronic degrees of freedom .
there are however some important quantitative differences .
whereas in ref .
@xcite it is found that the second energy minimum , corresponding to @xmath4 dominated matter , becomes more stable than the first minimum for @xmath19 , our current results indicate that this takes place at higher strangeness fractions @xmath214 . furthermore , the binding energy well for the @xmath4 system has a maximum depth of 56 mev in the present work while ref .
@xcite finds the largest depth to be 78 mev . in both calculations
the largest binding for the @xmath4 system takes place at @xmath215 .
the fact that the phase transition takes place at larger strangeness fractions may make it more difficult to observe experimentally , if it occurs at all .
finally , quark - quark correlations and excluded volume effects may become important at high baryon densities in particular when the bags start to overlap at baryon densities @xmath216 .
we shall consider these effects on the properties of strange hadronic matter in our future work .
p. a. m. guichon , phys .
b * 200 * , 235 ( 1988 ) .
p. g. blunden and g. a. miller , phys .
c * 54 * , 359 ( 1996 ) .
k. saito and a. w. thomas , phys .
b * 327 * , 9 ( 1994 ) .
x. jin and b. k. jennings , phys . rev .
c * 54 * , 1427 ( 1996 ) ; phys .
b * 374 * , 13 ( 1996 ) .
x. jin and b. k. jennings , phys . rev .
c * 55 * , 1567 ( 1997 ) ; h. mller and b. k. jennings , nucl . phys . * a 626 * , 966 ( 1997 ) ; h. mller , phys
c * 57 * , 1974 ( 1998 ) .
m. k. banerjee , phys .
c * 45 * , 1357 ( 1992 ) ; v. k. mishra , phys .
c * 46 * , 1143 ( 1992 ) ; e. naar and m. c. birse , j. phys . g. * 19 * , 555 ( 1993 ) i. zakout and h. jaqaman , phys.rev .
c * 59 * 962 ( 1999);_ibid _ * 59 * 968 ( 1999 ) .
i. zakout and h. r. jaqaman , j.phys .
g * 26 * 1095 ( 2000 ) .
i. zakout , h. r. jaqaman , s. pal , h. stcker and w. greiner , phys .
c * 61 * 055208 ( 2000 ) .
s. pal , m. hanauske , i. zakout , h. stcker and w. greiner phys .
c * 60 * 015802 ( 1999 ) .
s. fleck , w. bentz , k. shimizu , and k. yazaki , nucl .
a510 * , 731 ( 1990 ) . | the properties of strange hadronic matter are studied in the context of the modified quark - meson coupling model using two substantially different sets of hyperon - hyperon ( @xmath0 ) interactions .
the first set is based on the nijmegen hard core potential model d with slightly attractive @xmath0 interactions .
the second potential set is based on the recent @xmath1 extension of the nijmegen soft - core potential nsc97 with strongly attractive @xmath0 interactions which may allow for deeply bound hypernuclear matter .
the results show that , for the first potential set , the @xmath2 hyperon does not appear at all in the bulk at any baryon density and for all strangeness fractions .
the binding energy curves of the resulting @xmath3 system vary smoothly with density and the system is stable ( or metastable if we include the weak force ) . however , the situation is drastically changed when using the second set where the @xmath2 hyperons appear in the system at large baryon densities above a critical strangeness fraction .
we find strange hadronic matter undergoes a first order phase transition from a @xmath3 system to a @xmath4 for strangeness fractions @xmath5 and baryonic densities exceeding twice ordinary nuclear matter density .
furthermore , it is found that the system built of @xmath4 is deeply bound .
this phase transition affects significantly the equation of state which becomes much softer and a substantial drop in energy density and pressure are detected as the phase transition takes place .
bethlehem university preprint phys - hep001130 |
since hubble ( @xcite ) , galaxy number counts have been widely used as a statistical tool for probing the distant universe , with the hope of constraining both its geometry and the evolution of its content .
modern , high efficiency instruments and detectors reach impressive magnitude limits in the optical : @xmath4 ( tyson @xcite , metcalfe et al .
@xcite , smail et al .
@xcite ) , although on a limited area . galaxy counts done at brighter magnitudes ( @xmath2 ) are equally useful ; their interpretation is much less model - dependant because of the smaller lookback - times and weaker cosmological effects .
they thus constitute the link between models and deeper counts by providing a normalisation of both space densities and colours at low redshift .
the problem with bright galaxy counts is that they obviously require substantial solid angles to be surveyed in order to provide statistically significant samples . until very recently
, such areas could only be surveyed in a reasonable time using photographic schmidt plates . since the mid - eighties ,
fast microdensitometers coupled with image analysis computer programs have been employed to produce automatically highly complete catalogs from large areas of the high galactic latitude sky like the cosmos ( heydon - dumbleton et al .
@xcite ) , the mrsp ( seitter @xcite ) , or the apm ( maddox et al .
@xcite ) galaxy surveys .
the main difficulty when dealing with photographic material concerns flux measurements .
many scientific issues typically require the systematic errors to be kept @xmath5 mag .. this is quite a difficult task to achieve on large scales , and can be reached only if a large number of _ galaxy _ standards per schmidt plate , over the whole magnitude range of the counts , are observed to calibrate the data ( metcalfe et al .
having such a high density of photometric standards is practically not possible with very large schmidt plate surveys , because of the huge observing time it would require .
the catalogs extracted from these surveys are therefore likely to be hampered by systematic errors in their photometry , rather than by limited statistics .
in this paper we re - examine galaxy number counts in the blue and red photographic passbands over the magnitude range @xmath0 and @xmath6 .
our surveyed area is modest ( 140 sq .
deg . ) , but has been carefully calibrated using a fair density of ccd standards ( @xmath7 , half of them beeing galaxies ) , as an attempt to keep photometric systematic errors @xmath5 mag .
_ over the whole magnitude domain _ of the survey .
the procedure is described in detail in [ par : dataproc ] , as well as the general data processing .
the galaxy number counts are derived and compared to previous studies in [ par : numbercounts ] .
model predictions about number counts and galaxy colour distributions are tested in [ par : modcomp ] .
we finally discuss the implications of our results in [ par : discuss ] .
our optical catalog was originally built as part of a program of identification of faint iras sources on schmidt plates ( bertin , in preparation ) near both ecliptic poles .
the 7 selected fields are characterized by a very low infrared cirrus emission , which ensures that the interstellar extinction is negligible in the visible .
their coordinates are given in table [ tab : fields ] .
to each field corresponds one plate in the blue photographic passband , and one in the red .
northern plates were taken in france at the cerga 0.9 m schmidt telescope , while southern ones are copies from the eso and serc surveys .
[ cols="<,>,<,>,<,>,^,^ " , ] +
the apparent rapid increase in number counts of galaxies in the magnitude range @xmath8 reported by maddox et al .
( @xcite ) is thus absent from our counts ( or those of weir et al .
@xcite ) , although our slope of @xmath9 is still slightly higher than what predict no - evolution models at @xmath10 . as the comparison with the apm data seems to indicate , this might be essentially due to a difference in magnitude scale of @xmath11 0.1 - 0.2 mag./magnitude .
serious suspicions about systematic residual errors in the apm magnitudes at @xmath12 have recently been raised by metcalfe et al .
( @xcite ) . if our data are correct , they support and even strengthen this hypothesis .
other arguments also consolidate this idea . first , our data ( fig .
[ fig : galcol ] ) , as well as previous ones ( see koo & kron @xcite ) , do not present evidence for evolution of galaxy colours over the range @xmath13 . and rule out strong luminosity evolution scenarios in the visible ( mcleod & rieke @xcite ) for bright field galaxies , out to @xmath3 .
one might also think of a `` giant local void '' that would affect the bright end of the counts . although there are indications of large fluctuations in projected galaxy density at the level of a schmidt field ( fig .
[ fig : multicounts ] , picard @xcite ) for @xmath14 , or over the full sky for @xmath15 ( paturel et al .
@xcite ) , the analysis of a large spectroscopic subsample ( loveday et al .
@xcite ) seems to exclude this possiblity in the apm survey .
statistics of ccd calibrations ( maddox et al .
@xcite , loveday et al .
@xcite ) done on the apm magnitudes do not reveal any important trend like the one in fig .
[ fig : apmcompare ] and [ fig : apmstandcomp ] .
this is quite puzzling , and one might wonder if the effects seen here are not purely local and compensated somewhere else in the apm survey . if this were true , correlated systematic magnitude errors of @xmath16 mag .
, over such large areas ( 100 sq
. deg . in our case ) would artificially boost the two - point correlation function on scales @xmath17 in @xmath18 magnitude bins .
this does not appear prominently in fig . 2 of maddox ( @xcite ) .
another clue is that the trend found by metcalfe et al .
( @xcite ) concerns the whole area of the apm catalog , and therefore supports the hypothesis of a global magnitude scale error ( which does not exclude large scale variations of the zero - point ) . to partially bypass uncertainties in the optimum schechter parameters of the local luminosity function for field galaxies
, count models are generally `` renormalized '' by adjusting the schechter density parameter @xmath19 to counts at some bright magnitude . however , the @xmath20 parameter itself is directly affected by any systematic error in the magnitude zero - point of a redshift survey .
such errors are known to exist at bright magnitudes in photographic catalogs ( metcalfe @xcite , yasuda et al .
@xcite ) on which are essentially based all the determinations of the local field luminosity function .
in fact , decreasing the @xmath20 found by lpem by about 0.4 mag . , like our photometric comparison with the apm catalog suggests , brings their luminosity function in good agreement with both the one from eep and our bright galaxy counts ( fig .
[ fig : countmodcomp ] ) , without having to increase @xmath19 .
one should also note that such a``bright '' normalisation in @xmath20 of the luminosity function removes any significant luminosity evolution of luminous galaxies in the blue band to @xmath21 ( lonsdale & chokshi @xcite ) .
we have presented a new photometric survey of bright galaxies with @xmath22 and @xmath23 . comparison with our own ccd galaxy standards , as well as others from the literature , allowed us to estimate systematic errors in the photometry to be @xmath5 mag over the whole magnitude range of the survey .
our data reveal no evidence of strong galaxy evolution as had been reported by maddox et al .
( @xcite ) with the apm survey , although the slope of our number counts is still somewhat higher than what is predicted by no - evolution models brightward @xmath24 .
a comparison with the apm data and other arguments suggest that the apm catalog is affected by a large magnitude scale error , underestimating by about 0.4 mag the flux of galaxies with @xmath25 .
this would then justify the choice of a `` high '' normalization of the field luminosity function ( concerning @xmath20 as well as @xmath19 ) already used by several authors to improve the fit of models at @xmath26 ( e.g. metcalfe et al .
@xcite,@xcite , glazebrook et al .
@xcite ) .
but , given the difficulties inherent to photographic photometry , it would be good to confirm our analysis with linear detectors in order to conclude definitively .
forthcoming digital large scale surveys like the sdss ( e.g. kent @xcite ) should provide this possibility .
we thank the team from cerga , especially d.albanese and c.pollas for providing the northern schmidt plates ; the mama team , especially r. chesnel and o. moreau for their high - quality scans ; p.fouqu for enlightning discussions about the local universe and galaxy photometric systems ; j. loveday and g. dalton for providing the apm comparison subsamples ; and our referee for useful suggestions . | we present bright galaxy number counts in the blue ( @xmath0 ) and red ( @xmath1 ) passbands , performed over 145 sq .
degrees , both in the northern and southern galactic hemispheres .
the work was conducted on schmidt plates digitized with the mama machine , and individually calibrated using an adequate number of ccd sequences .
we find a relatively large density of bright galaxies implying a `` high '' normalization of the local luminosity function .
our counts and colour distributions exhibit no large departure from what standard no - evolution models predict to magnitudes @xmath2 , removing the need for evolution of the non - dwarf galaxy population in the optical , out to @xmath3 .
this result disagrees with that of maddox et al .
( 1990 ) on the apm catalog .
we show that the apm and similar catalogs may be affected by a systematic magnitude scale error which would explain this discrepancy . |
the widely used fourier analysis method in temporal analysis is to derive frequency spectra from a time series . after the fourier transform of a light curve @xmath1 @xmath2
we can get the power density spectrum @xmath3 from two light curves , @xmath4 and @xmath5 , observed simultaneously in two energy bands at times @xmath6 , and their fourier transforms @xmath7 and @xmath8 , we can construct the cross spectrum @xmath9 and then the time lag spectrum @xmath10/2\pi f_j\ ] ] and the coherence coefficient spectrum @xmath11 people usually take a fourier period @xmath12 as a timescale and use the fourier power spectrum to describe the distribution of variation amplitude at different timescales , the time lag spectrum to describe the distribution of the emission time difference between two energy bands at different timescales and the coherence coefficient spectrum to describe the distribution of degree of linear correlation between high and low energy processes at different timescales .
but it is not correct to equate the fourier period with the timescale ; a fourier component with a certain frequency @xmath13 of a light curve is not equal to the real process with the timescale @xmath12 .
for some important high - energy emission processes the fourier spectra distort the timescale distribution of real physical processes seriously .
it is needed to derive timescale spectra from observed light curves directly in the time domain without through fourier transforms .
the power density spectrum can be derived directly in the time domain ( li 2001 ) . for a counting series
@xmath14 obtained from a time history of observed photons with a time step @xmath15 , the definition of variation power is @xmath16 where @xmath17 .
the power density @xmath18 at a timescale @xmath15 can be then derived @xmath19 from ( 6 ) , ( 7 ) we can calculate the power density for poison noise @xmath20 and the signal power density can be defined as @xmath21 in fig .
1 the bottom panel shows the expected power spectrum ( solid line ) for a signal consisting of stochastic shots ( shown in the top panel ) , the timescale spectrum of power density ( crosses ) and fourier spectrum ( dashed line ) from the simulated data ( shown in the middle panel ) of the stochastic signal . as the rise and decay time constant of a shot
is randomly taken from the range between 5 ms and 0.2 s , there should exist significant variation power in this timescale region .
one can see from fig .
1 that the fourier spectrum significantly underestimates the power densities at the shorter timescales .
timescale spectra of power density can help to reveal the nature of physical process around compact stars ( li & muraki 2002 ) .
figure 2 shows the timescale spectra and fourier spectra of power density for a sample of accreting black holes ( left side ) and neutron stars ( right side ) .
the five black hole candidates demonstrate a significant power excess in the timescale spectra in comparison with their corresponding fourier spectra at timescales shorter than @xmath22 s , but the two kinds of power spectrum in the studied neutron stars are generally consistent with each other . assuming that there exist a stochastic process with a characteristic time of @xmath22 s in the black hole systems and that a significant variability of the accreting neutron stars comes from stochastic processes with characteristic time constants much shorter than @xmath23 ms can explain the observed power spectra for x - ray binaries .
measuring the relative time delay between photons in two energy bands can give useful clue to understanding the emission mechanism and emitting region ( e.g. , kazanas , hua & titarchuk 1997 ; hua , kazanas & titarchuk 1997 ; nowak et al .
1999 ) . with the cross - correlation analysis
one can calculate a time lag @xmath24 from two light curves .
traditional cross - correlation method fails to calculate any time lag shorter than the time step @xmath15 . in studying a complex process
only a time lag @xmath24 is not enough , we need to know time lags at different timescales , i.e. the timescale spectrum @xmath25 . using fourier analysis
we can get a time lag spectrum .
but as the fourier technique is powerless for detecting the variation power of a stochastic process at high frequencies ( short timescales ) range , the fourier analysis method is also powerless for detecting the time lags at high frequencies ( short timescales ) .
the timescale spectrum of time lag can be derived with a modified cross - correlation technique ( li , feng & chen 1999 ; li 2001 ) . for two light curves
@xmath26 with a time step @xmath15 the modified cross - correlation function at @xmath27 is defined as @xmath28 where @xmath29 , @xmath30 , @xmath31 is the counts in the time interval @xmath32 .
find the time lag @xmath24 letting @xmath33 for different timescale @xmath15 , we can derive the timescale spectrum of time lag , @xmath34 . for comparing the two kinds of spectrum for time lags , timescale spectrum and fourier spectrum , we construct two simulated time event series which consist of two kinds of random shot component : rapid components with typical timescale 1 ms and time lag 5 ms and slow component with timescale 0.01 s and time lag
0.02 s. figure 3 shows that the timescale spectrum can represent the time delay distribution in the real process , but the fourier analysis is powerless to detecting the time lags in the short timescale range .
the timescale spectral method for time lag analysis is a powerful tool in revealing the characteristic of emission process . applying the time domain technique , qu et al .
( 2001 ) detected time lags of hard x - ray photons from cyg x-1 in the short timescale region ( @xmath35 s or @xmath36 hz ) with pca / rxte data for the first time , giving strong constraint on emission mechanism ; feng et al .
( 2002 ) revealed the time delays of soft photons in terrestrial gamma - ray flashes ( tgfs ) observed by batse / cgro , strongly supporting the discharging mechanism of tgf production .
investigating temporal properties of grbs by the timescale spectral method is in process .
figure 4 shows the spectrum of time lag of soft photons and the energy dependence of time lag for grb910503 detected by batse .
the complex variability of high - energy emission shown in different time scales is a common character for x - ray binaries , super massive black holes and @xmath0-ray bursts .
the variability is caused by various physical processes at different timescales . for understanding the emission process of high - energy photons
it is necessary to know the variation characteristics in different timescales quantitatively , i.e. to derive timescale spectra from observed light curves .
in addition to the methods of making spectral analysis for power density and time lag in the time domain introduced in last two sections , the algorithms to calculate timescale spectra for coherence , hardness , variability duration , and correlation coefficient between two characteristic quantities have been also proposed ( li 2001 ) . there
now exist two kinds of spectral analysis : frequency analysis and timescale analysis .
as any observable physical process always occurs in the time domain , a frequency spectrum obtained by frequency analysis needs to be interpreted in the time domain . but a frequency analysis is based on a certain kind of time - frequency transformation .
different mathematically equivalent representations with different bases or functional coordinates in the frequency domain exist for certain time series , a fourier spectrum with the trigonometric basis does not necessarily represent the true distribution of a physical process in the time domain .
the rms variation vs. timescale of a time - varying process may differ substantially from its fourier spectrum .
figure 1 shows that the fourier transform distorts the power density distribution of a stochastic process at short timescales seriously .
the timescale analysis performed directly in the time domain can derive real timescale distribution for quantities characterizing temporal property . in comparison with the frequency analysis ,
timescale spectra from the timescale analysis can more exactly represent timescale distributions of a real physical process , and more sensitively reveal temporal characteristics at short timescales for a stochastic process . until now fourier analyses of variabilities of x - ray binaries observed by various instruments are all failure to detect hard x - ray lags at the high frequency region of @xmath37 hz .
our simulation study shows it is an intrinsic weakness of the fourier method , that more sensitive x - ray detectors of next generation can still not observe high frequency lags with fourier technique . on the other hand ,
the timescale analysis can already derive time lag spectra at short timescales reliably from existing data
. 1 li t.p .
( 2001 ) _ chin
. j. astron .
_ , 1 , 313 li t.p . , feng y.x .
& chen l. ( 1999 ) _ apj _ , 521 , 789 li t.p . & muraki y. ( 2002 ) _ apj _ , 578 , 374 feng h. , li t.p .
, wu m. , zha m. & zhu q.q .
( 2002 ) @xmath38 , 29 , no.3 hua x.m .
, kazanas d. & titarchuk l. ( 1997 ) @xmath39 , 482 , l57 kazanas d. , hua x.m . & titarchuk l. ( 1997 ) @xmath39 , 480 , 735 nowak m.a . , wilms j. , vaughan b.a . , dove j.b . & begelman c. ( 1999 ) @xmath39 , 510 , 874 qu j.l . & li t.p .
( 2001 ) _ acta astron . sin .
_ , 42 , 140 | a technique of timescale analysis performed directly in the time domain has been developed recently .
we have applied the technique to study rapid variabilities of hard x - rays from neutron star and black hole binaries , @xmath0-ray bursts and terrestrial @xmath0-ray flashes .
the results indicate that the time domain method of spectral analysis is a powerful tool in revealing the underlying physics in high - energy processes in objects .
methods : data analysis x - ray binaries @xmath0-ray bursts |
in dynamic asset pricing models , stochastic discount factors ( sdfs ) are stochastic processes that assign prices to claims to future payoffs over different investment horizons .
@xcite , @xcite and @xcite show that sdf processes may be decomposed into permanent and transitory components .
the _ permanent component _ is a martingale that induces an alternative probability measure which is used to characterize pricing over long investment horizons .
the _ transitory component _ is related to the return on a discount bond of ( asymptotically ) long maturity . @xcite and
the subsequent literature on bounds has found that sdfs must have nontrivial permanent and transitory components in order to explain a number of salient features of historical returns data .
@xcite show that the permanent - transitory decomposition obtains even in very general semimartingale environments , suggesting that the decomposition is a fundamental feature of arbitrage - free asset pricing models .
this paper introduces econometric methods to extract the permanent and transitory components of the sdf process .
specifically , we show how to estimate the solution to the perron - frobenius eigenfunction problem introduced by @xcite from a time series of data on state variables and the sdf process . by estimating directly the eigenvalue and eigenfunction
, one can reconstruct empirically the time series of the permanent and transitory components and investigate their properties ( e.g. the size of the components , their correlation , etc ) .
the methodology also allows one to estimate both the yield and the change of measure which characterizes pricing over long investment horizons .
this approach is fundamentally different from existing empirical methods for studying the permanent - transitory decomposition , which produce bounds on various moments of the permanent and transitory components as functions of asset returns @xcite .
the pathbreaking work of @xcite shows that the sdf decomposition may be obtained analytically in markovian environments by solving a perron - frobenius eigenfunction problem .
the permanent and transitory components are formed from the sdf , the eigenvalue , and its eigenfunction .
the eigenvalue is a long - run discount factor that determines the average yield on long - horizon payoffs .
the eigenfunction captures dependence of the price of long - horizon payoffs on the state .
the probability measure that characterizes pricing over long investment horizons may be expressed in terms of the eigenfunction and another eigenfunction that is obtained from a time - reversed perron - frobenius problem .
see @xcite , @xcite , @xcite , and @xcite for related theoretical developments .
the methodology introduced in this paper complements the existing theoretical literature by providing an _ empirical _
framework for estimating the perron - frobenius eigenvalue and eigenfunctions from a time series of data on the markov state and the sdf process .
empirical versions of the permanent and transitory components can then be recovered from the estimated eigenvalue and eigenfunction .
the methodology is _ nonparametric _ ,
i.e. , it does not place any tight parametric restrictions on the law of motion of state variables or the joint distribution of the state variables and the sdf .
this approach is coherent with the existing literature on bounds on the permanent and transitory components , which are derived without placing any parametric restrictions on the joint distribution of the sdf , its permanent and transitory components , and asset returns .
this approach is also in line with conventional moment - based estimators for asset pricing models , such as gmm @xcite and its various extensions .
examples include conditional moment based estimation methodology of @xcite which has been applied to estimate asset pricing models featuring habits @xcite and recursive preferences @xcite and the extended method of moments methodology of @xcite which is particularly relevant for derivative pricing . in structural macro - finance models , sdf processes ( and their permanent and transitory components )
are determined by both the preferences of economic agents and the dynamics of state variables .
several works have shown that standard preference and state specifications can struggle to explain salient features of historical returns data .
@xcite find that certain specifications appear unable to generate a sdf whose permanent component is large enough to explain historical return premia without also generating unrealistically large spreads between long- and short - term yields .
@xcite find that historical returns data support positive covariance between the permanent and transitory components , but that this positive association can not be replicated by workhorse models such as the long - run risks model @xcite .
the role of dynamics can be subtle , especially with recursive preferences that feature forward - looking components .
the nonparametric methodology introduced in this paper may be used together with parametric methods to better understand the roles of dynamics and preferences in building models whose permanent and transitory components have empirically realistic properties .
of course , if state dynamics are treated nonparametrically then certain forward - looking components , such as the continuation value function under @xcite recursive preferences , are not available analytically .
we therefore introduce nonparametric estimators of the continuation value function in models with @xcite recursive preferences with elasticity of intertemporal substitution ( eis ) equal to unity ..
our approach is different from theirs for a number of reasons discussed in section [ s : emp ] .
further , we study formally the large - sample properties of our estimators whereas @xcite do not . ]
this class of preferences is used in prominent empirical work , such as @xcite , and may also be interpreted as risk - sensitive preferences as formulated by @xcite ( see @xcite ) .
we reinterpret the fixed - point problem solved by the value function as a _ nonlinear _ perron - frobenius problem . in so doing ,
we draw connections with the literature on nonlinear perron - frobenius theory following @xcite .
the methodology is applied to study an environment similar to that in @xcite .
we assume a representative agent with @xcite preferences with unit elasticity of intertemporal substitution . however , instead of modeling consumption and earnings using a homoskedastic gaussian var as in @xcite , we model consumption growth and earnings growth as a general ( nonlinear ) markov process .
we recover the time series of the sdf process and its permanent and transitory components without assuming any parametric law of motion for the state .
the permanent component is large enough to explain historical returns on equities relative to long - term bonds , strongly countercyclical , and highly correlated with the sdf .
we also show that the permanent component induces a probability measure that tilts the historical distribution of consumption and earnings growth towards regions of low earnings and consumption growth and away from regions of high consumption growth . to understand better the role of dynamics
, we estimate the permanent and transitory components corresponding to different calibrations of preference parameters .
we find that the permanent and transitory components can be positively correlated for high ( but not unreasonable ) values of risk aversion .
in contrast , we find that parametric linear - gaussian and linear models with stochastic volatility fitted to the same data have permanent and transitory components that are negatively associated .
these findings suggest that nonlinear dynamics may have a useful role to play in explaining the long end of the term structure .
although the paper is presented in the context of sdf decomposition , the methodology can be applied to study more general processes such as the valuation and stochastic growth processes in @xcite , @xcite , and @xcite .
the sieve approach that we use for estimation builds on earlier work on nonparametric estimation of markov diffusions by @xcite and @xcite .
this approach approximates the infinite - dimensional eigenfunction problem by a low - dimensional matrix eigenvector problem whose solution is trivial to compute .
( @xmath1 is the sample size ) whereas with sieves the dimension is @xmath2 . ]
this approach also sidesteps nonparametric estimation of the transition density of the state .
the main theoretical contributions of the paper may be summarized as follows .
first , we study formally the large - sample properties of the estimators of the perron - frobenius eigenvalue and eigenfunctions .
we show that the estimators are consistent , establish convergence rates for the function estimators , and establish asymptotic normality of the eigenvalue estimator and estimators of other functionals .
these large - sample properties are established in a manner that is sufficiently general that it can allow for the sdf process to be either of a known functional form or containing components that are first estimated from data ( such as preference parameters and continuation value functions ) .
although the analysis is confined to models in which the state vector is observable , the main theoretical result applies equally to models in which components of the state are latent .
second , semiparametric efficiency bounds for the eigenvalue and related functionals are also derived for the case in which the sdf is of a known functional form and the estimators are shown to attain their bounds .
third , we establish consistency and convergence rates for sieve estimators of the continuation value function for a class of models with recursive preferences .
the derivation of the large sample properties of the eigenfunction / value estimators is nonstandard as the eigenfunction / value are defined implicitly by an unknown , nonselfadjoint operator .
the literature on nonparametric eigenfunction estimation to date has focused almost exclusively on the selfadjoint case ( see @xcite and @xcite for sieve estimation and @xcite , @xcite , and @xcite for a kernel approach ) .
the exception is @xcite and @xcite who establish asymptotic normality of kernel - based eigenfunction and eigenvalue estimators for nonparametric euler equation models .
the derivation of convergence rates for the time - reversed eigenfunction , semiparametric efficiency bounds , and asymptotic normality of the eigenvalue estimator ( and estimators of related functionals ) are all new .
the remainder of the paper is as follows .
section [ s : setup ] briefly reviews the theoretical framework in @xcite and related literature and discusses both the scope of the analysis and identification issues .
section [ s : est ] introduces the estimators of the eigenvalue , eigenfunctions , and related functionals and establishes their large - sample properties .
nonparametric continuation value function estimation is studied in section [ s : recursive ] .
section [ s : mc ] presents a simulation exercise which illustrates favorable finite - sample performance of the estimators .
section [ s : emp ] presents the empirical application and section [ s : conc ] concludes .
an appendix contains additional results on estimation , identification , and all proofs .
this subsection summarizes the theoretical framework in @xcite , @xcite ( hs hereafter ) , @xcite , and @xcite ( bhs hereafter ) .
we work in discrete time with @xmath3 denoting the set of non - negative integers . in arbitrage - free environments ,
there is a positive _ stochastic discount factor _
process @xmath4 that satisfies : @xmath5 = 1\ ] ] where @xmath6 is the ( gross ) return on a traded asset over the period from @xmath7 to @xmath8 , @xmath9 denotes the information available to all investors at date @xmath7 , and @xmath10 $ ] denotes expectation with respect to investors beliefs ( see , e.g. , @xcite ) . throughout this paper , we impose rational expectations by assuming that investors beliefs agree with the data - generating probability measure .
@xcite introduce the _ permanent - transitory decomposition _ : @xmath11 the permanent component @xmath12 is a martingale : @xmath13 = 1 $ ] ( almost surely ) .
hs show that the martingale induces an alternative probability measure which is used to characterize pricing over long investment horizons .
the transitory component @xmath14 is the reciprocal of the return to holding a discount bond of ( asymptotically ) long maturity from date @xmath7 to date @xmath15 .
@xcite provide conditions under which the permanent and transitory components exist .
@xcite show that the decomposition obtains in very general semimartingale environments . to formally introduce the framework in hs and bhs , consider a probability space @xmath16 on which there is a time homogeneous , strictly stationary and ergodic markov process @xmath17 taking values in @xmath18 .
let @xmath19 be the filtration generated by the histories of @xmath20 .
when we consider payoffs that depend only on future values of the state and allow trading at intermediate dates , we may assume the sdf process is a _ positive multiplicative functional _ of @xmath20 .
that is , @xmath21 is adapted to @xmath22 , @xmath23 for each @xmath24 ( almost surely ) and : @xmath25 where @xmath26 is the time - shift operator given by @xmath27 for each @xmath28 .
is a function of @xmath29 and @xmath30 is a function of @xmath31 .
the methodology applies to models in which the increments of the process @xmath32 are functionals of both @xmath20 and an additional noise process as in bhs .
we maintain this simpler presentation for notational convenience .
] define the operators @xmath33 which assign date-@xmath7 prices to date-@xmath34 payoffs @xmath35 by : @xmath36\,.\ ] ] it follows from the time - homogeneous markov structure of @xmath20 and the multiplicative functional property of @xmath32 that @xmath37 for each @xmath38 , where : @xmath39 \label{e : m : def } \\
\frac{m_{t+1}}{m_t } & = m(x_t , x_{t+1 } ) \label{e : sdf : m}\end{aligned}\ ] ] for some positive function @xmath40 . for convenience
, we occasionally refer to @xmath40 as the sdf .
hs introduce and study the perron - frobenius eigenfunction problem : @xmath41 where the eigenvalue @xmath42 is a positive scalar and the eigenfunction @xmath43 is positive .. ] we are also interested in the time - reversed perron - frobenius problem : @xmath44 where @xmath45 $ ] , @xmath42 is the eigenvalue from ( [ e : pev ] ) , and the eigenfunction @xmath46 is positive .
the permanent and transitory components are constructed using @xmath42 and @xmath43 .
the function @xmath46 will play a role in characterizing the long - run pricing approximation and the asymptotic variance of estimators of @xmath42 .
clearly @xmath42 , @xmath43 and @xmath46 are determined by both the form of the sdf process and the dynamics of the state process .
we will assume that the function @xmath40 is either pre - specified by the researcher or has been estimated from data on asset returns .
however , we will not restrict the dynamics of @xmath20 to be of any parametric form . given @xmath42 and @xmath43 which solve the perron - frobenius problem ( [ e : pev ] ) , we may define : @xmath47 it follows from ( [ e : pev ] ) that @xmath48 = 1 $ ] ( almost surely ) for each @xmath28 .
hs show that there may exist multiple solutions to ( [ e : pev ] ) , but only one solution leads to processes @xmath49 and @xmath50 that may be interpreted correctly as permanent and transitory components .
such a solution has a martingale term that induces a change of measure under which @xmath20 is _ stochastically stable _ ; see condition 4.1 in bhs for sufficient conditions
. loosely speaking , stochastic stability requires that conditional expectations under the distorted probability measure converge ( as the horizon increases ) to an unconditional expectation @xmath51 $ ]
. the expectation @xmath51 $ ] will typically be different from the expectation @xmath52 $ ] associated with the stationary distribution of @xmath20 . under stochastic stability ,
the one - factor representation : @xmath53 \phi(x)\ ] ] holds for each @xmath54 for which @xmath55 $ ] is finite .
when a result like ( [ e : lrr ] ) holds , we may interpret @xmath56 and @xmath14 from ( [ e : pctc ] ) as the permanent and transitory components .
we may also interpret @xmath57 as the _ long - run yield_. further , the result shows that @xmath43 captures state dependence of long - horizon asset prices .
the theoretical framework of hs may be used to characterize properties of the permanent and transitory components analytically by solving the perron - frobenius eigenfunction problem .
below , we describe an empirical framework to estimate the eigenvalue and eigenfunctions from time series data on @xmath20 and the sdf process . the markov state vector @xmath58 is assumed throughout to be fully observable to the econometrician .
however , we do not constrain the transition law of @xmath20 to be of any parametric form .
appendix [ ax : filter ] discusses potential extensions of our methodology to models in which @xmath58 or a subvector of @xmath58 is unobserved by the econometrician .
the main theoretical results on consistency and convergence rates ( theorem [ t : rate ] and [ t : fpest ] ) apply equally to such cases .
we assume the sdf function @xmath40 is either observable or known up to some parameter which is first estimated from data on @xmath20 ( and possibly asset returns ) .
[ [ case-1-sdf - is - observable ] ] case 1 : sdf is observable + + + + + + + + + + + + + + + + + + + + + + + + + here the functional form of @xmath40 is known ex ante .
for example , consider the ccapm with time discount parameter @xmath59 and risk aversion parameter @xmath60 both pre - specified by the researcher . here
we would simply take @xmath61 provided consumption growth @xmath62 is of the form @xmath63 for some function @xmath64 .
other structural examples include models with external habits and models with durables with pre - specified preference parameters .
[ [ case-2-sdf - is - estimated ] ] case 2 : sdf is estimated + + + + + + + + + + + + + + + + + + + + + + + + here the @xmath40 is of the form @xmath65 where the functional form of @xmath40 is known up to a parameter @xmath66 . here
@xmath66 could be of several forms : * a finite - dimensional vector of preference parameters in structural models ( e.g. @xcite and @xcite ) or risk - premium parameters in reduced - form models ( e.g. @xcite ) . * a vector of parameters @xmath67 together with a function @xmath68 , so @xmath69 .
one example is models with @xcite recursive preferences , where the continuation value function is not known when the transition law of the markov state is modeled nonparametrically ( see @xcite and the application in section [ s : emp ] ) . for such models , @xmath67 would consist of discount , risk aversion , and intertemporal substitution parameters and @xmath68 would be the continuation value function .
another example is @xcite in which @xmath67 consists of time discount and homogeneity parameters and @xmath68 is a nonparametric internal or external habit formation component .
* we could also take @xmath66 to be @xmath40 itself , in which case @xmath70 would be a nonparametric estimate of the sdf .
prominent examples include @xcite , @xcite , and @xcite . in this second case
, we consider a two - step approach to sdf decomposition . in the first step we estimate @xmath66 from a time series of data on the state , and possibly also contemporaneous data on asset returns . in the second step
we plug the first - stage estimator @xmath70 into the nonparametric procedure to recover @xmath42 , @xmath43 , @xmath46 , and related quantities . in this section
we present some sufficient conditions that ensure there is a unique solution to the perron - frobenius problems ( [ e : pev ] ) and ( [ e : pev : star ] ) .
the conditions also ensure that a long - run approximation of the form ( [ e : lrr ] ) holds .
therefore , the resulting @xmath49 and @xmath50 constructed from @xmath42 and @xmath43 as in ( [ e : pctc ] ) may be interpreted correctly as the permanent and transitory components . hs and bhs established very general identification , existence and long - run approximation results that draw upon markov process theory .
the operator - theoretic conditions that we use are more restrictive than the conditions in hs and bhs but they are convenient for deriving the large - sample theory that follows .
specifically , the conditions ensure certain continuity properties of @xmath42 , @xmath43 and @xmath46 with respect to perturbations of the operator @xmath71 .
our results are also derived for the specific parameter ( function ) space that is relevant for estimation , whereas the results in hs and bhs apply to a larger class of functions .
connections between our conditions and the conditions in hs and bhs are discussed in detail in appendix [ ax : id ] , which also treats separately the issues of identification , existence , and long - run approximation . for estimation ,
all that we require is for the conclusions of proposition [ p : id ] below hold .
therefore , the following conditions could be replaced by other sets of sufficient conditions .
let @xmath72 denote the space of all measurable @xmath73 such that @xmath74 , where @xmath75 denotes the stationary distribution of @xmath20 .
our parameter space for @xmath43 is the set of all positive functions in @xmath72 .
let @xmath76 and @xmath77 denote the @xmath72 norm and inner product .
we say that @xmath71 is bounded if @xmath78 and compact if @xmath71 maps bounded sets into relatively compact sets .
finally , let @xmath79 denote the product measure on @xmath80 .
[ a : id:0 ] let @xmath81 in ( [ e : mtau : def ] ) and @xmath71 in ( [ e : m : def ] ) satisfy the following : 1 .
@xmath71 is a bounded linear operator of the form : @xmath82 for some ( measurable ) @xmath83 that is positive ( @xmath79-almost everywhere ) 2 .
@xmath81 is compact for some @xmath38 .
* discussion of assumptions : * part ( a ) are mild boundedness and positivity conditions .
if the unconditional density @xmath84 and the transition density @xmath85 of @xmath20 exist , then @xmath86 is of the form : @xmath87 in this case , the positivity condition will hold provided @xmath40 and the densities are positive ( almost everywhere ) .
part ( b ) is weaker than requiring @xmath71 itself to be compact . to introduce the identification result ,
let @xmath88 denote the spectrum of @xmath71 ( see , e.g. , chapter vii in @xcite ) .
we say that @xmath42 is _ simple _ if it has a unique eigenfunction ( up to scale ) and _ isolated _ if there exists a neighborhood @xmath89 of @xmath42 such that @xmath90 .
as @xmath43 and @xmath46 are defined up to scale , we say that @xmath43 and @xmath46 are unique if they are unique up to scale , i.e. if @xmath91 is a positive eigenfunction of @xmath71 then @xmath92 ( almost everywhere ) for some @xmath93 .
[ p : id ] let assumption [ a : id:0 ] hold .
then : 1 .
there exists positive functions @xmath94 and a positive scalar @xmath42 such that @xmath95 solves ( [ e : pev ] ) and @xmath96 solves ( [ e : pev : star ] ) .
the functions @xmath43 and @xmath46 are the unique positive solutions ( in @xmath72 ) to ( [ e : pev ] ) and ( [ e : pev : star ] ) .
the eigenvalue @xmath42 is simple and isolated and it is the largest eigenvalue of @xmath71 .
4 . the representation ( [ e : lrr ] ) holds for all @xmath97 with @xmath98 given by : @xmath99 = { \mathbb{e}}\left [ \psi ( x_t ) \phi^*(x_t ) \right]\ ] ] under the scale normalization @xmath100 = 1 $ ] .
parts ( a ) and ( b ) are existence and identification results .
part ( c ) guarantees that @xmath42 is isolated and simple , which is used extensively in the derivation of the large sample theory .
finally , part ( d ) says that @xmath42 and @xmath43 are the relevant eigenvalue - eigenfunction pair for constructing the permanent and transitory components and links the expectation @xmath98 to @xmath46 .
in particular , if @xmath101 denotes the expectation under which @xmath98 is calculated , then the change of measure between @xmath101 and @xmath75 is : @xmath102 under the scale normalization @xmath100 = 1 $ ] . therefore , estimating @xmath43 and @xmath46 directly allows one to estimate this change of measure .
the above identification result is based on an extension of the classical krein - rutman theorem in the mathematics literature .
recently , similar operator - theoretic results have been applied to study identification in nonparametric euler equation models ( see @xcite , @xcite , and @xcite ) and semiparametric euler equation models featuring habits @xcite .
time - reversed perron - frobenius problems and long - run approximation do not feature in these other works , whereas they are important from our perspective .
identification under weaker , but related , operator - theoretic conditions is studied in @xcite .
this section introduces the estimators of the perron - frobenius eigenvalue @xmath42 and eigenfunctions @xmath43 and @xmath46 and presents the large - sample properties of the estimators .
we use a sieve approach in which the infinite - dimensional eigenfunction problem is approximated by a low - dimensional matrix eigenvector problem whose solution is easily estimated .
this methodology is an empirical counterpart to projection methods in numerical analysis ( see , e.g. , chapter 11 in @xcite ) .
let @xmath103 be a dictionary of linearly independent basis functions ( polynomials , splines , wavelets , fourier basis , etc ) and let @xmath104 denote the linear subspace spanned by @xmath105 .
the sieve dimension @xmath106 is a smoothing parameter chosen by the econometrician and should increase with the sample size .
let @xmath107 denote the orthogonal projection onto @xmath108 .
consider the projected eigenfunction problem : @xmath109 where @xmath110 is the largest real eigenvalue of @xmath111 and @xmath112 is its eigenfunction . under regularity conditions
, @xmath113 is unique ( up to scale ) for all @xmath114 large enough ( see lemma [ lem : exist ] ) .
the solution @xmath113 to ( [ e : symprob ] ) must belong to the space @xmath108 .
therefore , @xmath115 for a vector @xmath116 , where @xmath117 .
we may rewrite the eigenvalue problem in display ( [ e : symprob ] ) as : is nonsingular . ]
@xmath118 where the @xmath119 matrices @xmath120 and @xmath121 are given by : @xmath122 \label{e : gmat } \\ { \mathbf{m}}_k & = & { \mathbb{e}}[b^k(x_t ) m(x_t , x_{t+1 } ) b^k(x_{t+1 } ) ' ]
\label{e : mmat}\end{aligned}\ ] ] and where @xmath110 is the largest real eigenvalue of @xmath123 and @xmath124 is its eigenvector .
we refer to @xmath125 as the _ approximate solution _ for @xmath43 .
the approximate solution for @xmath46 is @xmath126 where @xmath127 is the eigenvector of @xmath128 corresponding to @xmath110 .
together , @xmath129 solve the generalized eigenvector problem:@xmath130 where @xmath110 is the largest real generalized eigenvalue of the pair @xmath131 .
we suppress dependence of @xmath121 and @xmath120 on @xmath114 hereafter to simplify notation . to estimate @xmath42 , @xmath43 and @xmath46 , we solve the sample analogue of ( [ e : gev ] ) , namely :
@xmath132 where @xmath133 and @xmath134 are described below and where @xmath135 is the maximum real generalized eigenvalue of the matrix pair @xmath136 .
, @xmath137 , @xmath138 can be computed using the command ` [ c , d , cstar]=eig(mhat , ghat ) ` where ` mhat ` and ` ghat ` are the estimators @xmath133 and @xmath134 . then @xmath135 is the maximum diagonal entry of ` d ` , @xmath137 is the column of ` c ` corresponding to @xmath135 , and @xmath139 is the column of ` cstar ` corresponding to @xmath135
. simultaneous computation of @xmath137 and @xmath138 is also possible in r using , for example , the function ` qz.dggev ` in the ` qz ` package . ]
the estimators of @xmath43 and @xmath46 are : @xmath140 under the regularity conditions below , the eigenvalue @xmath135 and its right- and left - eigenvectors @xmath137 and @xmath138 will be unique with probability approaching one ( see lemma [ lem : exist : hat ] ) . ) or when @xmath137 and @xmath138 are not unique , then we can simply take @xmath141 and set @xmath142 for all @xmath143 without altering the convergence rates or limiting distribution of the estimators .
this was not an issue in simulations or the empirical application . ] given a time series of data @xmath144 , a natural estimator for @xmath145 is : @xmath146 we consider two possibilities for estimating @xmath147 .
[ [ case-1-sdf - is - observable-1 ] ] case 1 : sdf is observable + + + + + + + + + + + + + + + + + + + + + + + + + first , consider the case in which the function @xmath148 is specified by the researcher . in this case
our estimator of @xmath147 is @xmath149 [ [ case-2-sdf - is - estimated-1 ] ] case 2 : sdf is estimated + + + + + + + + + + + + + + + + + + + + + + + + now suppose that the sdf is of the form @xmath65 where the functional form of @xmath40 is known up to the parameter @xmath66 which is to be estimated first from the data on @xmath20 and possibly also asset returns .
let @xmath70 denote this first - stage estimator . in this case ,
we take : @xmath150 recall that the _ long - run yield _ is @xmath151 .
we may estimate @xmath152 using : @xmath153 we may also estimate the size of the permanent component .
the _ entropy _ of the permanent component , namely @xmath154 - { \mathbb{e}}[\log(m^p_{t+1}/m^p_t)]$ ] , is known to be bounded from below by the expected excess return of any traded asset relative to a discount bond of asymptotically long maturity ( * ? ? ?
* proposition 2 ) .
previous empirical work has estimated this bound from data on equity returns and ( proxies for ) holding period returns for long - maturity discount bonds ( see , e.g. , @xcite and @xcite ) .
in contrast , here we can estimate directly the entropy and do so without data on returns on long - maturity discount bonds ( though we require data on @xmath20 and must take a stand on the sdf process ) .
in markovian environments , the entropy has the simple form @xmath155 $ ] ( see @xcite and @xcite ) . given @xmath135 , a natural estimator of @xmath156 is : @xmath157 in case 1 ; in case 2 we replace @xmath148 by @xmath158 in ( [ e : lhat1 ] ) .
the size of the permanent component may also be measured by other types of statistical discrepancies besides entropy ( e.g. cressie - read divergences ) which may be computed from the time series of the permanent component recovered empirically using @xmath135 and @xmath159 .
we confine our attention to entropy because the theoretical literature has typically used entropy to measure the size of sdfs and their permanent components over different horizons ( see , e.g. , @xcite and @xcite ) and for sake of comparison with the empirical literature on bounds .
here we establish consistency of the estimators and derive the convergence rates of the eigenfunction estimators under mild regularity conditions .
[ a : id ] @xmath71 is bounded and the conclusions of proposition [ p : id ] hold .
[ a : bias ] @xmath160 .
let @xmath161 denote the inverse of the positive definite square root of @xmath145 and @xmath162 denote the @xmath119 identity matrix .
define the `` orthogonalized '' matrices @xmath163 , @xmath164 , and @xmath165 .
let @xmath76 also denote the euclidean norm when applied to vectors and the operator norm ( largest singular value ) when applied to matrices .
[ a : var ] @xmath166 and @xmath167 .
* discussion of assumptions : * assumption [ a : bias ] requires that the space @xmath108 be chosen such that it approximates well the range of @xmath71 ( as @xmath168 ) .
similar assumptions are made in the literature on projection methods .
assumption [ a : bias ] requires that @xmath71 is compact , as has been assumed previously in the literature on sieve estimation of eigenfunctions ( see , e.g. , @xcite ) .
has range @xmath108 where @xmath169 .
if @xmath71 is not compact but @xmath170 is compact for some @xmath171 , then one can apply the estimators to @xmath170 in place of @xmath71 and estimate the solution @xmath172 to @xmath173 and similarly for @xmath46 .
consistency and convergence rates of @xmath174 , @xmath43 and @xmath46 would then follow directly from theorem [ t : rate ] . ]
assumption [ a : var ] ensures that the sampling error in estimating @xmath175 vanishes asymptotically .
this condition implicitly restricts the maximum rate at which @xmath114 can grow with @xmath1 , which will be determined by both the type of sieve and the weak dependence properties of the data .
see appendix [ ax : est : mat ] for sufficient conditions .
note that @xmath176 and @xmath177 are a proof device and do not need to be calculated in practice . before presenting the main result on convergence rates , we
first we introduce the sequence of positive constants @xmath178 , @xmath179 , @xmath180 , and @xmath181 that bound the approximation bias ( @xmath182 and @xmath179 ) and sampling error ( @xmath183 and @xmath181 ) . as eigenfunctions are only normalized up to scale , in what follows we impose the normalizations @xmath184 and @xmath185 . define : @xmath186 the quantities @xmath178 and @xmath179 measure the bias incurred by approximating @xmath43 and @xmath46 by elements of @xmath108 . and @xmath179 are available under standard smoothness assumptions ( see @xcite ) .
] let @xmath187 and @xmath188 and normalize @xmath124 and @xmath127 so that @xmath189 and @xmath190 ( these normalizations are equivalent to setting @xmath191 and @xmath192 ) . under assumption [ a : var ] , we may choose positive sequences @xmath183 and @xmath181 which are both @xmath193 , so that : @xmath194 ( see lemma [ lem : matcgce ] in the appendix for further details ) . the terms @xmath180 and @xmath181 will typically be increasing in @xmath114 and decreasing in @xmath1 : with more data we can estimate the entries of @xmath145 and @xmath147 more precisely , but with larger @xmath114 there are more parameters to estimate . [ t : rate ] let assumptions [ a : id][a : var ] hold . then : 1 .
@xmath195 2 .
@xmath196 3 .
@xmath197 where @xmath178 and @xmath179 are defined in ( [ e : deltas ] ) and @xmath180 and @xmath181 are defined in ( [ e : etas ] ) .
the convergence rates for @xmath159 and @xmath198 should be understood to hold under the scale normalizations @xmath184 , @xmath199 , @xmath200 and @xmath201 and sign normalizations @xmath202 and @xmath203 .
[ rmk : gen ] theorem [ t : rate ] holds for @xmath135 , @xmath159 and @xmath198 calculated from any estimators @xmath134 and @xmath133 that satisfy assumption [ a : var ] with @xmath134 positive definite and symmetric ( almost surely ) . theorem [ t :
rate ] is sufficiently general that it applies to models with latent state vectors without modification : all that is required is that one can construct estimators of @xmath145 and @xmath147 that satisfy assumption [ a : var ] . appendix [ ax : filter ] describes two approaches for extending the methodology to models with latent variables .
the relation ( [ e : gev ] ) may also be used to numerically compute @xmath42 , @xmath43 , and @xmath46 in models for which analytical solutions are unavailable .
for such models , the matrices @xmath147 and @xmath145 may be computed directly ( e.g. via simulation or numerical integration ) . the approximate solutions @xmath110 , @xmath113 and @xmath204 for @xmath42 , @xmath43 and @xmath46
can be recovered by solving ( [ e : gev ] ) .
lemma [ lem : bias ] gives the convergence rates @xmath205 , @xmath206 , and @xmath207 .
theorem [ t : rate ] displays the usual bias - variance tradeoff encountered in nonparametric estimation .
the bias terms @xmath178 and @xmath179 will be decreasing in @xmath114 ( since @xmath43 and @xmath46 are approximated over increasingly rich subspaces as @xmath114 increases ) . on the other hand , the variance terms @xmath180 and @xmath181 will typically be increasing in @xmath114 ( larger matrices ) and decreasing in @xmath1 ( more data ) . choosing @xmath114 to balance the bias and variance terms
will yield the best convergence rate . to investigate the theoretical properties of the estimators , we derive the convergence rate of @xmath159 in case 1 , where @xmath134 and @xmath133 are as in ( [ e : ghat ] ) and ( [ e : mhat1 ] ) under standard conditions from the statistics literature on optimal convergence rates . although the following conditions are not particularly appropriate in an asset pricing context , the result is informative about the convergence properties of @xmath159 relative to conventional nonparametric estimators .
[ c : rate ] let assumption [ a : id ] and the following conditions hold : ( i ) @xmath208 is compact , rectangular and has nonempty interior ; ( ii ) the stationary distribution @xmath75 of @xmath20 has a continuous and positive density ; ( iii ) @xmath71 is a bounded operator from @xmath72 into a hlder class of functions @xmath209 of smoothness @xmath210 @xcite ; ( iv ) @xmath211 with @xmath212 ; ( v ) @xmath213 <
\infty$ ] for some @xmath214 ; ( vi ) @xmath108 is a spanned by ( a tensor product of ) polynomial splines of degree @xmath215 with uniformly bounded mesh ratio ( * ? ? ?
* chapter 12 ) with @xmath216 ; and ( vii ) @xmath217 is exponentially rho - mixing . then : assumptions [ a : bias ] and [ a : var ] hold , @xmath218 , and @xmath219 . choosing @xmath220 yields : @xmath221 when @xmath40 is bounded ( i.e. @xmath222 ) , the convergence rate obtained in corollary [ c : rate ] is @xmath223 , which is the optimal @xmath72 convergence rate for nonparametric regression estimators with i.i.d .
data when the unknown regression function belongs to @xmath224 ( see , e.g. , @xcite ) . in this section
we establish the asymptotic normality of @xmath135 , which is also useful in establishing asymptotic normality of the estimator of the long - run yield and entropy of the permanent component ( see appendix [ ax : inf ] ) . the semiparametric efficiency bound in case 1 is also derived and @xmath135 is shown to be efficient .
to establish asymptotic normality of @xmath135 , we derive the representation : @xmath225 where the influence function @xmath226 is given by : @xmath227 with @xmath43 and @xmath46 normalized so that @xmath184 and @xmath228 .
the process @xmath229 is a martingale difference sequence ( relative to the filtration @xmath230 ) .
the asymptotic distribution of @xmath135 then follows from ( [ e : ale:1 ] ) by a central limit theorem for martingale differences . to formalize this argument , define : @xmath231 with @xmath232 and @xmath204 normalized so that @xmath191 and @xmath233 , and : @xmath234 [ a : asydist ] let the following hold : 1 .
@xmath235 2 .
@xmath236 and @xmath237 3 .
@xmath238 4 .
@xmath239 < \infty$ ] .
* discussion of assumptions : * assumption [ a : asydist](a ) is an undersmoothing condition which ensures that the approximation bias @xmath240 does not distort the asymptotic distribution of @xmath135 .
assumption [ a : asydist](b)(c ) ensures the remaining terms in ( [ e : ale:1 ] ) are @xmath241 ; sufficient conditions for assumption [ a : asydist](b ) are presented in appendix [ ax : est : mat ] .
assumption [ a : asydist](c ) requires that higher - order approximation error be asymptotically negligible .
this condition is mild : the summands in @xmath242 have expectation zero , and @xmath232 , @xmath204 and @xmath110 are converging to @xmath43 , @xmath46 , and @xmath42 by lemma [ lem : bias ] .
finally , assumption [ a : asydist](d ) ensures the asymptotic variance of @xmath135 is finite .
the following result establishes asymptotic normality of @xmath135 in case 1 .
appendix [ ax : inf ] contains further results on inference on functionals of @xmath43 and @xmath42 , such as the entropy of the permanent component .
define @xmath243 $ ] .
[ t : asydist:1 ] let assumptions [ a : id][a : asydist ] hold . then : the asymptotic linear expansion ( [ e : ale:1 ] ) holds and @xmath244 .
the following corollary lists some sufficient conditions for assumptions [ a : var][a : asydist ] .
define @xmath245 and let @xmath246 denote the @xmath247 norm @xmath248 ( so @xmath249 ) .
[ c : inf:1 ] let assumptions [ a : id ] and [ a : bias ] hold , and let : ( i ) @xmath217 be exponentially beta - mixing ( see appendix [ ax : est : mat ] ) ; ( ii ) @xmath250 for some @xmath251 , ( iii ) @xmath252 and @xmath253 for some @xmath254 , ( iv ) @xmath255 < \infty$ ] , @xmath256 and @xmath257 for some @xmath214 , ( v ) @xmath258 , and ( vi ) @xmath114 increase with @xmath1 such that @xmath259 and @xmath260 . then : ( [ e : ale:1 ] ) holds and @xmath244 .
we conclude by deriving the semiparametric efficiency bounds for case 1 . to derive the efficiency
bound we require a further technical condition ( see appendix [ ax : inf ] ) . [ t : eff ] let assumptions [ a : id][a : asydist ] and [ a : eff ] hold . then : the semiparametric efficiency bound for @xmath42 is @xmath261 and @xmath135 is semiparametrically efficient .
theorem [ t : eff ] provides further theoretical justification for using sieve methods to nonparametrically estimate @xmath42 , @xmath43 , and related quantities . in appendix
[ ax : inf ] we derive efficiency bounds for @xmath262 and @xmath263 and show that @xmath264 and @xmath265 attain their bounds .
for case 2 , we obtain the following expansion ( under regularity conditions ) : @xmath266 with @xmath226 from display ( [ e : inf : def ] ) with @xmath267 and : @xmath268 the expansion ( [ e : ale:2 ] ) indicates that the asymptotic distribution of @xmath135 and related functionals will depend on the properties of the first stage estimator @xmath70 .
we first suppose that @xmath66 is a finite - dimensional parameter and the plug - in estimator @xmath70 is root-@xmath1 consistent and asymptotically normally distributed .
the following regularity conditions are deliberately general so as to allow for @xmath70 to be any conventional parametric estimator of @xmath66 . to simplify notation , let @xmath269 .
[ a : parametric ] let the following hold : 1 .
@xmath270 for some @xmath271-valued random process @xmath272 2 .
@xmath273 satisfies a clt , i.e. : @xmath274})\ ] ] where the @xmath275 matrix @xmath276}$ ] is positive definite 3 .
there exists a neighborhood @xmath89 of @xmath66 upon which @xmath277 is continuously differentiable in @xmath278 for each @xmath279 with : @xmath280 < \infty\ ] ] for some @xmath214 4 .
@xmath281 < \infty$ ] and @xmath282 for some @xmath283 .
let @xmath284 } = ( 1\,,\ , { \mathbb{e}}[\phi^*(x_t ) \phi(x_{t+1})\frac{\partial m(x_t , x_{t+1};\alpha_0)}{\partial \alpha'}])'$ ] and define @xmath285 } = h_{[{\mathrm{2a}}]}'v_{[{\mathrm{2a}}]}^{\phantom \prime } h_{[{\mathrm{2a}}]}^{\phantom \prime}$ ] .
[ t : asydist:2a ] let assumptions [ a : id][a : parametric ] hold .
then : @xmath286})$ ] .
we now suppose that @xmath66 is an infinite - dimensional parameter .
the parameter space is @xmath287 ( a banach space ) equipped with some norm @xmath288 .
this includes the case in which ( 1 ) @xmath278 is a function , i.e. @xmath289 with @xmath290 a function space , and ( 2 ) @xmath278 consists of both finite - dimensional and function parts , i.e. @xmath291 with @xmath292 with @xmath293 . for example , under recursive preferences the vector @xmath294 could consist of discount , risk - aversion and eis parameters and @xmath295 could be the continuation value function .
inference in this case involves the ( nonlinear ) functional @xmath296 , given by : @xmath297\,.\ ] ] we focus on the case in which the functional @xmath298 is regular ( i.e. root-@xmath1 estimable ) .
we say that the functional @xmath296 is _ gateaux differentiable _ at @xmath66 in the direction @xmath299 $ ] if @xmath300 ) - \ell(\alpha_0))/\tau$ ] exists for every fixed @xmath301 .
if so , we denote the derivative by @xmath302 $ ] .
let @xmath303 and let @xmath304 .
let @xmath305 denote the centered empirical process on @xmath306 .
we say that @xmath306 is a _ donsker class _ if @xmath307 is absolutely convergent over @xmath306 to a non - negative quadratic form @xmath308 and there exists a sequence of gaussian processes @xmath309 indexed by @xmath306 with covariance function @xmath310 and a.s .
uniformly continuous sample paths such that @xmath311 as @xmath312 ( see @xcite ) .
[ a : nonpara ] let the following hold : 1 . @xmath313 is gateaux differentiable at @xmath66 and @xmath314| = o(\|\alpha - \alpha_0\|^2_{{\mathcal{a}}})$ ] as @xmath315 2 .
@xmath316 = \frac{1}{\sqrt n } \sum_{t=0}^{n-1 } \psi_{\ell , t } + o_p(1)$ ] for some @xmath317-valued random process @xmath318 , @xmath319 , and @xmath320 for some @xmath321 3 .
@xmath322 satisfies a clt , i.e. : @xmath323})\ ] ] where the @xmath324 matrix @xmath325}$ ] is positive definite 4 .
@xmath306 is a donsker class and @xmath326 is exponentially beta - mixing 5 .
@xmath327 < \infty$ ] and @xmath328 for some @xmath214 , @xmath329 and @xmath330 for some @xmath331 , and @xmath332 .
parts ( a ) and ( b ) are standard conditions for inference in nonlinear semiparametric models ( see , e.g. , theorem 4.3 in @xcite ) .
part ( c ) is a mild clt condition .
the covariance function @xmath310 is well defined under parts ( d ) and ( e ) .
sufficient conditions for the class @xmath306 to be donsker are well known ( see , e.g. , @xcite ) . for the following theorem ,
let @xmath333 } = ( 1,1)'$ ] and define @xmath334 } = h_{[{\mathrm{2b}}]}'v_{[{\mathrm{2b}}]}^{\phantom \prime } h_{[{\mathrm{2b}}]}^{\phantom \prime}$ ] .
sufficient conditions for assumption [ a : var ] are presented in appendix [ ax : est : mat ] for this case .
[ t : asydist:2b ] let assumptions [ a : id][a : asydist ] and [ a : nonpara ] hold .
then : @xmath335})$ ] . for case @xmath336
, the efficiency bounds will depend on the assumptions made about the model from which @xmath278 was estimated ( say , whether it is an unconditional or conditional moment restriction model ) .
we therefore do not derive efficiency bounds for this case .
this section describe how to nonparametrically estimate the value function and sdf in a class of recursive preference models by solving a _ nonlinear _ perron - frobenius eigenfunction problem .
we focus on @xcite recursive preferences with unit elasticity of intertemporal substitution ( eis ) .
this class of preferences may also be interpreted as risk - sensitive preferences as formulated by @xcite ( see @xcite ) . after describing the setup , we present some regularity conditions for local identification .
the local identification argument is constructive .
we then introduce the estimators and derive their large - sample properties . under epstein - zin preferences ,
the date-@xmath7 utility of the representative agent is defined via the recursion : @xmath337^{\frac{1-\theta}{1-\gamma } } \right\}^{\frac{1}{1-\theta}}\ ] ] where @xmath338 is date-@xmath7 consumption , @xmath339 is the eis , @xmath340 is the time discount parameter , and @xmath341 is the relative risk aversion parameter .
we maintain our assumption of a markov state process @xmath20 .
let consumption growth , namely @xmath342 , be a measurable function of @xmath343 .
@xcite show that the scaled continuation value @xmath344 may be written as @xmath345 where @xmath346 solves the fixed point equation : @xmath347^{\frac{1-\theta}{1-\gamma } } \bigg\}^{\frac{1}{1-\theta } } \,.\ ] ] with unit eis ( i.e. @xmath348 ) the fixed point equation ( [ e : ezrecur ] ) reduces to : @xmath349\ ] ] where @xmath350 .
the sdf is of the form : @xmath351}\,.\end{aligned}\ ] ] the dynamics of @xmath20 determine the value function @xmath352 and the conditional expectation in the denominator of the sdf .
the value function and conditional expectation are unknown when the dynamics of @xmath20 are treated nonparametrically .
the following reformulation is convenient for estimation . setting @xmath353 and rearranging ( [ e : fp : prelim ] ) yields : @xmath354 \,.\ ] ] this fixed - point equation may be written more compactly as @xmath355 , where : @xmath356 \end{aligned}\ ] ] ( as we seek a positive solution , taking an absolute value inside the conditional expectation does not change the fixed point ) .
we may then rewrite the sdf as : @xmath357 dividing @xmath358 by @xmath359 and using homogeneity allows us to reformulate the fixed - point problem as a nonlinear perron - frobenius problem : @xmath360 where @xmath361 is a positive eigenfunction of @xmath362 and @xmath363 is its eigenvalue . since @xmath362 is nonlinear , in this section we normalize the eigenfunction @xmath364 to have unit norm .. for example , @xmath365 is a positive eigenfunction of @xmath362 with eigenvalue @xmath366 for any @xmath367 . ]
combining equations ( [ e : efn : nl:0 ] ) and ( [ e : efn : nl ] ) yields : @xmath368 we show how to estimate @xmath364 and @xmath251 from time series data on @xmath20 .
the estimates @xmath369 and @xmath370 can be plugged into ( [ e : rec : sdf ] ) to obtain empirically a time series of the sdf process without assuming a parametric law of motion for @xmath20 . in this section
we provide sufficient conditions for local identification of @xmath295 .
we establish the results for the parameter ( function ) space @xmath72 because it is convenient for sieve estimation .
one can not establish ( global ) identification using contraction mapping arguments because @xmath362 is not a contraction on @xmath72 .
has a positive fixed point @xmath371 .
the function @xmath372 is also a fixed point .
therefore , @xmath362 is not a contraction on @xmath72 ( else the banach contraction mapping theorem would yield a unique fixed point ) . ]
some of the regularity conditions we require for estimation are sufficient for @xmath362 to satisfy a local ergodicity property .
this , in turn , is sufficient for local identification .
it may be possible to strengthen the conditions to achieve global identification . to describe the local ergodicity property , first choose some ( nonzero ) function @xmath97 and set @xmath373 .
then calculate : @xmath374 for @xmath375 .
proposition [ p : nl ] below shows that the sequence @xmath376 converges to @xmath364 for any starting value @xmath54 in a suitably defined region ( @xmath295 and @xmath251 can then be obtained by setting @xmath377 and @xmath378 ) .
this is similar to various `` stability '' results in the literature on balanced growth following @xcite .
space we study has empty interior .
if @xmath379 is bounded then these previous results may be used to derive ( global ) identification conditions in the space @xmath380 .
however , bounded support seems inappropriate for common choices of state variable , such as consumption growth and dividend growth .
] there , @xmath381 is a homogeneous of degree one input - output system , @xmath382 lists the proportions of commodities in the economy in period @xmath1 , and @xmath383 is normalized by its @xmath384 norm so that @xmath385 lists the proportions in period @xmath386 .
`` stability '' concerns convergence of the sequence @xmath387 to a positive eigenvector @xmath364 of @xmath362 ( representing balanced growth proportions ) .
write @xmath388 where @xmath389 and : @xmath390 \,.\ ] ] as is known from the theory of hammerstein operators ( see chapter 5 of @xcite ) , @xmath362 is a bounded ( respectively , compact ) operator on @xmath72 whenever @xmath391 is bounded ( compact ) on @xmath72 .
we say that @xmath391 is _ positive _ if @xmath392 is positive for any non - negative @xmath97 that is not identically zero .
positivity of @xmath391 ensures that the iterative procedure is well defined and that any non - zero fixed point of @xmath362 is strictly positive ( almost everywhere ) .
we say that @xmath362 is _ frchet differentiable _ at @xmath295 if there exists a bounded linear operator @xmath393 such that : @xmath394 if it exists , the frchet derivative @xmath395 of @xmath362 is given by : @xmath396 \,.\ ] ] let @xmath397 denote the spectral radius of @xmath395 [ p : nl ] let @xmath391 be positive and bounded and let @xmath362 be frchet differentiable at @xmath295 with @xmath398 .
then : there exists finite positive constants @xmath399 with @xmath400 and a neighborhood @xmath89 of @xmath364 such that : @xmath401 for any initial point @xmath54 in the cone @xmath402 .
we say that @xmath364 is _ locally identified _
if there exists a neighborhood @xmath89 of @xmath364 such that @xmath364 is the unique eigenfunction of @xmath362 belonging to @xmath403 where @xmath404 denotes the unit sphere in @xmath72 ( recall we normalize eigenfunctions of @xmath362 to have unit norm ) .
similarly , we say that @xmath295 is locally identified if it is the unique fixed point of @xmath362 belonging to some neighborhood @xmath405 of @xmath295 . to see why local identification follows from proposition [ p :
nl ] , suppose @xmath406 is a positive eigenfunction of @xmath362 belonging to @xmath403 .
proposition [ p : nl ] implies that @xmath407 for each @xmath1 , hence @xmath408 .
local identification of @xmath295 follows similarly .
[ c : local - id ] @xmath295 and @xmath364 are locally identified under the conditions of proposition [ p : nl ] .
local identification of @xmath364 and positive homogeneity of @xmath362 imply that @xmath295 is the unique fixed point of @xmath362 in the cone @xmath409 , which is stronger than local identification of @xmath295 .
existence and ( global ) identification of value functions in models with recursive preferences has been studied previously ( see @xcite , @xcite , and references therein ) . the most closely related work to ours
is @xcite , who study existence and uniqueness of value functions for markovian environments in @xmath410 spaces ( whose cones of non - negative functions also have empty interior ) . @xcite provide conditions under which a fixed point may exist when the eis is equal to unity but do not establish its uniqueness .
their existence conditions are based , in part , on existence of a positive eigenfunction of the operator @xmath391 .
there is also a connection between corollary [ c : local - id ] and the literature on local identification of nonlinear , nonparametric econometric models .
we can write @xmath358 as the conditional moment restriction : @xmath411 = 0\ ] ] ( almost surely ) .
the conditions of proposition [ p : nl ] ensure that the above moment restriction is frchet differentiable at @xmath295 with derivative @xmath412 .
the condition @xmath398 implies that @xmath412 is invertible on @xmath72 .
the conditions in proposition [ p : nl ] are therefore similar to the differentiability and rank conditions that @xcite use to study local identification in nonlinear conditional moment restriction models .
we again use a sieve approach to reduce the infinite - dimensional problem to a low - dimensional ( nonlinear ) eigenvector problem .
consider the projected fixed - point problem : @xmath413 where @xmath107 is the orthogonal projection onto the sieve space defined in section [ s : est ] . under the assumptions below ,
the approximate solution @xmath414 is well defined for all @xmath114 sufficiently large . as @xmath415
, we may write @xmath416 for some vector @xmath417 which solves : @xmath418 where @xmath419 $ ] . to simplify notation we drop dependence of @xmath120 and @xmath420 on @xmath114 hereafter . for estimation , we solve a sample analogue of ( [ e : fp : vec ] ) , namely : @xmath421 where @xmath134 is defined in display ( [ e : ghat ] ) and @xmath422 is given by : @xmath423 under the regularity conditions below , a solution @xmath424 on a neighborhood of @xmath425 necessarily exists wpa1 ( see lemma [ lem : fphat : exist ] ) .
the estimators of @xmath295 , @xmath364 and @xmath251 are : @xmath426 the estimators @xmath369 and @xmath370 can then be plugged into display ( [ e : rec : sdf ] ) to obtain the sdf consistent with preference parameters @xmath59 , @xmath60 and the observed law of motion of the state .
we say that @xmath362 is _ continuously frchet differentiable _ at @xmath295 if @xmath362 is differentiable on a neighborhood of @xmath295 and @xmath427 as @xmath428 .
[ a : fp : exist ] let the following hold : 1 .
@xmath362 has a unique positive fixed point @xmath371 2 .
@xmath391 is compact and positive 3 .
@xmath362 is continuously frchet differentiable at @xmath295 with @xmath398 .
[ a : fp : bias ] let the following hold : 1 .
@xmath429 2 .
@xmath108 is dense in the range of @xmath362 as @xmath168 .
let @xmath176 be as in assumption [ a : var ] .
let @xmath430 and @xmath431 .
[ a : fp : var ] @xmath432 and @xmath433 for each @xmath367 .
* discussion of assumptions : * assumption [ a : fp : exist ] imposes some mild structure on @xmath362 which ensures that fixed points of @xmath362 are continuous under perturbations . assumption [ a : fp : bias](a ) is analogous to assumption [ a : bias ] .
assumptions [ a : fp : exist ] and [ a : fp : bias ] are standard in the literature on solving nonlinear equations with projection methods .
finally , assumption [ a : fp : var ] is similar to assumption [ a : var ] and restricts the rate at which the sieve dimension @xmath114 can grow with @xmath1 .
sufficient conditions for assumption [ a : fp : var ] are presented in appendix [ ax : est : mat : fp ] .
let @xmath434 denote the bias in approximating @xmath295 by an element of the sieve space .
assumption [ a : fp : bias](b ) implies that @xmath435 . to control the sampling error , fix any small @xmath436 . by assumption [ a : fp : var ]
we may choose a sequence of positive constants @xmath437 with @xmath438 such that : @xmath439 [ t : fpest ] let assumptions [ a : fp : exist][a : fp : var ] hold .
then : 1 .
@xmath440 2 .
@xmath441 3 .
@xmath442 .
the convergence rates obtained in theorem [ t : fpest ] again exhibit a bias - variance tradeoff .
the bias terms @xmath443 are decreasing in @xmath114 , whereas the variance term @xmath437 is typically increasing in @xmath114 but decreasing in @xmath1 .
choosing @xmath114 to balance the terms will lead to the best convergence rate . for implementation
, we recommend the following iterative scheme based on proposition [ p : nl ] . set @xmath444 , then calculate : @xmath445 for @xmath446 .
if the sequence @xmath447 converges to @xmath448 ( say ) , we then set : @xmath449 this procedure converged in the simulations and empirical application and was more efficient than numerical solution of the sample fixed - point problem ( [ e : fp : sample ] ) .
the following monte carlo experiment illustrates the performance of the estimators in consumption - based models with power utility and recursive preferences .
the state variable is log consumption growth , denoted @xmath450 , which evolves as a gaussian ar(1 ) process : @xmath451 the parameters for the simulation are @xmath452 , @xmath453 , and @xmath454 .
the data are constructed to be somewhat representative of quarterly growth in u.s .
real per capita consumption of nondurables and services ( for which @xmath455 and @xmath456 ) .
however , we make the consumption growth process twice as persistent to produce more nonlinear eigenfunctions and twice as volatile to produce a more challenging estimation problem .
we consider a power utility design in which @xmath61 and a design with recursive preferences with unit eis , whose sdf is presented in display ( [ e : rec : sdf ] ) . for both designs we set @xmath457 and @xmath458 .
the parameterization @xmath457 and @xmath459 is typically used in calibrations of long - run risks models ( @xcite ) ; here we take @xmath458 to produce greater nonlinearity in the eigenfunctions . for each design
we generate 50000 samples of length 400 , 800 , 1600 , and 3200 .
results reported in this section use either a hermite polynomial or cubic b - spline basis of dimension @xmath460 .
the hermite polynomials are normalized by the mean and standard deviation of the data to be approximately orthonormal and the knots of the b - splines are placed evenly at the quantiles of the data .
the results were reasonably insensitive both to the choice of sieve and to the dimension of the sieve space .
we estimate @xmath43 , @xmath46 , @xmath42 , @xmath152 , and @xmath156 for both designs and we also estimate @xmath364 and @xmath251 for the recursive preference design . to implement the estimators @xmath135 , @xmath159 , and @xmath198
, we use the estimator @xmath134 in ( [ e : ghat ] ) for both preference specifications . for power utility
we use the estimator @xmath133 in ( [ e : mhat1 ] ) . for recursive preferences
we first estimate @xmath461 using the method described in the previous section , then construct the estimator @xmath133 as in display ( [ e : mhat2 ] ) , using the plug - in sdf : @xmath462 based on first - stage estimators @xmath463 of @xmath461 .
we normalize @xmath159
so that @xmath464 and @xmath198 so that @xmath465 . we also normalize @xmath369 so that @xmath466 .
the bias and rmse of the estimators are presented in tables [ tab : mc1 ] and [ tab : mc2 ] . , @xmath198 , and @xmath369 ,
for each replication we calculate the @xmath72 distance between the estimators and their population counterparts , then take the average over the mc replications . to calculate the bias we take the average of the estimators across the mc replications to produce @xmath467 , @xmath468 , and @xmath469 ( say ) ,
then compute the @xmath72 distance between @xmath470 , @xmath471 and @xmath406 and the true @xmath43 , @xmath46 and @xmath364 .
the use of the `` bias '' here is not to be confused with the bias term in the convergence rate calculations .
there `` bias '' measures how close @xmath113 , @xmath204 , and @xmath110 are to @xmath43 , @xmath46 , and @xmath42 . here
`` bias '' of an estimator refers to the distance between the parameter and the average of its estimates across the mc replications .
similar calculations are performed for @xmath135 , @xmath264 , @xmath265 , and @xmath370 . ]
table [ tab : mc1 ] shows that @xmath43 , @xmath46 and @xmath364 may be estimated with small bias and rmse using a reasonably low - dimensional sieve for both hermite polynomials and cubic b - splines .
table [ tab : mc2 ] presents similar results for @xmath135 , @xmath264 , @xmath265 and @xmath370 .
the rmses for @xmath159 and @xmath135 under recursive preferences are typically smaller than the rmses for @xmath159 and @xmath135 under power utility , even though with recursive preferences the continuation value must be first estimated nonparametrically .
in contrast , the rmse for @xmath198 is larger under recursive preferences , which is likely due to the fact that @xmath46 is much more curved for that design ( as evident from comparing the vertical scales figures [ fig : mc : sfig2 ] and [ fig : mc : sfig4 ] ) .
the results in table [ tab : mc1 ] also show that @xmath364 may be estimated with a reasonably small degree of bias and rmse in moderate samples .
figures [ fig : mc : sfig1][fig : mc : sfig5:bs ] also present ( pointwise ) confidence intervals for @xmath43 , @xmath46 and @xmath364 computed across simulations of different sample sizes . for each figure ,
the true function lies approximately in the center of the pointwise confidence intervals , and the widths of the intervals shrink noticeably as the sample size @xmath1 increases .
the results for hermite polynomials and cubic b - splines are similar . however , the intervals are somewhat narrower at their extremities for cubic b - splines than for hermite polynomials .
[ cols="^,^,^,^,^,^,^ " , ] table [ tab : emp ] presents the estimates and 90% confidence intervals . ,
@xmath60 , @xmath251 , @xmath364 , @xmath42 , @xmath152 , and @xmath156 for each bootstrap replication .
we discard the fraction of replications in which the estimator of @xmath472 failed to converge . in the right panel
we fix @xmath59 and @xmath60 and re - estimate @xmath251 , @xmath364 , @xmath42 , @xmath152 , and @xmath156 for each bootstrap replication .
] there are several notable aspects .
first , both state specifications are able to generate a permanent component whose entropy is consistent with a return premium of around 2.1% per quarter relative to the long bond , which is in the ballpark of empirically reasonable estimates .
second , the estimated long - run yield of around 2.2% per quarter is too large , which is explained by the low value of @xmath473 .
third , the estimated entropy of the sdf itself is @xmath474 for the bivariate specification and @xmath475 for the univariate specification .. ] therefore , the estimated horizon dependence ( the difference between the entropy of the permanent component and the entropy of the sdf ; see @xcite ) is about 0.02% and 0.07% per quarter , respectively .
these estimates are well inside the bound of @xmath4760.1% per month that @xcite argue is required to match the spread in average yields between short- and long - term nominal bonds .
finally , the estimates of @xmath60 are quite imprecise , in agreement with previous studies ( see , e.g. , @xcite who use a similar data set and a different , but related , estimation technique ) .
the confidence intervals for @xmath42 , @xmath152 and @xmath156 in the left panel of table [ tab : emp ] are quite wide because they reflect , in large part , the uncertainty in estimating @xmath59 and @xmath60 .
experimentation with different sieve dimensions resulted in estimates of @xmath60 between @xmath477 and @xmath478 , which is in agreement with previous studies . using aggregate consumption data and @xmath479 using stockholder consumption data .
further , with stockholder data their estimated eis is not significantly different from zero .
this suggests that our estimates of @xmath60 and maintained assumption of a unit eis are empirically plausible . ]
the right panel of table [ tab : emp ] presents estimates of @xmath42 , @xmath152 and @xmath156 fixing @xmath480 and @xmath481 , @xmath482 , and @xmath478 ( @xmath364 and @xmath251 are still estimated nonparametrically ) .
it is clear that the resulting confidence intervals are much narrower once the uncertainty in estimating @xmath59 and especially @xmath60 is shut down .
for example , with @xmath483 the estimate of @xmath156 is very similar to that obtained for @xmath484 , but the confidence interval shrinks from @xmath485 to @xmath486 .
confidence intervals for @xmath42 and @xmath152 are also considerably narrower .
we now turn to analyzing the time - series properties of the permanent and transitory components .
the upper panel of figure [ fig : ts ] presents time - series plots of the sdf under recursive preferences obtained at the parameter estimates @xmath487 for the bivariate state specification .
the upper panel also plots the time series of the permanent and transitory components , which are constructed as : @xmath488 as can be seen , the great majority of the movement in the sdf is attributable to movements in the permanent component .
both evolve closely over time and exhibit strong counter - cyclicality .
the transitory component for recursive preferences is small , coherent with the literature on bounds which finds that the transitory component should be substantially smaller than the permanent component .
the correlation of the permanent component series @xmath489 and gdp growth is approximately @xmath490 whereas the correlation of the transitory component series @xmath491 and @xmath492 is approximately @xmath493 .
the lower panel of figure [ fig : ts ] displays the time series of the sdf and permanent and transitory components obtained under power utility using the same @xmath494 .
this panel shows that the permanent component , which is similar to that obtained under recursive preferences , is much more volatile than the sdf series .
the large difference between the sdf and permanent component series under power utility is due to a very volatile transitory component , which implies a counterfactually large spread in average yields between short- and long - term bonds ( @xcite ) . to understand further the long - run pricing implications of the estimates of @xmath42 , @xmath43 and @xmath46 , figures [ fig : emp : phi_1][fig : emp : phistar_2 ] plot the estimated @xmath43 and @xmath46 under recursive preferences for both the bivariate and univariate state specifications .
it is clear that both estimates of @xmath43 are reasonably flat , which explains the small variation in the transitory component in figure [ fig : ts ] .
however , the estimated @xmath46 has a pronounced downward slope in @xmath495 .
the bivariate specification shows that this is only part of the story , as the estimated @xmath46 is also downward - sloping in @xmath496 , especially when consumption growth is relatively low .
what is the economic significance of these estimates ?
proposition [ p : id ] shows that @xmath497 is the radon - nikodym derivative of the measure corresponding to @xmath98 , say @xmath101 , with respect to @xmath75
. figures [ fig : emp : rn_1][fig : emp : rn_2 ] plot the estimated change of measure for the two state specifications . as the estimate of @xmath43 is relatively flat
, the estimated change of measure is characterized largely by @xmath198 .
therefore , in the bivariate case , the distribution @xmath101 assigns relatively more mass to regions of the state space in which there is low dividend and consumption growth than the stationary distribution @xmath75 , and relatively less mass to regions with high consumption growth .
.5 for @xmath498,title="fig : " ] .5 for @xmath498,title="fig : " ] + .5 for @xmath498,title="fig : " ] .5 for @xmath498,title="fig : " ] + .5 for @xmath498,title="fig : " ] .5 for @xmath498,title="fig : " ] finally , we turn to the role of nonlinearities and non - gaussianity in explaining certain features of the long - end of the term structure .
figure [ fig : emp : yield ] presents nonparametric estimates of ( a ) the ( quarterly ) long - run yield and ( b ) the covariance between the logarithm of the permanent and transitory components , namely @xmath499 and @xmath500 , recovered from the data on @xmath498 with @xmath457 and @xmath60 increased from @xmath501 to @xmath502 .
the nonparametric estimates are presented alongside estimates for two parametric specifications of the state process .
the first specification assumes @xmath498 evolves as a gaussian var(1 ) with constant variance .
the second specification is a gaussian ar(1 ) for log consumption growth with stochastic volatility : @xmath503 where @xmath504 is a first - order autoregressive gamma process ( i.e. a discrete - time versions of the feller square - root process ; see @xcite ) .
the state vector for the second specification is @xmath505 .
we refer to this second specification as sv - ar(1 ) .
the long - run yield and the covariance cov(@xmath506,@xmath507 ) were obtained analytically as functions of @xmath59 , @xmath60 , and the estimates of the var(1 ) and sv - ar(1 ) parameters .
, @xmath508 and the three autoregressive gamma process parameters for the sv - ar(1 ) model are estimated via indirect inference @xcite using a garch(1,1 ) auxiliary model .
analytic solutions for the stochastic volatility specification are available by following arguments in appendix h of @xcite . ] .5 , @xmath509),title="fig : " ] .5 , @xmath509),title="fig : " ] figure [ fig : emp : yld ] shows that the nonparametric estimates of the long - run yield are non - monotontic , whereas the parametric estimates are monotonically decreasing .
this non - monotonicity is not apparent in the nonparametric estimates using @xmath510 .
it is also clear that the nonparametric estimates of the long - run yield are much larger for larger values of @xmath60 than the corresponding results from either of the two parametric models .
figure [ fig : emp : cov ] displays the covariance between the log of the nonparametric estimates of the permanent and transitory components for different values of @xmath60 against those obtained for the two parametric state processes . the covariance of the nonparametric estimates of the permanent and transitory components is negative for low to moderate values of @xmath60 , but becomes positive for larger values of @xmath60 .
in contrast , the parametric estimates are negative and decreasing in @xmath60 .
a recent literature has emphasized the role of positive dependence between the permanent and transitory components in explaining excess returns of long - term bonds @xcite .
positive dependence also features in models in which the term structure of risk prices is downward sloping ( see , e.g. , the example presented in section 7.2 in @xcite ) .
however , positive dependence is known to be difficult to generate via conventional preference specifications in workhorse models with exponentially - affine dynamics , such as the long - run risks model of @xcite .
although the covariance is poorly estimated for large values of @xmath60 , this finding at least suggests that nonlinearities in state dynamics may have a role to play in explaining salient features of the long end of the yield curve .
this paper introduces econometric methods to extract the permanent and transitory components of the sdf process in the long - run factorization of @xcite , @xcite , and @xcite .
we show how to estimate the solution to the perron - frobenius eigenfunction problem directly from a time series of data on state variables and the sdf process . by estimating directly the perron - frobenius eigenvalue and eigenfunction
, we can ( 1 ) reconstruct the time series of the permanent and transitory components and investigate their properties ( e.g. the size of the components , their correlation , etc ) , and ( 2 ) estimate both the yield and the change of measure which characterizes pricing over long investment horizons .
this represents a useful contribution relative to existing empirical work which has established bounds on various moments of the permanent and transitory components as functions of asset returns , but has not extracted these components directly from data .
the methodology is nonparametric in that it does not impose tight parametric restrictions on the dynamics of the state .
we view this approach as being complementary to parametric methods and useful to help better understand the roles of dynamics and preferences in structural macro - finance models .
the main technical contributions of this paper are to introduce nonparametric sieve estimators of the eigenvalue and eigenfunctions and establish consistency , convergence rates and asymptotic normality of the estimators , and some efficiency properties .
we also introduce nonparametric estimators of the continuation value function under @xcite recursive preferences with unit eis and study their large - sample properties .
the econometric methodology may be extended and applied in several different ways .
first , the methodology can be applied to study more general multiplicative functional processes such as the valuation and stochastic growth processes in @xcite , @xcite , and @xcite .
second , the methodology can be applied to models with latent state variables .
the main theoretical results ( theorems [ t : rate ] and [ t : fpest ] ) are sufficiently general that they apply equally to such cases .
finally , our analysis was conducted within the context of structural models in which the sdf process was linked tightly to preferences .
a further extension would be to apply the methodology to sdf processes which are extracted flexibly from panels of asset returns data . to this end
, this methodology could be applied in conjunction with information projection - based sdf estimation procedures developed recently by @xcite and @xcite .
supplementary appendix for nonparametric stochastic discount factor decomposition this appendix contains material to support the paper `` nonparametric stochastic discount factor decomposition '' .
appendix [ ax : est ] presents further results on estimation of the eigenvalue and eigenfunctions used for sdf decomposition and on the estimation of the continuation value function under recursive preferences .
appendix [ ax : inf ] contains additional results on inference .
appendix [ ax : id ] further details on the relation between the identification and existence conditions in section [ s : id ] and the identification and existence conditions in @xcite and @xcite .
appendix [ ax : filter ] discusses extension of the sdf decomposition methodology to models with latent state variables .
proofs of all results are presented in appendix [ ax : proofs ] .
in this section we present some supplementary lemmas from which theorems [ t : rate ] and [ t : fpest ] follow immediately . we also present some sufficient conditions for assumption [ a : var ] and [ a : fp : var ] . when we refer to an eigenfunction / eigenvector as being unique , we mean unique up to sign and scale normalization .
the results below for @xmath42 and @xmath43 are derived using arguments from @xcite ( note that @xcite study a selfadjoint operator whereas here the operator is nonselfadjoint ) .
our results for @xmath46 are new .
the first result shows that the approximate solutions @xmath110 , @xmath232 and @xmath204 from the eigenvalue problem ( [ e : gev ] ) are well defined and unique for all @xmath114 sufficiently large .
it should be understood throughout that @xmath511 , @xmath127 , @xmath137 and @xmath138 are all real - valued eigenvectors .
[ lem : exist ] let assumptions [ a : id ] and [ a : bias ] hold
. then there exists @xmath512 such that for all @xmath513 , the maximum eigenvalue @xmath110 of the eigenvector problem ( [ e : gev ] ) is real and simple , and hence @xmath514 has unique right- and left - eigenvectors @xmath511 and @xmath127 corresponding to @xmath110 .
[ lem : bias ] let assumptions [ a : id ] and [ a : bias ] hold .
then : 1 .
@xmath515 2 .
@xmath206 3 . @xmath207 where @xmath178 and @xmath179 are defined in display ( [ e : deltas ] ) .
the rates should be understood to hold under the scale normalizations @xmath184 , @xmath516 , @xmath517 , and @xmath518 and the sign normalizations @xmath519 and @xmath520 .
the following result shows that the solutions @xmath135 , @xmath521 and @xmath138 to the sample eigenvalue problem ( [ e : est ] ) are well defined and unique with probability approaching one ( wpa1 ) .
[ lem : exist : hat ] let assumptions [ a : id ] , [ a : bias ] , and [ a : var ] hold . then wpa1 , the maximum eigenvalue @xmath135 of the generalized eigenvector problem ( [ e : est ] ) is real and simple , and hence @xmath522 has unique right- and left - eigenvectors @xmath137 and @xmath138 corresponding to @xmath135 .
[ lem : var ] let assumptions [ a : id ] , [ a : bias ] , and [ a : var ] hold .
then : 1 .
@xmath523 2 .
@xmath524 3 . @xmath525 where @xmath180 and @xmath181 are defined in display ( [ e : etas ] ) .
the rates should be understood to hold under the scale normalizations @xmath199 , @xmath516 , @xmath526 and @xmath518 and the sign normalizations @xmath527 and @xmath528 .
the following two lemmas are known in the literature on solution of nonlinear equations by projection methods ( see , e.g. , chapter 19 of @xcite ) .
the first result shows that the approximate solution @xmath414 is well defined for all @xmath114 sufficiently large and converges to the fixed point @xmath295 as @xmath168 .
[ lem : fp : exist ] let assumptions [ a : fp : exist ] and [ a : fp : bias ] hold
. then there exists @xmath512 and @xmath436 such that for all @xmath513 the projected problem @xmath529 has a unique solution @xmath414 in the ball @xmath530 .
moreover , @xmath531 as @xmath168 .
let @xmath532 and @xmath533 .
[ lem : bias : fp ] let assumptions [ a : fp : exist ] and [ a : fp : bias ] hold .
then @xmath534 , @xmath535 , and @xmath414 are each well defined for all @xmath114 sufficiently large , and : 1 .
@xmath536 2 .
@xmath537 3 .
@xmath538 .
we now show that the sample fixed - point problem has a solution wpa1 , and that the solution is a consistent estimator of @xmath295 .
the following two results are new .
[ lem : fphat : exist ] let assumptions [ a : fp : exist][a : fp : var ] hold .
then wpa1 , there exists a fixed point @xmath424 of @xmath539 on a neighborhood of @xmath425 and the estimator @xmath540 from ( [ e : fpest ] ) satisfies @xmath541 .
the following result bounds the sampling error .
notice that the `` bias term '' @xmath443 shows up due to the nonlinearity of @xmath362 .
in contrast , the bias term @xmath182 does not appear in the corresponding `` variance terms '' for the linear eigenvalue problem ( see lemma [ lem : var ] ) .
[ lem : var : fp ] let assumptions [ a : fp : exist][a : fp : var ] hold .
then wpa1 there exists a fixed point @xmath424 of @xmath539 such that the estimators @xmath370 , @xmath369 , and @xmath540 from ( [ e : fpest ] ) satisfy : 1 .
@xmath542 2 .
@xmath543 3 .
@xmath544 .
we derive the results assuming that the state process @xmath217 is either beta - mixing ( absolutely regular ) or rho - mixing .
we use these dependence concepts because a variety of models for macroeconomic and financial time series exhibit these types weak dependence .
examples include copula - based models @xcite and discretely sampled diffusion processes @xcite .
the beta - mixing coefficient between two @xmath545-algebras @xmath546 and @xmath547 is : @xmath548 with the supremum taken over all @xmath546-measurable finite partitions @xmath549 and @xmath547-measurable finite partitions @xmath550 .
the beta - mixing coefficients of @xmath551 are defined as : @xmath552 we say that @xmath553 is _ exponentially beta - mixing _ if @xmath554 for some @xmath555 and @xmath556 .
the rho - mixing coefficients of @xmath326 are defined as : @xmath557 = 0 , \|\psi\| = 1 } { \mathbb{e}}\big [ { \mathbb{e}}[\psi(x_{t+q})|x_t]^2\big]^{1/2}\,.\ ] ] we say that @xmath326 is _ exponentially rho - mixing _ if @xmath558 for some @xmath559 .
the following two lemmas derive convergence rates for the estimators @xmath134 in ( [ e : ghat ] ) and @xmath133 in ( [ e : mhat1 ] ) .
these results may be used to verify assumptions [ a : var ] and [ a : asydist](b ) and bound the terms @xmath183 and @xmath181 in display ( [ e : etas ] ) .
lemma [ lem : beta:1 ] uses an exponential inequality for sums of weakly - dependent random matrices derived in @xcite .
recall that @xmath560 .
[ lem : beta:1 ] let the following hold : 1 .
@xmath217 is strictly stationary and exponentially beta - mixing 2 .
@xmath561 < \infty$ ] for some @xmath562 3
. @xmath563 .
then : assumption [ a : var ] holds , @xmath564 and we may take @xmath565 .
the following lemma extends arguments from @xcite .
[ lem : rho:1 ] let the following hold : 1 .
@xmath217 is strictly stationary and exponentially rho - mixing 2 .
@xmath561 < \infty$ ] for some @xmath562 3 .
@xmath566 . then : assumption [ a : var ] holds , @xmath567 and we may take @xmath568 . the following lemma derives convergence rates for the estimators @xmath134 in ( [ e : ghat ] ) and @xmath133 in ( [ e : mhat2 ] ) when @xmath569 is a finite - dimensional parameter .
if the first - stage estimator @xmath70 converges at a root-@xmath1 rate ( which corresponds to taking @xmath570 in the following lemma ) , then the same convergence rates for @xmath135 , @xmath159 and @xmath198 are obtained as for case 1 . for brevity , in the following two results we consider only the case in which the state process is beta - mixing .
[ lem : beta:2 ] let the conditions of lemma [ lem : beta:1 ] hold for @xmath267 , and let : 1 .
@xmath571 for some @xmath572 2 .
@xmath573 be continuously differentiable in @xmath278 on a neighborhood @xmath89 of @xmath66 for all @xmath279 with : @xmath574 < \infty\ ] ] 3 .
@xmath575 . then :
assumption [ a : var ] holds , @xmath576 and we may take @xmath577 .
we now derive convergence rates for the estimators @xmath134 in ( [ e : ghat ] ) and @xmath133 in ( [ e : mhat2 ] ) for the semi / nonparametric case in which @xmath578 is an infinite - dimensional parameter .
the parameter space is @xmath287 ( a banach space ) equipped with some norm @xmath288 .
this includes the case in which @xmath278 is a function , i.e. @xmath289 with @xmath290 a function space , and in which @xmath278 consists of both finite - dimensional and function parts , i.e. @xmath291 with @xmath292 where @xmath579 . for each @xmath301
we define @xmath580 as the operator @xmath581 $ ] with the understanding that @xmath582 .
let @xmath583 .
we say that @xmath584 has an envelope function @xmath585 if there exists some measurable @xmath586 such that @xmath587 for every @xmath588 and @xmath589 .
let @xmath590 . the functions in @xmath591 are clearly bounded by @xmath592 .
let @xmath593}(u,{\mathcal{m}}^*,\|\cdot\|_{p})$ ] denote the entropy with bracketing of @xmath591 with respect to the norm @xmath594 .
finally , let @xmath595 and observe that @xmath596 .
[ lem : beta:3 ] let the conditions of lemma [ lem : beta:1 ] hold for @xmath267 , and let : 1 .
@xmath584 have envelope function @xmath585 with @xmath597 for some @xmath598 2 .
@xmath599}(u,{\mathcal{m}}^*,\|\cdot\|_{\frac{4vs}{2s - v } } ) \leq { \mathrm{const } } \times u^{-2\zeta}$ ] for some @xmath600 and @xmath601 3 .
@xmath602 be gateaux differentiable at @xmath66 with @xmath603| = o(\|\alpha - \alpha_0\|^2_{{\mathcal{a}}})$ ] as @xmath315 4 .
@xmath319 and @xmath604 = o_p(1)$ ] 5 .
@xmath605 . then : assumption [ a : var ] holds , @xmath606 and we may take @xmath607 .
the following conditions are sufficient for assumption [ a : fp : var ] and for bounding the term @xmath437 in display ( [ e : nudef ] ) .
let @xmath608 .
[ lem : beta:4 ] let the following hold : 1 .
@xmath217 is strictly stationary and exponentially beta - mixing 2 .
@xmath609 < \infty$ ] for some @xmath598 3 .
@xmath610 .
then : assumption [ a : fp : var ] holds , @xmath611 and we may take @xmath612 .
here we consider the asymptotic distribution the estimator @xmath265 of the entropy of the permanent component of the sdf , namely @xmath613 $ ] . in case 1 ,
the estimator of the long - run entropy is : @xmath614 recall that @xmath269 where the influence function @xmath226 is defined in ( [ e : inf : def ] ) .
define : @xmath615\ ] ] set @xmath616 .
[ p : asydist : l:1 ] let the assumptions of theorem [ t : asydist:1 ] hold and let @xmath617 satisfy a clt , i.e. : @xmath618 where the @xmath324 matrix @xmath619 is positive definite . then : @xmath620 where @xmath621 . in the preceding proposition , @xmath622 will be the long - run variance : @xmath623 where @xmath624 .
theorem [ t : eff : main ] below shows that @xmath622 is the semiparametric efficiency bound for @xmath156 . in case 2 , the estimator of the long - run entropy is : @xmath625 as with asymptotic normality of @xmath135 , the asymptotic distribution of @xmath265 will depend on the manner in which @xmath70 was estimated . for brevity
, we just consider the parametric case studied in theorem [ t : asydist:2a ] .
let @xmath626 and @xmath627 be as previously defined with @xmath628 .
recall @xmath629 and neighborhood @xmath89 from assumption [ a : parametric ] and define : @xmath630 } = \left ( \rho^{-1 } \ ; , \ ; - { \mathbb{e } } \left[\frac{1}{m(x_t , x_{t+1},\alpha ) } \frac{\partial m(x_t , x_{t+1},\alpha)}{\partial \alpha ' } \right ] \ ; , \ ; -1 \right)'\,.\ ] ] [ p : asydist : l:2a ] let the assumptions of theorem [ t : asydist:2a ] hold . also let ( i ) there exist a neighborhood @xmath631 of @xmath66 upon which the function @xmath632 is continuously differentiable in @xmath278 for all @xmath279 with : @xmath633 < \infty\ ] ] and ( ii ) @xmath634 satisfies a clt , i.e. : @xmath635})\ ] ] where the @xmath636 matrix @xmath637}$ ] is positive definite . then : @xmath638})\ ] ] where @xmath639 } = \hbar_{[{\mathrm{2a}}]}'w_{[{\mathrm{2a}}]}^{\phantom \prime } \hbar_{[{\mathrm{2a}}]}^{\phantom \prime}$ ] . to establish efficiency , we first obtain the efficient influence function for the parameter @xmath42 by showing that @xmath42 is differentiable relative to an appropriately chosen tangent set ( see chapter 25 of @xcite for details ) .
we then show that the efficient influence function for @xmath42 is precisely the influence function @xmath226 in expressions ( [ e : ale:1])([e : inf : def ] ) .
let @xmath640 denote the @xmath1-step transition probability of @xmath20 for any borel set @xmath641 .
we say that @xmath642 is _ uniformly ergodic _ if : @xmath643 where @xmath644 denotes total variation norm and @xmath75 denotes the stationary distribution of @xmath20 .
[ a : eff ] @xmath642 is uniformly ergodic .
equivalent conditions for assumption [ a : eff ] , such as doeblin s condition , are well known ( see , e.g. , theorem 16.0.2 of @xcite ) .
assumption [ a : eff ] also implies that @xmath642 is exponentially phi - mixing , and hence also exponentially beta - mixing and exponentially rho - mixing @xcite .
this assumption is used to construct the tangent set in markov process models ; see @xcite and references therein .
[ t : eff : main ] ( i ) let assumption [ a : id ] , [ a : asydist](d ) , and [ a : eff ] hold and let @xmath645 be continuously differentiable at @xmath42 with @xmath646 .
then : the efficiency bound for @xmath647 is @xmath648 .
+ ( ii ) if , in addition ,
@xmath649 < \infty$ ] , then : the efficiency bound for @xmath156 is @xmath622 .
the following result shows that @xmath650 behaves as a linear functional of @xmath651 .
this is used to derive the asymptotic distribution of @xmath135 in theorem [ t : asydist:1 ] .
it follows from assumption [ a : var ] that we can choose sequences of positive constants @xmath652 and @xmath653 such that : @xmath654 with @xmath655 and @xmath656 as @xmath657 .
[ lem : expansion ] let assumptions [ a : id ] , [ a : bias ] and [ a : var ] hold . then : @xmath658 with @xmath124 and @xmath127 normalized so that @xmath659 and @xmath660 . in particular , if @xmath661 and @xmath237 then : @xmath662
in this appendix we discuss separately existence and identification , and compare the conditions in the present paper with the stochastic stability conditions in @xcite ( hs hereafter ) and @xcite ( bhs hereafter ) .
[ a : id:1 ] let the following hold : 1 . @xmath71 is bounded 2 .
there exists positive functions @xmath94 and a positive scalar @xmath42 such that @xmath95 solves ( [ e : pev ] ) and @xmath96 solves ( [ e : pev : star ] ) 3 .
@xmath663 is positive for each non - negative @xmath97 that is not identically zero .
* discussion of assumptions : * parts ( a ) and ( c ) are implied by the boundedness of @xmath71 and positivity of @xmath86 in assumption [ a : id:0](a ) . proposition [ p : a : exist ] below shows that the conditions in assumption [ a : id:0 ] are sufficient for existence ( part ( b ) )
. no compactness condition is assumed .
[ p : a : id ] let assumption [ a : id:1 ] hold . then : the functions @xmath43 and @xmath46 are the unique solutions ( in @xmath72 ) to ( [ e : pev ] ) and ( [ e : pev : star ] ) , respectively .
we now compare the identification results with those in hs and bhs .
some of hs s conditions related to the generator of the semigroup of conditional expectation operators @xmath664 $ ] under the change of conditional probability induced by @xmath665 , namely : @xmath666 : = { \mathbb{e } } \bigg [ \frac{m_{t+ \tau}^p}{m_t^p } \psi(x_{t+\tau } ) \bigg| x_t = x\bigg ] \,.\ ] ] in discrete - time environments , both multiplicative functionals and semigroups are indexed by non - negative integers .
therefore , the `` generator '' in discrete - time is just the single - period distorted conditional expectation operator @xmath667 $ ] .
the following are discrete - time versions of assumptions 6.1 , 7.1 , 7.2 , 7.3 , and 7.4 in hs .
[ c : hs ] 1 .
@xmath668 is a positive multiplicative functional 2 .
there exists a probability measure @xmath669 such that @xmath670\,{\mathrm{d } } \hat \varsigma(x ) = \int \psi(x)\,{\mathrm{d } } \hat \varsigma(x)\ ] ] for all bounded measurable @xmath671 3 . for any @xmath672 with @xmath673 , @xmath674 > 0\ ] ] for all @xmath675 4 . for any @xmath672 with @xmath673 , @xmath676 for all @xmath675 , where @xmath677\,{\mathrm{d } }
\hat \varsigma(x)\ ] ] for each @xmath678 .
condition [ c : hs](a ) is satisfied by construction of @xmath49 in ( [ e : pctc ] ) . for condition
[ c : hs](b ) , let @xmath43 and @xmath46 be as in assumption [ a : id:1](b ) and normalize @xmath46 such that @xmath100 = 1 $ ] .
under this normalization we can define a probability measure @xmath669 by @xmath679 $ ] for all @xmath680 .
below shows that this probability measure is precisely the measure used to define the unconditional expectation @xmath98 in the long - run approximation ( [ e : lrr ] ) . ]
recall that @xmath75 is the stationary distribution of @xmath20 .
we then have : @xmath681\,{\mathrm{d } } \hat \varsigma(x ) \\ & = \int { \mathbb{e } } \left [ \left .
\rho^{-1 } m(x_t , x_{t+1 } ) \frac{\phi(x_{t+1})}{\phi(x_t ) } \psi(x_{t+1 } ) \right| x_t = x \right ] \phi(x ) \phi^*(x)\,{\mathrm{d } } q(x ) \\ & = \rho^{-1 } { \mathbb{e } } \left [ \phi^*(x_t ) ( { \mathbb{m}}(\phi \psi ) ( x_t ) ) \right ] \\ & = \rho^{-1 } { \mathbb{e } } \left [ ( ( { \mathbb{m}}^ * \phi^*)(x_{t+1 } ) ) \phi(x_{t+1 } ) \psi(x_{t+1 } ) \right ] \\ & = { \mathbb{e}}[\phi^*(x_{t+1 } ) \phi(x_{t+1 } ) \psi(x_{t+1 } ) ] = \int \psi(x ) \,{\mathrm{d } } \hat \varsigma(x)\,.\end{aligned}\ ] ] therefore , condition [ c : hs](b ) is satisfied .
a similar derivation is reported for continuous - time semigroups in an preliminary 2005 draft of hs with @xmath75 replaced by an arbitrary measure . for condition
[ c : hs](c ) , note that @xmath673 implies @xmath682 under our construction of @xmath669 .
therefore , @xmath673 implies @xmath683 is positive on a set of positive @xmath75 measure .
moreover , by definition of @xmath98 we have : @xmath684 & = & \frac{1}{\phi(x ) } \sum_{t=1}^\infty \rho^{-t } \mathbb m_t ( \phi(\cdot ) { 1\!\mathrm{l}}\ { \cdot \in \lambda\})(x ) \\ & \geq & \frac{1}{\phi(x ) } \sum_{t=1}^\infty \lambda^{-t } \mathbb m_t(\phi(\cdot ) { 1\!\mathrm{l}}\ { \cdot \in \lambda\})(x ) \end{aligned}\ ] ] for any @xmath685 where @xmath686 denotes the spectral radius of @xmath71 .
assumption [ a : id:1](c ) implies @xmath71 is irreducible and , by definition of irreducibility , @xmath687 ( almost everywhere ) holds for @xmath688 .
therefore , assumption [ a : id:1](c ) implies condition [ c : hs](c ) , up to the `` almost everywhere '' qualification .
part ( d ) is a harris recurrence condition which does not translate clearly in terms of the operator @xmath689 .
when combined with existence of an invariant measure and irreducibility ( condition [ c : hs](b ) and ( c ) , respectively ) , it ensures both uniqueness of @xmath669 as the invariant measure for the distorted expectations as well as @xmath43-ergodicity , i.e. , @xmath690 - \int \frac{\psi(x)}{\phi(x)}\,{\mathrm{d } } \hat \varsigma(x ) \right| = 0\ ] ] ( almost everywhere ) where the supremum is taken over all measurable @xmath54 such that @xmath691 ( * ? ? ?
* proposition 14.0.1 ) . result ( [ e : hscgce ] ) is a discrete - time version of proposition 7.1 in hs , which they use to establish identification of @xmath43 .
assumption [ a : id:1 ] alone is not enough to obtain a convergence result like ( [ e : hscgce ] ) . on the other hand , the conditions in the present paper assume existence of @xmath46 whereas no positive eigenfunction of the adjoint of @xmath71 is guaranteed under the conditions in hs . to belong to a banach space ) .
] this suggests the harris recurrence condition is of a very different nature from assumption [ a : id:1 ] .
bhs assume that @xmath20 is ergodic under the @xmath692 probability measure , for which conditions [ c : hs](b)(d ) are sufficient .
also notice that condition [ c : hs](a ) is satisfied by construction in bhs .
the identification results in hs and the proof of proposition 3.3 in bhs shows that uniqueness is established in the space of functions @xmath54 for which @xmath693 $ ] is finite , where @xmath98 denotes expectation under the stationary distribution corresponding to ( [ e : twist ] ) . under assumption
[ a : id:1 ] , their result establishes identification in the space of functions @xmath54 for which @xmath694 = { \mathbb{e } } [ \psi(x_t ) \phi^*(x_t)]\ ] ] is finite .
the right - hand side is finite for all @xmath97 ( by cauchy - schwarz ) .
so in this sense the identification result in hs and bhs applies to a larger class of functions than our result .
we obtain the following existence result by replacing assumption [ a : id:1](b)(c ) by the slightly stronger quasi - compactness and positivity conditions in assumption [ a : id:0 ] .
the following result is essentially theorems 6 and 7 of @xcite.(a ) in proposition [ p : a : exist ] by the condition that @xmath71 is _ quasi - positive _
, i.e. for each non - negative @xmath54 and @xmath695 in @xmath72 that are not identically zero there exists @xmath38 such that @xmath696 . notice
that quasi - compactness also requires that @xmath697 .
assumption [ a : id:0](a ) is sufficient for these two conditions ( i.e. quasi - positivity and @xmath697 ) .
the condition @xmath697 together with power - compactness of @xmath71 ( assumption [ a : id:0](b ) ) is sufficient for quasi - compactness
. ] say that @xmath71 is _ quasi - compact _ if @xmath71 is bounded and there exists @xmath38 and a bounded linear operator @xmath698 such that @xmath699 is compact and @xmath700 .
quasi - compactness of @xmath71 is implied by assumption [ a : id:0 ] .
[ p : a : exist ] let assumption [ a : id:0](a ) hold and let @xmath71 be quasi - compact .
then : 1 .
there exists positive functions @xmath94 and a positive scalar @xmath42 such that @xmath95 solves ( [ e : pev ] ) and @xmath96 solves ( [ e : pev : star ] ) .
the functions @xmath43 and @xmath46 are the unique solutions ( in @xmath72 ) to ( [ e : pev ] ) and ( [ e : pev : star ] ) , respectively .
the eigenvalue @xmath42 is simple and isolated and it is the largest eigenvalue of @xmath71 . a similar existence result to part ( a ) was presented in a 2005 preliminary version of hs . for that result , hs assumed that @xmath686 was positive and that the ( continuous - time ) semigroup of operators had an element which was compact .
the further properties of @xmath42 that we establish in part ( c ) of proposition [ p : a : exist ] are essential to our derivation of the large - sample theory .
a similar proposition was derived under different conditions in @xcite .
hs establish existence of @xmath43 in possibly non - stationary , continuous - time environments by appealing to the theory of ergodic markov processes .
equivalent conditions for discrete - time environments are now presented and compared with our identification conditions . as with the identification conditions , we use analogues of generators and resolvents for discrete - time semigroups where appropriate .
[ c : hs : ex ] 1 .
there exists a function @xmath701 with @xmath702 and a finite constant @xmath703 such that @xmath704 for all @xmath675 2 .
there exists a measure @xmath705 on @xmath706 such that @xmath707 for any @xmath672 with @xmath708 , where @xmath709 is given by @xmath710 for @xmath711 3 .
the operators @xmath709 and @xmath712 are bounded , where @xmath712 is given by @xmath713 where @xmath714 is such that @xmath715 and @xmath716 for all @xmath717 ( @xmath718 exists by part ( b ) ) , @xmath719 , and @xmath720 . hs show that @xmath721 is a positive eigenfunction of @xmath71 under the preceding conditions ( see their lemma d.3 ) . condition [ c : hs : ex](b ) is satisfied under assumption [ a : id:0 ] with @xmath722 whenever @xmath723 . to see this ,
take @xmath672 with @xmath724 and observe that : @xmath725 ( almost everywhere ) where the first inequality is by positivity and the second is by irreducibility .
it follows that @xmath726 ( almost everywhere ) .
this verifies part ( b ) , up to the `` almost everywhere '' qualification .
on the other hand , conditions [ c : hs : ex](a)(c ) seem quite different from the conditions of proposition [ p : a : exist ] .
for instance , assumption [ a : id:0 ] does not presume existence of the function @xmath727 but imposes a quasi - compactness condition .
hs do not restrict the function space for @xmath71 ex ante so there is no notion of a bounded or power - compact operator on the space to which @xmath43 belongs .
the requirement that @xmath712 be bounded ( or the sufficient conditions for this provided in hs ) do not seem to translate clearly in terms of the operator @xmath71 . we now present a version of the long - run pricing approximation of hs that holds under our existence and identification conditions .
we impose the normalizations @xmath728 = 1 $ ] and @xmath100 = 1 $ ] and define the operator @xmath729 by : @xmath730 [ p : lr ] let assumption [ a : id:0 ] hold . then : there exists @xmath367 such that : @xmath731 as @xmath732 .
proposition [ p : lr ] is similar to proposition 7.4 in hs . proposition [ p : lr ]
establishes convergence of @xmath733 to @xmath734 , with the approximation error vanishing exponentially in the payoff horizon @xmath1 . a similar proposition ( without the rate of convergence )
was reported in a 2005 draft of hs . there , hs assumed directly that the distorted conditional expectations converged to an unconditional expectation characterized by @xmath43 , @xmath46 , and an arbitrary measure .
proposition [ p : lr ] shows that in stationary environments the unconditional expectation is characterized by @xmath43 , @xmath46 and @xmath75 .
more generally , section 7 of hs establishes long - run approximations of the form : @xmath735 \phi(x)\ ] ] as @xmath732 , where @xmath98 is defined in terms of some probability measure , say @xmath101 . by proposition [ p : lr ] ,
we obtain : @xmath736 \phi(x ) \,.\ ] ] comparing ( [ e - meas - hs ] ) and ( [ e - meas - stat ] ) , we see that the probability measure @xmath101 is given by : @xmath737 therefore , in environments for which our identification conditions hold , the expectation @xmath738 $ ] is characterized by the stationary distribution of @xmath20 and the positive eigenfunctions of @xmath71 and @xmath739 .
in this appendix we describe two approaches that may be used to extend the methodology presented in this paper to models with latent state variables .
theorem [ t : rate ] will continue to apply provided that one can construct estimators of @xmath145 and @xmath147 that satisfy assumption [ a : var ] .
formal verification of assumption [ a : var ] for the following approaches requires a nontrivial extension of the statistical theory in appendix [ ax : est : mat ] , which we defer to future research .
a first approach to dealing with models with a latent volatility state variable is via _ high - frequency proxies .
_ suppose that the state vector may be partitioned as @xmath740 where @xmath741 is observable and @xmath742 is spot volatility or integrated volatility . here we could use an estimator of @xmath742 , say @xmath743 , based on high - frequency data .
the estimators @xmath134 and @xmath133 may be formed as described in section [ s : est ] but using @xmath744 in place of @xmath58 .
a second approach is via _ filtering and smoothing .
_ fully nonparametric models with latent variables are not , in general , well identified .
nevertheless , with sufficient structure ( e.g. by specifying a semiparametric state - space representation for the unobservable components in terms of an auxiliary vector of observables ) it may be feasible to use a filter and smoother to construct an estimate of the distribution of the latent state variables given the time series of observables .
suppose that the state vector @xmath58 may be partitioned as @xmath745 where @xmath741 is observable and @xmath746 is latent . to simplify presentation
, we assume that the joint density of @xmath747 given @xmath748 factorizes as : @xmath749 where we assume that @xmath750 is unknown and @xmath751 is known up to a parameter @xmath752 .
we further assume that the econometrician observes a time series of data on an auxiliary vector @xmath753 and that the joint density for @xmath754 given @xmath755 factorizes as : @xmath756 with @xmath757 known up to some parameter @xmath758 .
note that ( [ e : latent : z ] ) describes a state - space representation in which @xmath751 is the state equation and @xmath757 is the observation equation .
finally , the sequence @xmath759 is assumed to be ( jointly ) strictly stationary and ergodic .
the econometrician observes the vector @xmath760 at dates @xmath761 . to estimate @xmath145
we could proceed as follows : 1 . calculate the maximum likelihood estimate @xmath762 of @xmath763 using the time series @xmath764 .
2 . 3 .
partition @xmath764 into @xmath765 blocks of length @xmath766 .
that is , for each @xmath767 , let @xmath768 } = ( z_{jq},\ldots , z_{(j+1)q-1})$ ] . fix a prior distribution @xmath769 for each @xmath770 , @xmath771 , such as the unconditional distribution for @xmath770 implied by @xmath772 .
for each @xmath771 and for each @xmath773 we compute the posterior density @xmath774};\hat \theta,\pi)$ ] of @xmath775 given @xmath768}$ ] , @xmath762 and @xmath769 .
4 . 5 . let @xmath776 be a @xmath777-vector of basis functions for @xmath778 and let @xmath779 be a @xmath780-vector of basis functions for @xmath781 .
we form basis functions for @xmath782 by taking the tensor product @xmath783 with @xmath784 .
the estimator of @xmath145 is : @xmath785 } \\ { \widehat{{\mathbf{g}}}}_{[j ] } & = \frac{1}{q } \sum_{t = jq}^{(j+1)q-1 } \left ( \big ( b_1^{k_1}(y_t)b_1^{k_1}(y_t ) ' \big ) \otimes { { \mathbb{e}}}_{\hat \theta,\pi } \big [ b_2^{k_2}(\zeta_t)b_2^{k_2}(\zeta_t ) ' \big|\vec z_{[j ] } \big ] \right ) \end{aligned}\ ] ] + where @xmath786 } \big]$ ] denotes expectation under the posterior @xmath787 } ; \hat \theta,\pi)$ ] . to gain some intuition for this estimator , suppose that the true @xmath67 is known .
then : @xmath788 } \big ] \approx \frac{1}{q } \sum_{t = jq}^{(j+1)q-1 } { { \mathbb{e } } } \big [ b_2^{k_2}(\zeta_t)b_2^{k_2}(\zeta_t ) ' \big|\vec z_{[j ] } \big]\ ] ] where `` @xmath789 '' is because , as the block length @xmath790 increases , the posterior distribution will typically become independent of the prior @xmath769 and the posterior expectation should approach the true conditional expectation . then for each @xmath791
we have : @xmath792 } \big ] & = \frac{1}{q } \sum_{t = jq}^{(j+1)q-1 } { \mathbb{e}}\big [ b^k(x_t ) b^k(x_t ) ' \big| \vec x^{obs}_{[j ] } \big]\end{aligned}\ ] ] where @xmath793 } = ( \vec x_{jq}^{obs},\ldots,\vec x_{(j+1)q-1}^{obs})$ ] . but notice that the term on the right - hand side is a @xmath119 random matrix which is a function of @xmath793}$ ] and whose expected value is @xmath145 ( by iterated expectations ) .
therefore , the estimator @xmath134 is approximately the average of @xmath794 such random matrices .
as the sequence @xmath795}^{obs},\ldots,\vec x_{[n]}^{obs}$ ] is strictly stationary and ergodic , its average ought to approach @xmath145 as @xmath89 increases .
when @xmath294 is unknown but the mle @xmath762 is consistent , the same intuition goes through except it would also have to be established that : @xmath796 } \big ] - { { \mathbb{e}}}_{\theta_0,\pi } \big [ b_2^{k_2}(\zeta_t)b_2^{k_2}(\zeta_t ) ' \big|\vec z_{[j ] } \big]\ ] ] becomes negligible ( in an appropriate sense ) as @xmath797 . to estimate @xmath147 we proceed similarly : 1 . with @xmath762 and @xmath769 as before ,
for each @xmath771 and each @xmath798 we compute the posterior density @xmath799};\hat \theta,\pi)$ ] of @xmath800 given @xmath768}$ ] , @xmath762 and @xmath769 .
2 . 3 .
when the function @xmath40 is known , the estimator of @xmath147 is : @xmath801 } \\ { \widehat{{\mathbf{m}}}}_{[j ] } & = \frac{1}{q-1 } \sum_{t = jq}^{(j+1)q-2 } \bigg ( \big ( b_1^{k_1}(y_t)b_1^{k_1}(y_{t+1 } ) ' \big ) \\ & \quad \quad \quad \quad \quad \otimes \bigg ( \int b_2^{k_2}(\zeta_t)b_2^{k_2}(\zeta_{t+1})'m(y_t,\zeta_t , y_{t+1},\zeta_{t+1 } ) f(\zeta_t;\zeta_{t+1}|
\vec z_{[j]},\hat \theta,\pi ) { \mathrm{d } } \zeta_{t+1 } { \mathrm{d } } \zeta_t \bigg ) \bigg ) \end{aligned}\ ] ] when @xmath40 contains an estimated component @xmath70 we replace @xmath802 in the above display by @xmath803 .
for any vector @xmath804 , define : @xmath805 or equivalently @xmath806 . for any matrix @xmath807
we define : @xmath808 we also define the inner product weighted by @xmath120 , namely @xmath809 .
the inner product @xmath810 and its norm @xmath811 are germane for studying convergence of the matrix estimators , as @xmath812 is isometrically isomorphic to @xmath813 .
first note that : @xmath814 where the second line is by assumption [ a : asydist](a ) and the third line is by lemma [ lem : expansion ] and assumption [ a : asydist](b ) ( under the normalizations @xmath659 and @xmath660 ) . by identity ,
we may write the first term on the right - hand side of display ( [ a : asydist : pf:1 ] ) as : @xmath815 where the second line is by assumption [ a : asydist](c ) . we verify the conditions of theorem [ t : asydist:1 ] .
since the data are exponentially beta - mixing , @xmath250 , and @xmath40 has finite @xmath718th moment , lemma [ lem : beta:1 ] implies that @xmath816 and @xmath817 .
both terms are @xmath818 under condition ( vi ) , verifying assumptions [ a : var ] and [ a : asydist](b ) . part ( a ) of assumption [ a : asydist ] is satisfied by conditions ( iii ) and ( vi ) and part ( d ) is satisfied by condition ( iv ) .
finally , for part ( c ) , by stationarity and the triangle inequality we have : @xmath819 & \leq \sqrt n { \mathbb{e } } [ | \psi_{\rho , k}(x_t , x_{t+1 } ) - \psi_\rho(x_t , x_{t+1 } ) | ] \notag \\ & \leq \sqrt n \big ( | \rho_k - \rho| \times { \mathbb{e } } [ \phi ^ * ( x_t ) \phi(x_t ) ] + \rho_k \times { \mathbb{e}}\big [ | \phi^*(x_t)\phi(x_t ) - \phi^*_k(x_t ) \phi_k^{\phantom * } ( x_t)|\big ] \big ) \notag \\ & \quad + \sqrt n \big ( { \mathbb{e}}\big [ m(x_t , x_{t+1 } ) | \phi^*(x_t)\phi(x_{t+1 } ) - \phi^*_k(x_t ) \phi_k^{\phantom * } ( x_{t+1})|\big ] \big)\ , .
\label{e : corrnorm:0}\end{aligned}\ ] ] for the leading term , lemma [ lem : bias](a ) and the conditions on @xmath182 and @xmath114 yield : @xmath820 & = \sqrt n \times o(\delta_k ) \times 1 \notag \\ & = \sqrt n \times k^{-\omega } \times 1 = o(1 ) \label{e : corrnorm:1 } \,.\end{aligned}\ ] ] for the second term , the cauchy - schwarz inequality , lemma [ lem : bias](b)(c ) and the conditions on @xmath178 , @xmath179 and @xmath114 yield : @xmath821 & \leq \sqrt n \times \rho_k \times \big ( \| \phi^ * - \phi_k^*\| \|\phi\| + \|\phi_k^*\| \| \phi - \phi_k\| \big ) \notag \\ & = \sqrt n \times o(1 ) \times o ( \delta_k^ * + \delta_k^{\phantom * } ) \notag \\ & = \sqrt n \times o(1 ) \times o ( k^{-\omega } ) = o(1 ) \label{e : corrnorm:2}\,.\end{aligned}\ ] ] we split the final term in ( [ e : corrnorm:0 ] ) in two , to obtain : @xmath822 \big ) \\ & \quad \quad \leq \sqrt
n \big ( { \mathbb{e } } [ m(x_t , x_{t+1 } ) | \phi(x_{t+1 } ) - \phi_k(x_{t+1})| \phi^*(x_{t+1 } ) ] \\ & \quad \quad \quad + { \mathbb{e } } [ m(x_t , x_{t+1 } ) | \phi^*(x_t ) - \phi_k^*(x_t)| |\phi_k(x_{t+1})| ] \big ) \,.\end{aligned}\ ] ] using the cauchy - schwarz inequality and lemma [ lem : bias](b)(c ) as above , we may deduce that these terms are @xmath823 and @xmath824 provided that @xmath825 < \infty$ ] ( which holds under the moment conditions on @xmath46 and @xmath40 ) and @xmath826 < \infty$ ] .
this latter condition holds under the moment conditions on @xmath43 and @xmath40 because : @xmath827 \\ & \leq 2{\mathbb{e}}[m(x_t , x_{t+1})^2 \phi(x_{t+1})^2 ] + 2{\mathbb{e}}[m(x_t , x_{t+1})^2 ( \phi(x_{t+1 } ) - \phi_k(x_{t+1}))^2 ] \\ & \leq 2{\mathbb{e}}[m(x_t , x_{t+1})^2 \phi(x_{t+1})^2 ] + 2{\mathbb{e}}[m(x_t , x_{t+1})^s]^{2/s } \times \|\phi - \phi_k\|_{\frac{2s}{s-2 } } ^2\end{aligned}\ ] ] and @xmath258 . therefore : @xmath828 \big ) = o ( \sqrt n \times k^{-\omega } ) = o(1 ) \label{e : corrnorm:3}\ ] ] under the conditions on @xmath114 . substituting ( [ e : corrnorm:1 ] ) , ( [ e : corrnorm:2 ] ) , and ( [ e : corrnorm:3 ] ) , into ( [ e : corrnorm:0 ] ) and using markov s inequality yields @xmath238 , as required . as in the proof of theorem [ t : asydist:1 ] , we have : @xmath829 by adding and subtracting terms and using assumption [ a : asydist](c ) , we obtain : @xmath830 let @xmath831 , @xmath832 , @xmath833 , @xmath834 , and @xmath835 .
we decompose the second term on the right - hand side of ( [ e:2a : pf1 ] ) as : @xmath836 for term @xmath837 , we know that @xmath838 wpa1 .
whenever @xmath838 we may take a mean value expansion to obtain : @xmath839 wpa1 , where @xmath840 is in the segment between @xmath70 and @xmath66 . since @xmath841 , it suffices to show that : @xmath842 let @xmath843 be a compact neighborhood of @xmath66 .
assumption [ a : parametric](c)(d ) implies that : @xmath844 < \infty\ ] ] and so by dominated convergence we may deduce that the @xmath845-valued function : @xmath846\ ] ] is continuous at @xmath66 .
it is also straightforward to show that : @xmath847 \right| = o_p(1)\ ] ] by ( [ e : moments ] ) , the ergodic theorem , and compactness of @xmath631 .
therefore , @xmath848 . for @xmath849 , using the fact that @xmath850 and root-@xmath1 consistency of @xmath70 , it follows that for any @xmath851 we have : @xmath852\end{aligned}\ ] ] where the second - last line is by the union bound and cauchy - schwarz , and the final line is by markov s inequality and assumption [ a : parametric](c ) and where the @xmath193 term is independent of @xmath853 .
therefore @xmath854 and hence : @xmath855 where the second line is by markov s inequality the cauchy - schwarz inequality and the third is by lemma [ lem : bias](b)(c ) .
it follows by assumptions [ a : asydist](a ) and [ a : parametric](d ) that @xmath856 .
substituting into ( [ e:2a : pf1 ] ) and using assumption [ a : parametric](a ) : @xmath859 } ' \bigg ( \begin{array}{c } \psi_{\rho , t } \\
\psi_{\alpha , t } \end{array }
\bigg ) + o_p(1)\ ] ] and the result follows by assumption [ a : parametric](b ) . by arguments to the proof of theorem [ t : asydist:1 ] and [ t : asydist:2a ] and using assumption [ a : nonpara](a)(b ) , we may deduce : @xmath860 + t_{1,n } + t_{2,n } + o_p(1)\end{aligned}\ ] ] where : @xmath861 the result will follow by assumption [ a : nonpara](b)(c ) provided we can show that the terms @xmath837 and @xmath849 are both @xmath241 .
similar arguments to the proof of corollary [ c : inf:1 ] imply that @xmath848 under assumption [ a : nonpara](e ) . for term @xmath849 ,
notice that : @xmath862 where @xmath305 denotes the centered empirical process on @xmath306 .
let @xmath863 denote the norm defined on p. 400 of @xcite .
display ( a ) in lemma 2 of @xcite shows that , under exponential beta - mixing ( condition ( d ) ) , the norm @xmath863 is dominated by the @xmath247 norm for any @xmath864 .
it follows from condition ( e ) that @xmath865 for some @xmath321 .
therefore , @xmath306 is uniformly bounded under @xmath863 and @xmath310 is well defined on @xmath306 since @xmath866 @xcite .
it follows by condition ( b ) that @xmath867 . appropriately modifying the arguments of lemma 19.24 in @xcite (
i.e. replacing the @xmath72 norm by the norm induced by @xmath310 which is the appropriate norm for the weakly dependent case ) give @xmath868 , as required .
take @xmath869 , @xmath870 , and @xmath89 as in lemma [ lem : fpiter ] .
fix @xmath436 and consider @xmath871 . as @xmath872 , choose @xmath853 sufficiently small that @xmath873 .
( @xmath874 is the neighborhood in the statement of the proposition . )
take any @xmath875 .
for any such @xmath54 we can write @xmath876 where @xmath877 . write @xmath878 . by homogeneity of @xmath362 : @xmath879 for each @xmath375 , where @xmath880 ( note positivity of @xmath391 ensures that @xmath881 for each @xmath1 and each @xmath882 ) .
it follows from lemma [ lem : fpiter ] that : @xmath883 as required .
[ lem : fpiter ] let the conditions of proposition [ p : nl ] hold .
then : there exists finite positive constants @xmath399 with @xmath400 and a neighborhood @xmath89 of @xmath295 such that : @xmath884 for all @xmath885 .
fix some constant @xmath886 such that @xmath887 . by the gelfand formula
, there exists @xmath888 such that @xmath889 .
frchet differentiability of @xmath362 at @xmath295 together with the chain rule for frchet derivatives implies that : @xmath890 hence : @xmath891 we may therefore choose @xmath321 such that @xmath892 for all @xmath893 where @xmath894 .
then for any @xmath893 and any @xmath895 we have : @xmath896 it is straightforward to show via induction that boundedness of @xmath391 and homogeneity of degree @xmath59 of @xmath362 together imply : @xmath897 for any @xmath898 . the result for @xmath364 is stated in the text .
for @xmath295 , let @xmath869 , @xmath870 , and @xmath89 be as in lemma [ lem : fpiter ] .
suppose @xmath902 is a fixed point of @xmath362 belonging to @xmath89 .
then by lemma [ lem : fpiter ] : @xmath903 hence @xmath904 .
first note that @xmath42 is a simple isolated eigenvalue of @xmath71 under assumption [ a : id ] .
therefore , there exists an @xmath321 such that @xmath906 for all @xmath907 .
let @xmath310 denote a positively oriented circle in @xmath908 centered at @xmath42 with radius @xmath909 .
let @xmath910 denote the resolvent of @xmath71 evaluated at @xmath911 , where @xmath912 is the identity operator .
note that the number @xmath913 , given by : @xmath914 because @xmath915 is a holomorphic function on @xmath310 and @xmath310 is compact .
by assumption [ a : bias ] , there exists @xmath512 such that @xmath916 for all @xmath917 .
therefore , for all @xmath513 the inequality : @xmath918 holds .
it follows by theorem iv.3.18 on p. 214 of @xcite that whenever @xmath513 : ( i ) the operator @xmath919 has precisely one eigenvalue @xmath110 inside @xmath310 and @xmath110 is simple ; ( ii ) @xmath920 ; and ( iii ) @xmath921 lies on the exterior of @xmath310 .
note that @xmath110 must be real whenever @xmath513 because complex eigenvalues come in conjugate pairs .
thus , if @xmath110 were complex - valued then its conjugate would also be in @xmath310 , which would contradict the fact that @xmath110 is the unique eigenvalue of @xmath919 on the interior of @xmath310 .
step 2 : any nonzero eigenvalue of @xmath919 is also a nonzero eigenvalue of @xmath922 with the same multiplicity .
so by step 1 we know that the largest eigenvalue @xmath110 of @xmath922 is positive and simple whenever @xmath513 .
recall that @xmath923 where @xmath127 solves the left - eigenvector problem in ( [ e : gev ] ) . here , @xmath204 is the eigenfunction corresponding to @xmath110 of the adjoint of @xmath919 with respect to the space @xmath108 .
we now introduce the term @xmath924 , which is the eigenfunction corresponding to @xmath110 of the adjoint of @xmath919 with respect to the space @xmath72 ( it follows from lemma [ lem : exist ] that @xmath204 and @xmath924 are uniquely defined ( up to scale ) for all @xmath114 sufficiently large under assumptions [ a : id ] and [ a : bias ] ) . that is : @xmath925 & = \rho_k { \mathbb{e}}[\phi_k^+(x ) \psi(x ) ] \\ { \mathbb{e}}[\phi_k^*(x ) \pi_k { \mathbb{m } } \psi_k(x ) ] & = \rho_k { \mathbb{e}}[\phi_k^*(x ) \psi_k(x)]\end{aligned}\ ] ] for all @xmath97 and @xmath926 .
notice that @xmath924 does not necessarily belong to @xmath108 whereas @xmath204 does belong to @xmath108 .
however , it follows from the preceding display that @xmath927 .
the proof of lemma [ lem : bias ] shows that @xmath204 and @xmath924 converge to @xmath46 at the same rate .
step 1 : proof of part ( b ) .
we use similar arguments to the proof of proposition 4.2 of @xcite .
take @xmath917 ( where @xmath928 is from lemma [ lem : exist ] ) and let @xmath929 denote the spectral projection of @xmath919 corresponding to @xmath110 .
by lemma 6.4 on p. 279 of @xcite , we have : @xmath930 where @xmath310 is defined in the proof of lemma [ lem : exist ] .
it follows that : @xmath931 moreover , for each @xmath932 we have : @xmath933 where the first inequality is by theorem iv.3.17 on p. 214 of @xcite and the second is by display ( [ e : crdef ] ) .
this inequality holds uniformly for @xmath932 . substituting ( [ e : resbd ] ) into ( [ e : projbd ] ) : @xmath934 where the final equality is by definition of @xmath182 in display ( [ e : deltas ] ) .
since @xmath110 is simple , the spectral projection @xmath929 is of the form : under the normalizations @xmath191 and @xmath935 . here
we are imposing the normalizations @xmath191 and @xmath936 and and so we scale the definition of the projection accordingly . ]
@xmath937 under the normalizations @xmath191 and @xmath936 . by the proof of proposition 4.2 of @xcite , under the sign normalization @xmath938 we have
: @xmath939 whence @xmath940 by ( [ e : resbd:00 ] ) , proving ( b ) .
step 2 : proof of part ( a ) .
here we use similar arguments to the proof of corollary 4.3 of @xcite . by the triangle inequality : @xmath941 because @xmath942 by part ( b ) , @xmath943 by definition of @xmath182 , and @xmath944 because @xmath71 is bounded and @xmath945 is a ( weak ) contraction on @xmath72 .
first observe that @xmath948 holds for all @xmath932 ( where @xmath949 denotes the conjugate of @xmath950 ) because @xmath951 ( * ? ? ?
* theorem 6.22 , p. 184 ) .
similarly , @xmath952 holds for all @xmath932 whenever @xmath513 . by similar arguments to the proof of part ( b ) : @xmath953 where the final equality is by definition of @xmath179 ( see display ( [ e : deltas ] ) ) .
now we use the fact that @xmath954 is of the form : @xmath955 under the normalizations @xmath191 and @xmath936 . by similar arguments to the proof of part ( b ) , we have : @xmath956 under the sign normalization @xmath957 .
it follows by ( [ e : projbdstar ] ) that @xmath958 .
step 4 : proof that @xmath207 .
recall that @xmath959 . then by the triangle inequality and the fact that @xmath945 is a weak contraction
, we have : @xmath960 where the final equality is by definition of @xmath179 ( see display ( [ e : deltas ] ) ) and step 3 .
if @xmath134 is invertible we have : @xmath964 part ( b ) follows by the triangle inequality , noting that @xmath965 whenever @xmath962 . substituting @xmath966 into the preceding display yields : @xmath967 as required .
step 1 : let @xmath968 denote the restriction of @xmath919 to @xmath108 ( i.e. @xmath968 is a linear operator on the space @xmath108 rather than the full space @xmath72 ) .
we show that : @xmath969 holds for all @xmath970 .
fix any such @xmath950 .
then for any @xmath926 we have @xmath971 where @xmath972 is given by @xmath973 .
similarly , for any @xmath97 we have @xmath974 where @xmath975 is given by @xmath976 .
therefore , for any @xmath926 we must have @xmath977 , i.e. , @xmath978 holds for all @xmath926 . therefore : @xmath979 step 2 : we now show that @xmath136 has a unique eigenvalue @xmath135 inside @xmath310 wpa1 , where @xmath310 is from the proof of lemma [ lem : exist ] . as the nonzero eigenvalues of @xmath919 , @xmath968 , and @xmath922 are the same , it follows from the proof of lemma [ lem : exist ] that for all @xmath513 the curve @xmath310 separates @xmath110 from @xmath980 and @xmath110 from @xmath981 . taking @xmath917 from lemma [ lem :
exist ] , by step 1 and display ( [ e : resbd ] ) , we have : @xmath982 recall that @xmath983 is isomorphic to @xmath984 on @xmath812 .
let @xmath985 denote the resolvent of @xmath986 on the space @xmath812 . by isometry and display ( [ e : resbd
: proj ] ) : @xmath987 by lemma [ lem : matcgce](b ) , assumption [ a : var ] , and boundedness of @xmath71 , we have : @xmath988 hence : @xmath989 holds wpa1 .
the proof of lemma [ lem : exist ] shows that @xmath986 has a unique eigenvalue @xmath110 inside @xmath310 whenever @xmath513 and that @xmath110 is a simple eigenvalue of @xmath986 .
therefore , whenever ( [ e : kato ] ) holds we have that : ( i ) @xmath990 has precisely one simple eigenvalue @xmath135 inside @xmath310 ; ( ii ) @xmath991 ; and ( iii ) @xmath992 lies on the exterior of @xmath310 ( see theorem iv.3.18 on p. 214 of @xcite ) .
again , @xmath135 must be real and simple whenever ( [ e : kato ] ) holds ( because complex eigenvalues come in conjugate pairs ) hence the corresponding left- and right - eigenvectors @xmath138 and @xmath137 are unique .
step 1 : proof of part ( b ) .
let @xmath993 denote the spectral projection of @xmath990 corresponding to the eigenvalue @xmath994 . by similar arguments to the proof of lemma [ lem : bias](b ) , @xmath995 and
so : @xmath996 note that @xmath997 because @xmath310 is centered at @xmath42 and @xmath998 by lemma [ lem : bias ] .
further , whenever ( [ e : kato ] ) holds , for each @xmath932 we have : @xmath999 which is @xmath1000 by assumption [ a : var ] and display ( [ e : resbd : k ] ) .
substituting into ( [ e : hatbd ] ) and using the definition of @xmath183 ( cf . display ( [ e : etas ] ) ) yields : @xmath1001 let @xmath1002 .
observe that @xmath993 is given by : @xmath1003 under the normalizations @xmath1004 and @xmath1005 . by similar arguments to ( [ e : oblique ] ) , we obtain : @xmath1006 under the sign normalization @xmath1007 .
it follows that @xmath1008 by ( [ e : ckhatbd ] ) .
step 2 : proof of part ( a ) .
similar arguments to the proof of lemma [ lem : bias](a ) yield : @xmath1009 the second term on the right - hand side of ( [ e : rhatineq ] ) is @xmath1010 ( cf . display ( [ e : etas ] ) ) . for the first term
, assumption [ a : var ] implies that @xmath1011 . but @xmath1012 .
therefore , @xmath1013 and the result follows by ( [ e : ckineq ] ) .
step 3 : proof of part ( c ) .
identical arguments to the proof of part ( b ) yield : @xmath1014 where : @xmath1015 and @xmath181 is from display ( [ e : etas ] ) .
therefore , under the sign normalization @xmath1016 we have : @xmath1017 whence @xmath1018 by ( [ e : projbdstarhat ] ) .
we verify the conditions of theorem 19.1 in @xcite .
we verify the conditions noting that , in our notation , @xmath1019 is a neighborhood of @xmath295 , @xmath1020 , @xmath1021 , and @xmath1022 ( the restriction of @xmath1023 to @xmath108 ) .
@xmath362 is continuously frchet differentiable on a neighborhood @xmath89 of @xmath295 by assumption [ a : fp : exist](c ) . as a consequence , @xmath1024 is continuously frchet differentiable on @xmath1025 .
assumption [ a : fp : exist](b)(c ) implies that @xmath1026 is continuously invertible in @xmath72 .
also notice that @xmath1027 by assumption [ a : fp : bias](b ) ( this verifies condition ( 19.8 ) in @xcite ) .
moreover : @xmath1028 by continuity of @xmath362 .
moreover , @xmath1029 by continuous frchet differentiability of @xmath362 and assumption [ a : fp : bias](a ) .
this verifies conditions ( 19.9 ) in @xcite , and their condition ( 19.10 ) is trivially satisfied . fix any @xmath321 . by continuous frchet differentiability , there exists @xmath1030 such that @xmath1031 whenever @xmath1032
. we may also choose @xmath1033 such that @xmath1034 and @xmath1035 for all @xmath1036 .
whenever @xmath1036 and @xmath1037 we have : @xmath1038 and hence : @xmath1039 as required .
theorem 19.1 in @xcite then ensures ( 1 ) existence of @xmath414 and uniqueness as a fixed point of @xmath1023 on a neighborhood of @xmath295 for all sufficiently large @xmath114 and ( 2 ) the upper bound : @xmath1040 ( see equations ( 19.12)(19.13 ) on p. 295 ) .
it follows by continuity of @xmath945 and @xmath362 and assumption [ a : fp : bias](b ) that @xmath1041 .
we first prove part ( c ) by standard arguments ( see p. 310 in @xcite ) .
take @xmath513 from lemma [ lem : fp : exist ] .
we then have : @xmath1042 frchet differentiability of @xmath362 at @xmath295 ( assumption [ a : fp : exist](c ) ) implies : @xmath1043 as @xmath398 and @xmath1044 ( assumptions [ a : fp : exist](c ) and [ a : fp : bias](a ) ) , a similar argument to lemma [ lem : exist ] ensures that there exists @xmath321 such that @xmath1045 for all @xmath114 sufficiently large . therefore @xmath1046 is invertible and @xmath1047 holds for all @xmath114 sufficiently large . substituting ( [ e : bias : fp:1 ] ) and
( [ e : bias : fp:2 ] ) into ( [ e : bias : fp:0 ] ) yields : @xmath1048 hence @xmath1049 .
part ( b ) then follows from the inequality : @xmath1050 finally , part ( a ) follows from the fact that @xmath1051 and continuous differentiability of @xmath1052 at each @xmath1053 . by definition of @xmath176 , @xmath1056 , and @xmath1057 , we have @xmath1058 as @xmath965 holds wpa1 ( by the first part of assumption [ a : fp : var ] )
, we have : @xmath1059 wpa1 , hence : @xmath1060 wpa1 .
the result follows by assumption [ a : fp : var ] .
lemma [ lem : fp : exist ] implies that @xmath1023 has a unique fixed point @xmath414 in @xmath1067 and no fixed point on @xmath1068 for all @xmath513 .
it also follows from the proof of lemma [ lem : fp : exist ] that @xmath1069 is invertible for all @xmath513 ( increasing @xmath928 if necessary ) .
let @xmath1070 denote the rotation of @xmath1071 on @xmath1068 .
then for all @xmath513 we have @xmath1072 and hence @xmath1073 by propositions ( 3)(5 ) on pp .
299 - 300 of @xcite .
we now show that the inequality : @xmath1074 holds wpa1 .
the left - hand side is @xmath241 by lemma [ lem : fp : matcgce ] and assumption [ a : fp : var ] .
we claim that @xmath1075 .
suppose the claim is false .
then there exists a subsequence @xmath1076 with @xmath1077 such that @xmath1078 .
since @xmath1076 is bounded and @xmath362 is compact , there exists a convergent subsequence @xmath1079 .
let @xmath1080 . then : @xmath1081 as @xmath1082 , where the first term vanishes by definition of @xmath1083 , the second vanishes by definition of @xmath1084 , and the third vanishes because @xmath108 is dense in the range of @xmath362 .
it follows that @xmath1085 .
finally , by continuity of @xmath362 and definition of @xmath695 : @xmath1086 as @xmath1082 , hence @xmath1085 is a fixed point of @xmath362 . but
this contradicts the fact that @xmath295 is the unique fixed point of @xmath362 in @xmath310 .
this proves the claim .
it follows by proposition ( 2 ) on p. 299 of @xcite that whenever inequality ( [ e : fpindex : ineq ] ) holds , we have : @xmath1087 by isomorphism : @xmath1088 hence @xmath1089 holds wpa1 . therefore , @xmath1090 has at least one fixed point @xmath1091 wpa1 .
clearly , for @xmath1092 we must have @xmath1093 . repeating
the argument for any @xmath1094 implies that @xmath1095 wpa1 .
therefore @xmath541 .
we first prove part ( c ) . by lemma [ lem : fphat : exist ] , @xmath540 is well defined ( wpa1 ) and @xmath1099 . recall that @xmath1100 and @xmath1101 .
then wpa1 , we have : @xmath1102 so by lemma [ lem : fp : matcgce](b ) and the triangle inequality : @xmath1103 notice that : @xmath1104 moreover , @xmath1026 is continuously invertible by assumption [ a : fp : exist](b ) .
therefore : @xmath1105 holds for all @xmath114 sufficiently large .
also notice that : @xmath1106 where the first inequality is because @xmath945 is a ( weak ) contraction on @xmath72 , the second is by the triangle inequality , and the final line is by frechet differentiability of @xmath362 at @xmath295 ( first and third terms ) and continuity of @xmath1107 at @xmath295 ( second term ) . substituting ( [ e : var : fp:1 ] ) and ( [ e : var : fp:2 ] ) into ( [ e : var : fp:0 ] ) and rearranging ,
we obtain : @xmath1108 as required .
parts ( a ) and ( b ) follow by similar arguments to the proof of lemma [ lem : bias : fp ] . for part ( a ) we use a truncation argument .
let @xmath1115 be a sequence of positive constants to be defined subsequently , and write : @xmath1116 where , with @xmath1117 and @xmath1118 denoting the indicator function , we have : @xmath1119 \\
\xi_{2,t , n } & = & n^{-1 } \tilde b^k(x_t ) m(x_t , x_{t+1 } ) \tilde b^k(x_{t+1 } ) ' { 1\!\mathrm{l}}\ { \|\tilde b^k(x_t ) m(x_t , x_{t+1 } ) \tilde b^k(x_{t+1})'\| > t_n\ } \\ & & - { \mathbb{e}}[n^{-1 } \tilde b^k(x_t ) m(x_t , x_{t+1 } ) \tilde b^k(x_{t+1 } ) ' { 1\!\mathrm{l}}\ { \|\tilde b^k(x_t ) m(x_t , x_{t+1 } ) \tilde b^k(x_{t+1})'\| > t_n\ } ] \,.\end{aligned}\ ] ] note @xmath1120 = 0 $ ] and @xmath1121 by construction .
let @xmath1122 .
for any @xmath1123 and any @xmath1124 , we have : @xmath1125v| & \leq n^{-2 } \xi_k^2 { \mathbb{e}}[|u'\tilde b^k(x_t ) m(x_t , x_{t+1 } ) m(x_s , x_{s+1 } ) \tilde b^k(x_s)'v| ] \\ & \leq n^{-2 } \xi_k^2 { \mathbb{e}}[|m(x_t , x_{t+1})|^r]^{1/r } { \mathbb{e}}[|m(x_s , x_{s+1})|^r]^{1/r } \\ & \quad \quad \times { \mathbb{e}}[|(u'\tilde b^k(x_t))|^q]^{1/q } { \mathbb{e}}[|(v'\tilde b^k(x_s))|^q]^{1/q } \end{aligned}\ ] ] by hlder s inequality with @xmath790 chosen such that @xmath1126 .
using the fact that @xmath1127 and @xmath1128 = \|u\|^2 = 1 $ ] ( and similarly for @xmath352 ) , we obtain : @xmath1129^{1/q } \leq ( \xi_k^{q-2 } { \mathbb{e}}[(u'\tilde b^k(x_t))^2])^{1/q } = \xi_k^{1 - 2/q}\ ] ] and hence : @xmath1130v| \leq \frac{\xi_k^{2 + 4/r } } { n^2 } { \mathbb{e}}[|m(x_0,x_1)|^r]^{2/r}\,.\ ] ] as this bound holds uniformly for @xmath1123 , it follows from the variational characterization of the operator norm that : @xmath1131\| = \sup_{u , v \in s^{k-1}}|u'{\mathbb{e } } [ \xi_{1,t , n}^{\phantom \prime } \xi_{1,s , n } ' ] v| = o(\xi_k^{2 + 4/r}/n^2)\,.\ ] ] this bound holds uniformly for @xmath1124 , and also holds for @xmath1132\|$ ] .
applying corollary 4.2 of @xcite yields : @xmath1133 provided @xmath1134 .
now consider the remaining term .
if @xmath1135 then we can set @xmath1136 by taking @xmath1137 for sufficiently large @xmath869 .
now suppose @xmath1138 . by the triangle and jensen inequalities : @xmath1139 \\ &
\leq \frac{2}{n } \sum_{t=0}^{n-1 } { \mathbb{e } } [ \| \tilde b^k(x_t ) m(x_t , x_{t+1 } ) \tilde b^k(x_{t+1 } ) ' \|{1\!\mathrm{l}}\ { \|\tilde b^k(x_t ) m(x_t , x_{t+1 } ) \tilde b^k(x_{t+1})'\| > t_n\ } ] \\ & \leq \frac{2}{n t_n^{r-1 } } \sum_{t=0}^{n-1 } { \mathbb{e } } [ \| \tilde b^k(x_t ) m(x_t , x_{t+1 } ) \tilde b^k(x_{t+1 } ) ' \|^r { 1\!\mathrm{l}}\ { \|\tilde b^k(x_t ) m(x_t , x_{t+1 } ) \tilde b^k(x_{t+1})'\| > t_n\ } ] \\ & \leq \frac{2\xi_k^{2r}}{t_n^{r-1 } } { \mathbb{e}}[|m(x_0,x_1)|^r]\,.\end{aligned}\ ] ] so by markov s inequality : @xmath1140 we choose @xmath1141
so that : @xmath1142 so that @xmath1143
. solving the above display for @xmath1141 , we obtain @xmath1144 .
the condition @xmath1134 is , with this choice of @xmath1141 , equivalent to @xmath1145 .
this condition is therefore implied by the conditions @xmath1146 and @xmath1147 .
we use similar arguments to lemmas 4.8 and 4.9 of @xcite . for part ( b ) , by the covariance inequality for exponentially rho - mixing processes , there is a finite positive constant @xmath869 ( depending on the rho - mixing coefficients ) such that : @xmath1154 \leq \frac{c}{n } \sum_{l=1}^k { \mathbb{e } } \left [ \tilde b_{kl}(x_t)^2 m(x_t , x_{t+1})^2 ( \tilde b^k(x_{t+1})'\tilde c_k)^2 \right]\,.\ ] ] it follows by definition of @xmath1155 , cauchy - schwarz , and hlder s inequality that : @xmath1156 & \leq \frac{c \xi_k^2}{n } { \mathbb{e}}[|m(x_0,x_1)|^r]^{2/r } { \mathbb{e } } [ ( \tilde b^k(x_0 ) ' \tilde c_k)^{2r/(r-2)}]^{(r-2)/r } \notag \\ & \leq \frac{c \xi_k^{(2r+4)/r}}{n } { \mathbb{e}}[|m(x_0,x_1)|^r]^{2/r } \label{e - mhatvec}\end{aligned}\ ] ] where the final line is because @xmath189 and @xmath1157 = \|\tilde c_k\|^2 = 1 $ ] .
this proves @xmath1152 .
the result for @xmath1158 follows similarly . for part ( a ) ,
let @xmath1159 be an orthonormal basis for @xmath1160 .
using the fact that the frobenius norm @xmath1161 dominates the @xmath72 norm , we have : @xmath1162 & \leq & { \mathbb{e } } [ \|{\widehat{{\mathbf{m}}}}^o - { \mathbf{m}}^o\|^2_f ] \\ & = & \sum_{l=1}^k { \mathbb{e } } [ \|({\widehat{{\mathbf{m}}}}^o - { \mathbf{m}}^o)u_l\|^2 ] \\ & \leq & \frac{ck \xi_k^{(2r+4)/r}}{n } { \mathbb{e}}[|m(x_0,x_1)|^r]^{2/r } \end{aligned}\ ] ] by similar arguments to ( [ e - mhatvec ] ) .
the result follows by chebyshev s inequality .
the convergence rate of @xmath134 is as in lemma [ lem : beta:1 ] .
first write : @xmath1163 lemma [ lem : m : beta](a ) yields @xmath1164 . for the convergence rate for the remaining term , condition ( a )
implies that @xmath838 wpa1 . a mean - value expansion ( using conditions ( a ) and ( b ) ) then yields : @xmath1165 wpa1 , for @xmath840 in the segment between @xmath70 and @xmath66 .
let @xmath1122 .
therefore , wpa1 we have : @xmath1166 where the first line is because @xmath1167 and the second and third lines are by several applications of the cauchy - schwarz inequality .
finally , notice that @xmath1168 by lemma [ lem : g : beta](a ) and condition ( c ) , and : @xmath1169 by the ergodic theorem and condition ( b ) .
the bounds for @xmath183 and @xmath181 follow by lemma [ lem : matcgce](a ) .
the convergence rate for @xmath134 is from lemma [ lem : beta:1 ] . to establish the rate for @xmath133
, it suffices to bound : @xmath1170 let @xmath1171 and let @xmath1115 be a sequence of positive constants to be defined .
define the functions : @xmath1172 then by the triangle inequality : @xmath1173 \right\| \\ & + \sup_{\alpha \in { \mathcal{a } } } \left\| \frac{1}{n } \sum_{t=0}^{n-1 } \tilde b^k(x_t ) h^{tail}_\alpha(x_t , x_{t+1 } ) \tilde b^{k}(x_{t+1})\right\| \\ & + \sup_{\alpha \in { \mathcal{a } } } \left\| { \mathbb{e } } [ \tilde b^k(x_t ) h^{tail}_\alpha(x_t , x_{t+1 } ) \tilde b^{k}(x_{t+1 } ) ] \right\| + \left\| { \mathbb{e } } [ \tilde b^k(x_t ) h_{\hat \alpha}(x_t , x_{t+1 } ) \tilde b^{k}(x_{t+1 } ) ] \right\| \\
= : & \ , { \widehat{t}}_1 + { \widehat{t}}_2 + { \widehat{t}}_3 + { \widehat{t}}_4 \,.\end{aligned}\ ] ] to control @xmath1174 , consider the class of functions : @xmath1175 where @xmath1176 is the unit sphere in @xmath1160 .
each function in @xmath1177 is bounded by @xmath1141 .
moreover : @xmath1178 where @xmath305 is the centered empirical process on @xmath1177 . by theorem 2 of @xcite : @xmath1179 = o \left ( \varphi(\sigma_{n , k } ) + \frac{t_n q \varphi^2(\sigma_{n , k})}{\sigma_{n , k}^2 \sqrt n } + \sqrt n
t_n \beta_q \right)\ ] ] where @xmath790 is a positive integer , @xmath1180 for the norm @xmath863 is defined on p. 400 of @xcite , and : @xmath1181}(u,{\mathcal{h}}_{n , k},\|\cdot\|_{2,\beta } ) } \ , { \mathrm{d } } u \,.\ ] ] exponential @xmath59-mixing and display ( a ) in lemma 2 of @xcite ( with @xmath1182 ) imply : @xmath1183 for some constant @xmath1184 that depends only on the @xmath59-mixing coefficients . therefore : @xmath1185 where @xmath1186 is finite by conditions ( a ) and ( b ) .
take @xmath1187 .
define @xmath1188 and @xmath1189 .
each function in @xmath1190 is of the form : @xmath1191 for @xmath1192 we have : @xmath1193 where : @xmath1194 for any @xmath1195 .
let @xmath1196 . by theorem 2.7.11 of @xcite , we have : @xmath1197}(u,{\mathcal{b}}_k^*,\|\cdot\|_{q } ) \leq n\big ( u/(2(k/\xi_k^2)^{1/q}),s^{k-1},\|\cdot\| \big ) \leq \bigg ( \frac{4 ( k/\xi_k^2)^{1/q } } { u } + 1\bigg)^k\ ] ] since the @xmath853-covering number for the unit ball in @xmath1160
is bounded above by @xmath1198 .
it follows by lemma 9.25(ii ) in @xcite that : @xmath1199}(3u,{\mathcal{h}}_{n , k}^*,\|\cdot\|_{q } ) & \leq \bigg ( \frac{4 ( k/\xi_k^2)^{1/q } } { u } + 1\bigg)^{2k } n_{[\,\,]}(u,{\mathcal{h}}_n^*,\|\cdot\|_{q})\ , .
\label{e : bracket : euclid}\end{aligned}\ ] ] let @xmath1200 $ ] be a @xmath853-bracket for @xmath1190 under @xmath1201 . then @xmath1202 $ ] is a @xmath1203-bracket for the @xmath1204 norm for @xmath1177 , because : @xmath1205 by hlder s inequality . taking @xmath1206 in ( [ e : bracket : euclid ] ) and using the fact that truncation of @xmath591 does nt increase its bracketing entropy , we obtain : @xmath1207 } ( u , { \mathcal{h}}_{n , k } , \|\cdot\|_{2v } ) & \leq n_{[\,\ , ] } \big ( \frac{u}{\xi_k^2 \| e \|_{4s } } , { \mathcal{h}}_{n , k}^ * , \|\cdot\|_{\frac{4vs}{2s - v } } \big ) \notag \\ & \leq \big ( \frac{12 \|e\|_{4s } \xi_k^{2 } ( k/\xi_k^2)^{\frac{2s - v}{4vs}}}{u } + 1\big)^{2k } n_{[\,\,]}\big ( \frac{u}{3 \xi_k^2 \|e\|_{4s}},{\mathcal{m}}^*,\|\cdot\|_{\frac{4vs}{2s - v}}\big ) \ , .
\label{e : entropy : hnk}\end{aligned}\ ] ] finally , by displays ( [ e : beta : norm ] ) and ( [ e : entropy : hnk ] ) and condition ( b ) : @xmath1208}(u,{\mathcal{h}}_{n , k},\|\cdot\|_{2,\beta } ) } \ , { \mathrm{d } } u \\ & \leq \int_0^\sigma \sqrt{\log
n_{[\,\,]}(u / c,{\mathcal{h}}_{n , k},\|\cdot\|_{2v } ) } \ , { \mathrm{d } } u \\ & \leq \int_0^\sigma \sqrt{2 k \log \big(1 + 12 c \|e\|_{4v } \xi_k^{2 } ( k/\xi_k^2)^{\frac{2s - v}{4vs } } /u \big ) } \ , { \mathrm{d } } u + { \mathrm{const}}^{1/2 } ( 3c \xi_k^2 \|e\|_{4s})^\zeta \frac{\sigma^{1-\zeta}}{1-\zeta } \\ & \leq 12\sqrt{2k } c \|e\|_{4s } \xi_k^{2 } ( k/\xi_k^2)^{\frac{2s - v}{4vs } } \int_0^{\sigma/(12c \|e\|_{4s } \xi_k^{2 } ( k/\xi_k^2)^{\frac{2s - v}{4vs } } ) } \sqrt { \log ( 1 + 1/u ) } \ , { \mathrm{d } } u \\ & \quad + { \mathrm{const}}^{1/2 } ( 3 \xi_k^2 \|e\|_{4s})^\zeta \frac{\sigma^{1-\zeta}}{1-\zeta } \\ & \leq 24 \sqrt{2k } c \|e\|_{4s } \xi_k^{2 } ( k/\xi_k^2)^{\frac{2s - v}{4vs } } \bigg ( \frac{\sigma}{12c \|e\|_{4s } \xi_k^{2 } ( k/\xi_k^2)^{\frac{2s - v}{4vs } } } \vee 1 \bigg ) \\ & \quad + { \mathrm{const}}^{1/2 } ( 3 \xi_k^2 \|e\|_{4s})^\zeta \frac{\sigma^{1-\zeta}}{1-\zeta } \,.\end{aligned}\ ] ] using the fact that @xmath1209 and @xmath1210 , we obtain @xmath1211 . substituting into ( [ e : dmr : bd ] ) and using markov s inequality : @xmath1212 similar arguments to the proof of lemma [ lem : m : beta ] yield @xmath1213 and @xmath1214 .
we choose @xmath1141 so that : @xmath1215 which makes @xmath1216 ( since @xmath1217 ) . we will also choose @xmath1218 for sufficiently large @xmath1219 , so that @xmath1220 .
this also ensures that the term @xmath1221 .
therefore , @xmath1174 , @xmath1222 , and @xmath1223 are all @xmath1224 . for the remaining term , by conditions ( c ) and ( d ) of the lemma we may deduce : @xmath1225 + o ( \|\hat \alpha - \alpha_0\|_{{\mathcal{a}}}^2 ) \\ & = o_p(n^{-1/2 } ) + o ( \|\hat \alpha - \alpha_0\|_{{\mathcal{a}}}^2 ) = o_p(n^{-1/2 } ) \,.\end{aligned}\ ] ] combining the above bounds for @xmath1226 yields @xmath1227 . the bounds for @xmath183 and @xmath181 follow by lemma [ lem : matcgce](a ) .
the convergence rate for @xmath176 is from lemma 2.2 of @xcite . for @xmath1056 ,
let @xmath1115 be a sequence of positive constants to be defined
. also define : @xmath1228 we then have : @xmath1229 \right\| \\ & \quad + \sup_{v : \| v\| \leq c } \left\| \frac{1}{n } \sum_{t=0}^{n-1 } \tilde b^k(x_t ) g^{tail}_{t+1 } |\tilde b^{k}(x_{t+1})'v|^\beta \right\| \\ & \quad + \sup_{v : \| v\| \leq c } \left\| { \mathbb{e } } [ \tilde b^k(x_t ) g^{tail}_{t+1 } |\tilde b^{k}(x_{t+1})'v|^\beta ] \right\| \quad = : \quad { \widehat{t}}_1 + { \widehat{t}}_2 + { \widehat{t}}_3 \,.\end{aligned}\ ] ] let @xmath1230 denote the @xmath1231th element of @xmath1232 . also define @xmath1233 observe that @xmath1234 where @xmath305 is the centered empirical process on @xmath1177 . by construction , each @xmath1235 is uniformly bounded by @xmath1236 .
theorem 2 of @xcite gives the bound : @xmath1237 = o \left ( \varphi(\sigma_{n , k } ) + \frac{c^\beta t_n q \varphi^2(\sigma_{n , k})}{\sigma_{n , k}^2 \sqrt n } + \sqrt nc^\beta t_n \beta_q \right)\ ] ] where @xmath790 is a positive integer , @xmath1180 where the norm @xmath863 is defined on p. 400 of @xcite , and : @xmath1181}(u,{\mathcal{h}}_{n , k},\|\cdot\|_{2,\beta } ) } \ , { \mathrm{d } } u \,.\ ] ] exponential @xmath59-mixing and display ( a ) in lemma 2 of @xcite ( with @xmath1238 ) , imply : @xmath1239 for some constant @xmath1240 that depends only on the @xmath59-mixing coefficients .
therefore : @xmath1241 where @xmath1242 is finite by condition ( b ) . set @xmath1243 . to calculate the bracketing entropy integral , first fix @xmath1195 and let @xmath1244 be a @xmath853-cover for @xmath1176 and @xmath1245 be a @xmath1246-cover for @xmath1247 .
for any @xmath1248 and @xmath1249 there exist @xmath1250 and @xmath1251 such that : @xmath1252 where : @xmath1253 therefore : @xmath1197}\big(u , { \mathcal{h}}_{n , k } , \|\,\cdot\,\|_{2s } \big ) \leq \big(\frac{2k\xi_k^{1+\beta } } { u } + 1\big)^k \big(\frac{2(ck\xi_k^{1+\beta})^{1/\beta } } { u^{1/\beta } } + 1\big)^k\ ] ] since the @xmath853-covering number for the unit ball in @xmath1160 is bounded above by @xmath1198 .
it follows by ( [ e : beta : norm : fp ] ) and the above display that : @xmath1208}(u,{\mathcal{h}}_{n , k},\|\cdot\|_{2,\beta } ) } \ , { \mathrm{d } } u \\ & \leq \int_0^\sigma \sqrt{\log
n_{[\,\,]}(u / c,{\mathcal{h}}_{n , k},\|\cdot\|_{2s } ) } \ , { \mathrm{d } } u \\ & \leq \int_0^\sigma \sqrt{k \log \big(1 + 2 c k \xi_k^{1+\beta}/u
\big ) } \ , { \mathrm{d } } u + \int_0^\sigma \sqrt{k \log \big(1 + 2 ( c ck \xi_k^{1+\beta}/u)^{1/\beta } \big ) } \ , { \mathrm{d } } u \\ & \leq 4 c k \sqrt k \xi_k^{1+\beta } \bigg ( \frac{\sigma}{4ck\xi_k^{1+\beta } } \vee 1 \bigg ) \\ & \quad \quad + 2^\beta c c k \sqrt k \xi_k^{1+\beta } \int_0^{\sigma/(2^\beta c c k \xi_k^{1+\beta } ) } \sqrt{\log(1+u^{-1/\beta } ) } \ , { \mathrm{d } } u \\ & \leq 4 c k \sqrt k \xi_k^{1+\beta } \bigg ( \frac{\sigma}{4ck\xi_k^{1+\beta } } \vee 1 \bigg ) + 2^\beta c c k \sqrt k \xi_k^{1+\beta } \frac{\big ( \sigma/(2^\beta c c k \xi_k^{1+\beta } ) \big)^{1-\frac{1}{2\beta}}}{1-\frac{1}{2\beta } } \,.\end{aligned}\ ] ] where the final line is because @xmath1254 . since @xmath1243
, we obtain @xmath1255 . substituting into ( [ e : dmr : bd ] ) and using markov s inequality : @xmath1256 by similar arguments to the proof of lemma [ lem : m : beta ] , we have @xmath1257 and @xmath1258 .
we choose @xmath1141 so that : @xmath1259 which makes @xmath1216 ( since @xmath1217 ) . we will also choose @xmath1218 for sufficiently large @xmath1219 , so that @xmath1260 .
this also ensures that the term @xmath1261 .
therefore , @xmath1174 , @xmath1222 , and @xmath1223 are all @xmath1262 .
step 1 : normalize @xmath124 and @xmath127 so that @xmath1263 .
we first show that : @xmath1264 notice that dividing by @xmath1265 establishes the result under the normalization @xmath1266 in the statement of the lemma .
recall the definitions of @xmath993 in display ( [ e : hat : pk : def ] ) . also define : @xmath1267 whenever @xmath513 ( from lemma [ lem : exist ] ) and @xmath135 is positive and simple ( which it is wpa1 by lemma [ lem : exist : hat ] ) then @xmath1268 and @xmath993 are well defined and we have : @xmath1269 by linearity of trace .
observe that : @xmath1270 but we have : @xmath1271 by cauchy - schwarz , display ( [ e : ckhatbd ] ) , and the normalization @xmath1272 .
it follows from ( [ e : pgbd0 ] ) , ( [ e : pgbd1 ] ) , ( [ e : pgbd2 ] ) , and lemma [ lem : var](a ) that : @xmath1275 but notice that : @xmath1276 whenever @xmath513 ( * ? ? ?
* expression ( 6.19 ) on p. 178 ) where @xmath310 is as in the proof of lemma [ lem : exist ] .
finally , by display ( [ e : resbd : k ] ) we have : @xmath1277 this completes the proof of step 1 .
step 2 : here we show that @xmath1278 by lemma [ lem : matcgce](a ) we have : @xmath1279 which completes the proof of step 2 .
the result follows by combining steps 1 and 2 and using @xmath1280 ( which follows from lemma [ lem : matcgce](b ) and definition of @xmath183 ) . first note that : @xmath1281 \right ) \\
& = \frac{1}{\sqrt
n } \sum_{t=0}^{n-1 } ( \rho^{-1 } \psi_{\rho , t } - \psi_{lm , t } ) + o_p(1)\end{aligned}\ ] ] where the second line is by display ( [ e : ale:1 ] ) and a delta - method type argument .
the result now follows from the joint convergence in the statement of the proposition .
similar arguments to the proof of proposition [ p : asydist : l:1 ] yield : @xmath1282 by continuous differentiability of @xmath1283 on a neighborhood of @xmath66 and the dominance condition in the statement of the proposition , we may deduce : @xmath1284 where @xmath1285 \,.\ ] ] substituting into the expansion for @xmath265 and using assumption [ a : parametric](a ) yields : @xmath1286 the result follows by the joint clt assumed in the statement of the proposition .
we prove part ( i ) first .
let @xmath1287 denote the stationary distribution of @xmath343 .
under assumption [ a : eff ] and the maintained assumption of strict stationarity of @xmath642 , the tangent space is @xmath1288 < \infty$ ] and @xmath1289 = 0 $ ] almost surely@xmath1290 endowed with the @xmath1291 norm ( see pp . 878880 of @xcite which is for a real - valued markov process , but trivially extends to vector - valued markov processes ) .
take any bounded @xmath1292 and consider the one - dimensional parametric model which we identify with the collection of transition probabilities @xmath1293 where each transition probability @xmath1294 is dominated by @xmath1295 and is given by : @xmath1296 where : @xmath1297 for each @xmath1298 we define the linear operator @xmath1299 on @xmath72 by : @xmath1300 observe that : @xmath1301 which is a bounded linear operator on @xmath72 ( since @xmath71 is a bounded linear operator on @xmath72 and @xmath295 is bounded ) . by taylor s theorem and boundedness of @xmath295
we can deduce that @xmath1302 and therefore : @xmath1303 where the @xmath1304 term is uniform in @xmath1305 .
it then follows by this and boundedness of @xmath295 that @xmath1306 .
similar arguments to the proof of lemma [ lem : exist ] imply that there exists @xmath321 and @xmath1307 such that the largest eigenvalue @xmath1308 of @xmath1299 is simple and lies in the interval @xmath1309 for each @xmath1310 .
it follows from a perturbation expansion of @xmath1308 about @xmath1311 ( see , for example , equation ( 3.6 ) on p. 89 of @xcite ) that : @xmath1312 + o(\tau^2 ) \notag \\ & = \tau \int m(x_t , x_{t+1})\phi(x_{t+1})\phi^*(x_t ) h(x_t , x_{t+1}){\mathrm{d}}q_2(x_t , x_{t+1 } ) + o(\tau^2 ) \label{e : eff : pf3}\end{aligned}\ ] ] under the normalization @xmath1313 , where the second line is by ( [ e : eff : pf1 ] ) and ( [ e : eff : pf2 ] ) .
expression ( [ e : eff : pf3 ] ) shows that the derivative of @xmath1308 at @xmath1311 is @xmath1314 . since bounded functions @xmath295 are dense in @xmath1315
, we have that @xmath42 is differentiable relative to the tangent set @xmath1315 with derivative @xmath1316 .
the efficient influence function is the projection of @xmath1316 onto @xmath1315 , namely : @xmath1317
= \psi_\rho(x_t , x_{t+1})\ ] ] because @xmath1318 = \phi^*(x_{t } ) { \mathbb{m } } \phi(x_t ) = \rho \phi(x_t ) \phi^*(x_t)$ ] .
it follows that @xmath1319 $ ] is the efficiency bound for @xmath42 .
we now prove part ( ii ) . by linearity ,
the efficient influence function of @xmath156 is : @xmath1320 where @xmath1321 is the efficient influence function for @xmath1322 $ ] . to simplify notation ,
let @xmath1323 .
it is well known that : @xmath1324 - { \mathbb{e}}[l(x_t , x_{t+1})|x_0 = x_0 ] \big ) \label{e - phim}\end{aligned}\ ] ] ( see , e.g. , p. 879 of @xcite ) .
it may be verified using the telescoping property of the sum ( [ e - phim ] ) that @xmath1325 $ ] , as required .
we first show that any positive eigenfunction of @xmath71 must have eigenvalue @xmath42 .
suppose that there is some positive @xmath97 and scalar @xmath251 such that @xmath1326 .
then we obtain : @xmath1327 with @xmath1328 because @xmath46 and @xmath54 are positive , hence @xmath1329 .
a similar argument shows that any positive eigenfunction of @xmath739 must correspond to the eigenvalue @xmath42 .
it remains to show that @xmath43 and @xmath46 are the unique eigenfunctions ( in @xmath72 ) of @xmath71 and @xmath739 with eigenvalue @xmath42 .
we do this in the following three steps .
let @xmath1330 .
we first show that if @xmath1331 then the function @xmath1332 given by @xmath1333 also is in @xmath1334 . in the second step
we show that @xmath1331 implies @xmath1335 or @xmath1336 .
finally , in the third step we show that @xmath1337 . for the first step , first observe that @xmath1338 because @xmath1339 by assumption [ a : id:1](b ) .
then by assumption [ a : id:1](c ) , for any @xmath1331 we have @xmath1340 and so @xmath1341 ( almost everywhere ) . on the other hand , @xmath1342 which implies that @xmath1343 and hence @xmath1344 . for the second step , take any @xmath1331 that is not identically zero .
suppose that @xmath1335 on a set of positive @xmath75 measure ( otherwise we can take @xmath1345 in place of @xmath54 ) .
we will prove by contradiction that this implies @xmath1346 .
assume not , i.e. @xmath1347 on a set of positive @xmath75 measure .
then @xmath1348 ( almost everywhere ) and @xmath1349 .
but by step 1 we also have that @xmath1350 . then for any @xmath688 we have @xmath1351 ( almost everywhere ) by assumption [ a : id:1](c ) .
therefore , @xmath1352 ( almost everywhere ) .
this contradicts the fact that @xmath1335 on a set of positive @xmath75 measure .
a similar proof shows that if @xmath1353 holds on a set of positive @xmath75 measure then @xmath1353 . for the third step we use an argument based on the archimedean axiom ( see , e.g. , p. 66 of @xcite ) .
take any positive @xmath1331 and define the sets @xmath1354 and @xmath1355 ( where the inequalities are understood to hold almost everywhere ) .
it is easy to see that @xmath1356 and @xmath1357 are convex and closed .
we also have @xmath1358 \subseteq s_+$ ] so @xmath1356 is nonempty .
suppose @xmath1357 is empty .
then @xmath1359 on a set of positive measure for all @xmath1360 .
by step 2 we therefore have @xmath1359 ( almost everywhere ) .
but then because @xmath72 is a lattice we must have @xmath1361 for all @xmath1360 which is impossible because @xmath97
. therefore @xmath1357 is nonempty . finally , we show that @xmath1362 .
take any @xmath1363 .
clearly @xmath1364 .
by claim 2 we know that either : @xmath1365 ( almost everywhere ) which implies @xmath1366 or @xmath1367 ( almost everywhere ) which implies @xmath1368
. therefore @xmath1362 .
the archimedean axiom implies that the intersection @xmath1369 must be nonempty .
therefore @xmath1370 ( the intersection must be a singleton else @xmath1371 and @xmath1372 with @xmath1373 ) and so @xmath1374 ( almost everywhere ) .
this completes the proof of the third step .
assumption [ a : id:0](a ) implies that @xmath697 ( see proposition iv.9.8 and theorem v.6.5 of @xcite ) .
the result now follows by theorems 6 and 7 of @xcite with @xmath1375 . that @xmath42 is isolated follows from the discussion on p. 1030 of @xcite .
consider the operator @xmath1376 with @xmath1375 .
proposition [ p : a : exist ] implies that @xmath1377 .
further , since @xmath71 is power compact it has discrete spectrum ( * ? ? ?
* theorem 6 , p. 579 ) .
we therefore have @xmath1378 and hence @xmath1379 where @xmath1380 and @xmath1381 , @xmath734 and @xmath698 commute ( see , e.g. , p. 331 of @xcite or pp .
1034 - 1035 of @xcite ) . since these operators commute , a simple inductive argument yields : @xmath1382 for each @xmath38 .
by the gelfand formula , there exists @xmath321 such that : @xmath1383 let @xmath1384 be the maximal subset of @xmath3 for which @xmath1385 . if this subsequence is finite then the proof is complete .
if this subsequence is infinite , then by expression ( [ gelfand ] ) , @xmath1386 therefore , there exists a finite positive constant @xmath1387 such that for all @xmath443 large enough , we have : @xmath1388 and hence : @xmath1389 as required . | stochastic discount factor ( sdf ) processes in dynamic economies admit a permanent - transitory decomposition in which the permanent component characterizes pricing over long investment horizons .
this paper introduces econometric methods to extract the permanent and transitory components of the sdf process .
we show how to estimate the solution to the perron - frobenius eigenfunction problem of @xcite using data on the markov state and the sdf process . estimating directly the eigenvalue and eigenfunction allows one to ( 1 ) construct empirically the time series of the permanent and transitory components of the sdf process and
( 2 ) estimate the yield and the change of measure which characterize pricing over long investment horizons .
the methodology is nonparametric , i.e. , it does not impose any tight parametric restrictions on the dynamics of the state variables and the sdf process .
we derive the large - sample properties of the estimators and illustrate favorable performance in simulations .
the methodology is applied to study an economy where the representative agent is endowed with recursive preferences , allowing for general ( nonlinear ) consumption and earnings growth dynamics .
* keywords : * nonparametric estimation , sieve estimation , stochastic discount factor , permanent - transitory decomposition , nonparametric value function estimation .
* jel codes : * c13 , c14 , c58 . |
atlas ( a toroidal lhc apparatus ) @xcite is one of the four major experiments at the forthcoming lhc ( large hadron collider ) , in which protons will collide at a center of mass energy of 14 tev .
it consists of three main sub - systems : the inner detector ( i d ) , the calorimetry system ( electromagnetic and hadronic calorimeters ) and the muon spectrometer ( ms ) .
it uses a superconducting magnet system with a central solenoid around the inner detector and large air - core toroid magnets for the muon spectrometer .
[ atlas ] shows the overall detector layout .
+ the commissioning of the atlas detector with physics data already started while the detector was being mechanically and electrically completed by collecting cosmic rays with the parts of the detectors that were becoming available .
global cosmic rays runs with the complete detector with different magnetic field configurations are now being recorded .
the full system was ready to collect data during the three days in which a single beam of the lhc at the injection energy of 450 gev circulated through atlas . + in addition to put in place the trigger and data acquisition chains , commissioning of the full software chain is a main goal .
this is interesting not only to ensure that the reconstruction , monitoring and simulation chains are ready to deal with lhc collisions data , but also to understand the detector performance in view of achieving the physics requirements .
furthermore , they have been used to validate and improve the atlas simulation comparing the results obtained from real data and those from the specific simulations .
the commissioning period is being used to put in place the full detector operation chain , i.e. from the lvl1 trigger and data acquisition to the analysis in the grid tier-2 computing centers .
the software plays an important role in this chain .
+ the software chain is shown in fig .
[ softwarechain ] .
cosmic rays and events recorded during lhc single beam operations are reconstructed using the full atlas software chain , with specific modifications to account for the lack of synchronization of these kind of events with the readout clock ( even during single beam operations due to the not yet operational rf capture ) and the fact that particles do not come from the center of the detector .
the reconstruction and monitoring algorithms have been continuously running online in the atlas control room to provide online event displays and histograms monitoring the data quality during detector operations .
event displays in which a cosmic ray track is reconstructed is shown in fig .
one can see hits in the trigger and precision muon spectrometer chambers , energy deposited in the calorimeters and hits in the inner detector .
examples of the type of histograms produced to check the data quality are number of tracks reconstructed , energy of calorimeter cells , hit occupancies , synchronization between the different sub - detectors , etc .
some examples are shown in fig .
[ monitoring ] . in the left plot ,
the occupancy in the semiconductor tracker ( sct ) modules is shown .
the right one shows the difference of the @xmath0 track parameter measured in the inner detector and the muon spectrometer as a function of the event number .
it can be seen that both sub - detectors are synchronized .
+ the high level trigger algorithms have been providing data streams adequate for different purposes such as alignment and calibrations running at the cern analysis facility ( caf ) .
the first offline data processing takes place with a latency of less than one hour at the tier-0 and further re - processings with new software versions and updated conditions are done at the grid tier-1 centers . at the end of the chain , event summary data ( esd ) , monitoring histograms , combined ntuples ( cbnt ) , analysis data objects ( aod ) and tag files to allow for a selection of events are produced .
in addition to the cosmic rays and single beam data taken by atlas in the pit , simulated data has also been made available .
this is important to check the reconstruction software and allows for data / mc comparisons in terms of detector response , efficiencies , etc . beam gas and beam halo events have been simulated at 5 tev .
however , as already mentioned , the energy of the lhc proton beam that circulated through atlas was of 450 gev .
consequently , comparisons data / mc have only been made for cosmic rays events .
+ the simulation chain is shown in fig .
[ simulation ] .
a monte carlo generator was used for simulating muons from cosmic ray events , based on measurements of the differential vertical muon cross section and analytical calculations extrapolated to low muon energy @xcite .
single muons are generated at the surface .
+ the simulation toolkit geant4 was used in order to simulate the passage of particles through the detector , giving rise to energy depositions in the detector . in order to reduce the simulation time ,
only those events pointing to the atlas detector are passed to the geant4 simulation and only those events that have energy depositions in a given volume -in this case , in the transition radiation tracker ( trt ) volume- are sent to the next step .
this final state consists in emulating the electronics response ( the so - called digitization process ) in order to end up with simulated raw data .
cosmic rays have allowed us to study the atlas detector in terms of efficiencies , resolutions , channel integrity and alignment and calibrations .
some examples of these studies are shown in fig .
[ detectorstudies ] . on the left plot , the energy response and track length measured in the hadronic calorimeter ( tilecal )
the @xmath1 dependence that is observed is well understood by the variation in the length traversed by the muons in the calorimeter .
+ + furthermore , cosmic ray runs have allowed for studies of the performance of combined algorithms . the atlas experiment will identify and measure muons in the muon spectrometer ( in a region @xmath22.7 ) . however , it is not enough .
because of various acceptance gaps in the muon spectrometer and the decrease in the momentum resolution for tracks with low momenta , algorithms that combine the information from the different atlas sub - detectors are essential .
[ ed2 ] shows a cosmic ray event in which a combined track ( inner detector and muon spectrometer ) is reconstructed .
the combined tracker matches first tracks of the inner detector with those reconstructed in the muon system and then performs a global @xmath3 fit using all hits . + the difference between the @xmath0 track parameter measured in these two sub - detectors is shown in the right plot of fig . [ detectorstudies ] .
the mean of these distributions gives an idea of the relative alignment between them .
the distributions are more centered for mc , as expected . in addition , the momentum reconstructed in the muon spectrometer and in the inner detector is compared in fig . [ momentum ] for tracks reconstructed in the top ( left plot ) and bottom ( right plot ) part of the muon spectrometer .
the difference between the momentum reconstructed in both sub - detectors is the energy deposited in the calorimeters .
data taken during the lhc single beam period has also been analysed .
[ eta ] ( left ) shows a single beam ( halo - like muons ) event . a comparison of the @xmath0 track parameter measured by the inner detector in cosmic rays and single beam runs is shown in the right plot of fig .
cosmics tracks are mainly vertical while single beam tracks are more horizontal . + +
the complete software chain is being commissioned making use of the different type of data taken by atlas , i.e. cosmic rays and lhc single beam data .
all these data have allowed us to study the atlas detector in terms of efficiencies , resolutions , channel integrity and alignment and calibration corrections .
they have also allowed us to test and optimize the different sub - system reconstruction as well as the muon combined performance algorithms , such as combined tracking tools and different muon identification algorithms based on measurements in the inner detector and muon spectrometer or calorimeters .
+ the commissioning work described here requires many things to be working , from the detectors systems through to the offline software .
i would like to acknowledge the huge contribution from the whole atlas collaboration . | looking towards first lhc collisions , the atlas detector is being commissioned using the physics data available : cosmic rays and data taken during the lhc single beam operations at 450 gev . during the installation of the atlas detector in the cavern ,
cosmic rays were collected with the different parts of the detector that were available .
combined cosmic runs taken with the full installed detector with and without magnetic field as well as a few single beam events recently recorded are being used to commission the full system prior to the first proton collisions . |
the prospect of quantum computing ( qc ) has caused great excitement in condensed matter physics .
if a set of qubits can be maintained in a coherent , controllable many - body state , certain very difficult computational problems become tractable . in particular , successful qc would mean a revolution in the areas of cryptography @xcite and data - base searching @xcite .
in addition , it would mean a great advance in general technical capabilities , since the control of individual quantum systems and their interactions would represent a new era in nanotechnology .
however , from a practical point of view , a dilemma presents itself immediately . on the one hand ,
one wishes to control quantum degrees of freedom using external influences , since that is how a quantum algorithm is implemented , and to measure them , since that is the output step .
on the other hand , the system must be isolated from the environment , since random perturbations will destroy the quantum coherence that is the whole advantage of qc .
this is the isolation - control dilemma , and it leads to a very rough figure - of - merit @xmath1 for any quantum computer . if we define the decoherence time @xmath2 as the time it takes to lose quantum coherence , and the clock speed @xmath3 ( roughly the inverse of the time to run a logic gate ) , then the figure of merit is @xmath4 and a practical machine should satisfy @xmath5 at least . if the clock speed is limited only by standard electronics , then we may be able to achieve @xmath6 .
this would imply that @xmath7 is a lower limit for the decoherence time .
the dilemma has not yet been solved , though a number of solutions have been proposed .
a particularly attractive solution is to use spin degrees of freedom as qubits .
nuclear spins interact relatively weakly with their environment because the coupling , proportional to the magnetic moment , is small .
yet there has grown up a sophisticated technology ( nmr ) for the manipulation of the spins , and some rudimentary computations have been performed @xcite .
readout is the main difficulty with this approach , since the field created by a single moment is tiny , and pure states can not be achieved in the macroscopic samples used .
electron spins also interact weakly with the environment in some circumstances : relaxation times in excess of @xmath8 have been measured for donor bound states of phosphorus - doped silicon ( si : p ) @xcite . the corresponding manipulation technology ( esr )
has also reached a high level of sophistication , but the magnetic moment of the electron exceeds that of the nucleus by 3 orders of magnitude , which presents problems of isolation .
readout should be easier than for nuclei , since detection of single electron charges is certainly possible @xcite , and spin - dependent detection is not far out of the reach of current technology .
solid - state implementations of qc are particularly attractive because of the possibility of using existing computer technology to scale small numbers of qubits up to the @xmath9 or so that would be needed for nontrivial computations .
the first paper to propose using the electron spin in a quantum dot subjected to a strong dc magnetic field was that of loss and divincenzo @xcite .
kane @xcite proposed employing the nuclear spin in the si : p system as the qubit .
a specific structure consisting of silicon - germanium ( sige ) layers was proposed by vrijen _
this structure incorporates the idea that the @xmath10-factor of an electron can be changed by moving it in a ge concentration gradient , allowing individual electron to be addressed by the external ac field .
a different sige structure has been proposed by friesen _
this structure is designed so that the electron number on the dots , and the coupling between the dots , can be carefully controlled .
solid - state implementations must also face the isolation - control dilemma .
decoherence times must exceed the @xmath11 number in the actual physical structures that are needed for the operation of quantum algorithms . in this paper , we examine whether this can be the case for some of the existing proposals based on electron - spin qubits . in the process , we hope to learn something about modifications to these structures that can increase @xmath2 .
we shall focus on low temperature operation , since , as we shall see , this will probably be necessary in order to obtain sufficiently large @xmath2 .
we can build on a large body of work , both theoretical and experimental , from the 1950 s and 1960 s on esr in doped semiconductors . in a series of papers , feher and collaborators @xcite investigated the relaxation time for the spin of electrons bound on donor sites in
lightly doped si . at sufficiently low temperatures , the relaxation time
@xmath12 is dominated by single - phonon emission and absorption . in the presence of spin - orbit coupling ( soc ) , this can relax the spin , causing decoherence .
the theory was worked out by hasegawa @xcite and roth @xcite .
it must of course be recognized that this spin - lattice relaxation time is not necessarily to be identified with the decoherence time .
the decoherence time is the shortest time for any process to permanently erase the phase information in the wavefunction .
this may mean the phase for a single spin , but it also means that the relative phases of the wavefunctions of different spins must also be preserved , so that processes that cause mutual decoherence must also be taken into account .
the actual decoherence time is the minimum of all of these times .
a spin relaxation time in excess of @xmath11 is a necessary , not a sufficient , condition for the viability of a solid - state electron - spin qc proposal .
a qc must have precise input as well as an accurate algorithm .
preparation of the spin state is often proposed to be done by thermalization of the spin system at a low temperature .
the time to do this actually sets an _
limit on the relaxation time of whatever processes thermalize the spins to the lattice .
a limit of perhaps @xmath13 @xmath3 is a reasonable requirement .
this paper focuses on @xmath12 , the time for relaxation of the longitudinal component of the magnetization , by spin - phonon interactions .
these processes cause real spin - flip transitions .
they occur at random times and thus indubitably cause decoherence .
in addition to these processes characterized fully by @xmath12 , there are processes which introduce random phase changes in the spin wavefunctions .
to characterize all such processes by a single dephasing time
@xmath14 will usually not be sufficient for understanding the operation of a multi - qubit system @xcite .
difficulties of definition arise , and care must be taken to specify which phase is involved and to what extent it is randomized . for a single spin system there is no ambiguity .
the @xmath15 density matrix @xmath16for the qubit with cylindrical symmetry involves only 2 independent parameters @xmath17 , and @xmath18 the time dependence of @xmath17 after a system preparation is exponential with decay constant @xmath12 and represents the return of the longitudinal component of the magnetization to its equilibrium value .
the decay is due to inelastic transitions of the type calculated in this paper .
@xmath19 on the other hand , is nonzero only if the preparation of the spin state has a transverse component : @xmath20 the decay of this quantity represents the irreversible conversion of this state to an incoherent mixure of up and down states . again , this is genuine decoherence of the spin state , since the phase information can not be recovered .
the time dependence of @xmath19 when the spin is in a strong field has been calculated by mozyrsky _
@xcite using a markovian approximate master equation .
they find that the time scale of the decay due to spin - phonon coupling is very short , of the order of the time for a phonon to cross the electron s wavefunction , which is about @xmath21 .
but the decay is incomplete , with @xmath19 retaining all but @xmath22 of its original value . their calculation was for si : p , but a very similar result should hold for the dot case .
decoherence at this level is certainly acceptable for quantum computing .
these authors also pointed out that the decay of the remainder of @xmath23 is due to the spin - flip processes computed in this paper . if @xmath14 is defined as the dominant decay time of the off - diagonal density matrix element , then @xmath24 for spin - phonon processes . a quite different source of decoherence is the hyperfine coupling to nuclear spins .
the nuclear spins produce an effective random magnetic field on the electrons .
very recent calculations using semiclassical averaging techniques @xcite have obtained a very short relaxation time @xmath25 for gaas - based dot systems in a strong field .
this represents the decay of the transverse magnetization of an ensemble of dots .
this is a dephasing time , but _ not _ a decoherence time .
the electrons spins precess in what is effectively the frozen field of the nuclei .
this field is spatially random , and the differential precession of the electron spins leads to the magnetization decay .
however , this is not an irreversible loss of the phase information of the collective wavefunction .
spin echo exeriments are a very beautiful demonstration of precisely this point .
inhomogeneous broadening presents challenges for the calibration and operation of quantum computers , but does not destroy coherence . finally , in any implementation based on electron
spin qubits , there will certainly exist small interactions between the spins themselves .
the dipole - dipole interaction , for one , can not be avoided , and there may be indirect spin - spin interactions mediated by the gates .
a recent paper suggests that these interactions set the fundamental time scale @xmath26 for si quantum dot implementations of qc @xcite .
these interactions do produce experimental broadening of esr lines in experiments on bulk systems , and this might be taken as decoherence . in our view , however , these interactions do not destroy the coherence of a state .
the system is the set of all the qubits . during the course of a quantum algorithm they are collectively in a pure state ( in principle ) .
any decoherence that destroys the purity of the state comes from averaging over the unknown states of the environment .
the broadening that comes from dipole - dipole interactions comes , in nmr and esr calculations , from averaging over the states of the system itself , which is not an appropriate method for calculating decoherence .
the effect of qubit - qubit interactions that can not be turned off is to complicate the quantum algorithm .
a quantum algorithm is a unitary transformation that must always include the effect of the system hamiltonian ( including dipole - dipole interactions ) in addition to external operations . in every case except for very simple ones , this algorithm must first be computed , presumably with the help of a classical computer .
this step in qc may be termed `` quantum compilation '' .
the issue that qubit - qubit interactions raise is not one of decoherence , but rather whether the determination of the algorithm , the complation step , becomes prohibitively difficult .
this could happen for two reasons .
one is that the interactions are so poorly known that they can not be corrected for .
it seems likely that quantum error correction can resolve this difficulty .
a second and more interesting possibility that the interactions convert the computation of the algorithm itself into a problem that grows exponentially with the size of the system .
we regard this as an open question and a deep one , that combines many - body theory with algorithm design and error correction .
we note that in nmr implementations the interaction between the qubits also can not be turned off , but it can be canceled by refocusing @xcite . in this paper ,
our aim is to evaluate the importance of spin - phonon coupling as a source of decoherence in quantum dot qubits .
fundamentally , the issue is whether the long relaxation times @xmath27 observed at low temperatures in bulk si : p carry over to sige dots proposed for qc . in sec .
[ sec : well ] we introduce the structures that we are interested in . in sec .
sec : decoherence the physics of the spin - phonon relaxation mechanism is described qualitatively . in sec . [
sec : si ] calculations and results for pure si systems are given as a function of the critical design and operating parameters . in sec .
[ sec : ge ] are found the correponding results for structures containing ge .
[ sec : conclusion ] is the conclusion .
si - ge heterostructures are utilized widely in the digital electronics industry , and presently have the shortest switching times of any device .
one reason for their success lies in the ability to engineer structures of near perfect purity , with control over thicknesses and interfaces that approaches atomic precision a technological tour - de - force .
an equally key achievement has been the harnessing of strain as a tool to control band offsets in heterostructure devices .
this paper presents calculations of spin relaxation for real sige structures such as those proposed by vrijen _
_ @xcite and friesen _ et al .
accordingly , we have calculated the electron wavefunctions in quantum wells , which is needed as input for these calculations .
details of these calculations are presented in ref .
@xcite , and will not be repeated here . in this section
we only describe those aspects of the calculations that are germane to spin relaxation .
quantum wells are constructed by sandwiching a very thin layer of one material between two others .
electrons can be confined in the quantum well layer when the conduction band offsets produce a potential well .
the key to this technology is therefore to understand the band structure of the various layers . in this section
we will consider a particular class of wells formed of pure si , sandwiched between barrier layers of sige .
we will find that this is optimal from the standpoint of spin coherence .
metallic gates or impurities create zero - dimensional bound states that define a quantum dot .
we first review briefly effective mass theory for dots in pure , unstrained si , then unstrained sige , and finally strained si . in pure si ,
the @xmath28 conduction band minima occur near the symmetry points @xmath29 , in the directions @xmath30 . in a perfect si crystal
these minima are six - fold degenerate , the valleys being equivalent . in the dot
the electron feels a potential @xmath31 in addition to the atomic potential , which lifts the degeneracy , though the splittings are not large .
the spatial variation of @xmath31 is on length scales generally much longer than the lattice spacing . for the moment
, we shall assume that the electron is in the ground state of @xmath31 and ignore mixing with any excited states .
to the extent that the scale of variation of @xmath32 is much longer than the atomic spacing there are six nearly degenerate ground states .
this is referred to as the valley degeneracy .
the wavefunctions can be written as@xcite @xmath33here , @xmath34 is a bloch function of the form @xmath35where @xmath36 are the six @xmath28 minima @xmath37 ( we shall always use this ordering ) , and @xmath38 are periodic functions with the same periodicity as the crystal potential @xmath39 .
the @xmath40 are envelope functions that satisfy the schrdinger - like equation
@xmath41 f_{\pm z}(\vec{r% } ) ) = ( e - e_{k_{z}}^{(\delta ) } ) f_{\pm z}(\vec{r } ) , \label{eq : envelope}\]]and are independently normalized to unity , similar to wavefunctions .
analogous equations can be given for the @xmath42 and @xmath43 minima .
we see that @xmath44 , @xmath45 , and @xmath46 , so only three independent envelope functions must be computed . @xmath47 and @xmath48 are the longitudinal and transverse effective masses associated with the anisotropic conduction band valleys .
@xmath49 is the @xmath28 conduction band edge at @xmath50 .
the splitting of the degeneracy comes from corrections to this envelope fuction approximation .
different choices of the constants @xmath51 determine the six states . their values will be discussed in sec . [
sec : decoherence ] .
this formalism is a good approximation for both dot and impurity bound states , as the valley splitings are much smaller than the energy scales in [ eq : envelope ] .
germanium is completely miscible in si , forming a random alloy . for variable ge content @xmath52 , si@xmath53ge@xmath54 exhibits materials properties that vary gradually over the composition range .
the alloy lattice constant , @xmath55 , follows a linear interpolation between pure si and ge , known as vegard s law , quite accurately for all @xmath52 : @xmath56 @xcite .
electronic properties show an abrupt change in behavior near @xmath57 , where the si - like @xmath28 minima cross over to four - fold degenerate , ge - like , @xmath58 minima . in this work we focus on the range @xmath59 , which is strictly si - like , though we will have some remarks below on ge - rich structures . throughout this range , properties such as effective mass and the dielectric constant vary only slightly from pure si values . for our calculations ,
the most important parameter is the conduction band edge , @xmath60 , which remains 6-fold degenerate in the range @xmath59 .
the theory of the variation of @xmath61with @xmath52 is not germane to the present work , and we simply quote the empirical result , linear in @xmath52 , which is consistent with ref .
rieger93 : @xmath62(we note , however , that ref .
@xcite suggests a slope for @xmath63 of opposite sign . )
the relatively weak variations of the effective mass and the dielectric constant will be ignored here .
we consider thin si wells in which the si layer grows pseudomorphically .
the in - plane lattice constant , @xmath64 , must be the same for all the layers , causing a tetragonal distortion in the strained layer(s ) . here
we consider the case of strained si grown on the ( 001 ) surface of relaxed si@xmath53ge@xmath54 .
the in - plane si lattice constant depends on @xmath52 as @xmath65since @xmath66 , the si is under tensile strain in the plane .
hence , the out - of - plane si lattice constant , @xmath67 , is reduced according to continuum elastic theory : @xmath68 , \label{eq : aperp}\]]where @xmath69 and @xmath70 are elastic constants for pure si .
strain produces shifts of the @xmath28 band proportional to the strain variables @xmath71with proportionality constants called the dilational and uniaxial deformation potentials , @xmath72 and @xmath73 , respectively . because of the anisotropic nature of the strain ,
the two @xmath74 minima are shifted down relative to the @xmath75 and @xmath76 minima , resulting in a splitting of the @xmath28 conduction band .
the net shifts with respect to the unstrained si @xmath28 band are given by @xcite @xmath77the first terms in eqs .
( [ eq : deperp ] ) and ( [ eq : depar ] ) are hydrostatic strain terms , which lower the conduction edge compared to unstrained si . the second terms in eqs .
( [ eq : deperp ] ) and ( [ eq : depar ] ) produce the splitting , associated with uniaxial strain . to perform our calculations
, we use the materials parameters given in table [ tab : params ] .
however we note that the deformation potentials , particularly @xmath78 , are very difficult to measure experimentally .
considerable disagreement exists in the literature as to the value and even the sign of @xmath72 @xcite .
the value given in table [ tab : params ] was reported ( but not endorsed ! ) in ref .
fischetti96 , and provides energy band variations in general agreement with refs .
@xcite and @xcite .
we arrive at the following strain - induced shifts of the conduction band edge for pure si : @xmath79the corresponding shift in the relaxed barrier layers , due to the presence of ge , was given in eq .
( [ eq : unstrained ] ) .
together , these results describe the conduction band offsets for the quantum well that are used in our simulations .
we now apply our results to two specific quantum well designs of interest for quantum computing .
design ( 1 ) , shown in the inset of fig . [ fig : donor ] , is a version of that proposed by vrijen _
et al . _
@xcite , in which electrons are trapped on donor ions ( usually p ) , implanted in a semiconductor matrix . in that work ,
the quantum well is split into ge - rich and si - rich regions to facilitate single qubit operations . for simplicity , we consider here a uniform quantum well , formed of pure si , with a single dopant ion located at the center of the well .
in such a device , single qubit operations can be accomplished using a coded qubit scheme @xcite .
design ( 2 ) , proposed by friesen et al .
@xcite , is shown in fig .
[ fig : qw ] .
the confinement potential for the electrons is much softer than in design ( 1 ) .
electrons are trapped vertically by the quantum well , and laterally by the electrostatic potential arising from lithographically patterned , metallic top - gates .
additionally , the quantum dot is tunnel - coupled to a degenerate doped back - gate .
the dimensions for both designs are given in the figures .
the wave function of the bound electron is computed in the envelope function formalism , eq .
( [ eq : envelope ] ) .
couplings between the different valleys are introduced through the perturbation theory described in sec .
[ sec : si ] .
this procedure provides the specific values of @xmath80 for the ground state , which we use in our calculations .
the abrupt conduction band offsets are handled by matching the ground state wave function @xmath81 and @xmath82 at the interfaces .
( remember that we have equated effective masses on both sides of interfaces . )
due to linear independence of the bloch functions , the boundary conditions do not cause mixing of the envelope functions
. solutions of eq .
( [ eq : envelope ] ) and the analogous @xmath83 equation are obtained , using commercial 3d finite element software . as will be seen in sec .
[ sec : ge ] , the key quantity for the computation of @xmath27 is @xmath84 , which describes the probability for the bound electron to be on a ge atom .
ge is associated with reduced coherence times , by virtue of its large spin - orbit coupling . referring to eq .
( 1 ) , one deduces that this probability may be expressed as @xmath85 , \]]where @xmath86 is the volume outside the quantum well , if the well is pure si .
the subscript @xmath10 refers to the ground state .
the term in square brackets reflects the probability of finding the bound electron in a barrier region , while @xmath52 gives the probability that the electron is on a ge site .
[ fig : donor ] shows the results of our calculations for @xmath84 in design ( 1 ) , as a function of the ge concentration , @xmath52 , in the si@xmath53ge@xmath87 barriers . for @xmath88
, @xmath84 decreases with @xmath52 for two reasons .
first , as @xmath52 increases , the conduction band offset at the quantum well also increases , allowing less of the wavefunction to penetrate the barrier .
second , the spatial extent the electron in the @xmath89 direction is greater for @xmath83 than @xmath90 , because of the anisotropic effective mass .
however , less of @xmath83 is mixed into the wavefunction for large @xmath52 , since @xmath91 becomes very small . for @xmath92
, @xmath84 drops quickly to zero , due to the absence of ge in the barriers . in the actual design of vrijen _ et al .
_ , @xcite there is ge in the active layer .
to give an idea of the effect of this , we include an equivalent value of @xmath84 for such a structure .
[ fig : qw ] shows results of the @xmath84 calculation for design ( 2 ) , as a function of the quantum well thickness , @xmath93 . to perform the calculations , we have considered a fixed ge concentration , @xmath94 , and taken the limit of large strain , so that @xmath95 and @xmath96 for the ground state
. as @xmath93 increases , less of the wavefunction penetrates the barrier regions , causing @xmath84 to decrease .
in this section we give the method for calculating @xmath12 , the spin - flip time of a spin qubit in the ground orbital state due to emission or absorption of a phonon , following the logic used by hasegawa @xcite and roth @xcite for bulk si .
consider a single impurity with a unit positive charge , such as a phosphorus atom , at the origin . in the absence of central cell corrections , there is a 12-fold degenerate ground state , including spin .
this valley degeneracy of the ground state is reduced to two by these corrections , and the splitting between the 2-fold spin - degenerate ground state and the higher states is of order @xmath97 we shall discuss the detailed linear combinations ( @xmath98-values ) of the states below , as the coefficients giving the various valley amplitudes play an important role in the calculation of matrix elements .
these twelve states may all be thought of as hydrogenic @xmath99 states .
the splitting of @xmath99 and @xmath100 is about @xmath101 , larger than the @xmath102 valley splittings .
let us now split the 2-fold degenerate ground state by applying a dc magnetic field in the @xmath93-direction .
the transition rates between these states are denoted by @xmath103 and @xmath104 .
the relaxation time @xmath12 is defined by @xmath105 @xmath106 .
the transitions are caused by phonons , but there are important approximate symmetries that suppress these transitions .
these are : ( 1 ) spin rotation symmetry , meaning that the electron spin can not be flipped by a phonon ; this symmetry is broken by soc ; ( 2 ) time - reversal symmetry , meaning that one state can not be changed into its time - reversed partner by emission or absorption of a phonon ; this symmetry is broken by the external magnetic field ; ( 3 ) point group symmetries ; these are partially broken when strain is applied .
the spin rotation symmetry would rule out phonon - mediated transitions between the two states entirely if there were no spin - orbit coupling ( soc ) .
this means that the effects of soc on the wavefunctions , even though these effects are small in relatively low-@xmath107 si , must be taken into account .
when we refer to a state as @xmath108 or @xmath109 , these symbols must be taken to refer to the majority spin content of the state , not to a pure spin state .
transition rates are roughly proportional to @xmath110 , [ more precisely @xmath111 , where @xmath112 is the transverse ( longitudinal ) @xmath10-factor , see below for definitions ] .
the time - reversal symmetry implies that transitions can not take place directly between kramers - degenerate states even in the presence of spin - orbit coupling . the direct phonon - mediated transitions between the two states of interest
to us are strongly suppressed by this approximate symmetry .
it is broken only by the external field @xmath113 .
the fastest processes then involve virtual excitation to higher - energy states that are mixed into the ground state by @xmath113 .
hence @xmath114 involves a factor @xmath115 .
there is an additional factor of @xmath116 from the phonon density of states , giving an overall rate @xmath117 in the limit of small @xmath113 .
the point group symmetry is reduced from cubic to tetragonal under strain .
this has complex effects that we will explain below . before giving actual calculations ,
we summarize those differences between the electrons in donor impurity states and in an artificial dot that affect @xmath118 .
the most obvious is the single - particle potential that binds the electron .
the gate potential is much smoother than the hydrogenic potential of the impurity .
this implies that the corrections to the effective mass approximation are much weaker , and @xmath119 will be much reduced .
it is difficult to compute the energy splittings precisely , but considerations based on the method of sham and nakayama @xcite give splittings in the range @xmath120 in the structure of friesen et al . @xcite .
this increases the relaxation rate .
( in fact a naive estimate of the enhancement is a factor of 400 . )
on the other hand , the structures we consider have strong lattice strain .
this partly lifts the valley degeneracy and also reduces the matrix elements , which decreases the rate .
another aspect of some of the proposed designs is the presence of ge with its much stronger soc .
this will act to decrease the spin relaxation time .
we consider first the case of pure si under uniaxial strain .
the ingredients of the calculation are as follows . from sec . [
sec : well ] we have the solutions to the schrdinger equation @xmath121\phi_{n0}(\vec{r } ) = e_{n}\phi_{n0}(\vec{r})$ ] .
@xmath122 is the unperturbed crystal hamiltonian without soc and it has the full space group symmetry .
@xmath123 is the gate and/or impurity potential . to calculate @xmath12
, we must also include soc , which we treat as a perturbation : @xmath124 .
the resulting states @xmath125 are two - fold degenerate because of time - reversal symmetry .
let us denote these states as @xmath126 and @xmath127 .
they are not eigenstates of spin , so the arrows denote pseudo - spin
. we may define the pseudo - up state as that which evolves from the spin - up state as spin - orbit coupling is turned on adiabatically , and similarly for the pseudo - down state .
because of valley and pseudospin degeneracy , there are 2 ground states @xmath128 and 10 excited states @xmath129 the 12-fold degeneracy in the effective mass approximation is broken by central - cell corrections in the impurity case and smaller corrections in the quantum dot .
there is also the zeeman hamiltonian of the external field @xmath130 in the field @xmath131 and @xmath127 are no longer degenerate . note that the energy splitting may depend on the direction of @xmath132 .
finally , we have the electron - phonon coupling hamiltonian @xmath133 a phonon represents a time - dependent perturbation .
this will create transitions whose rate is given by the fermi golden rule .
we are interested in the transitions between @xmath134 and @xmath135 however , @xmath136 in the absence of the external field .
thus we need to calculate in next order in perturbation theory using an effective hamiltonian@xmath137 + \left [ \left . \mathcal{h}% _ { ep}|\phi _ { rs}(\vec{r})\right\rangle \left\langle \phi _ { rs}(\vec{r})|% \mathcal{h}_{z}\right .
\right ] \right\}\]]here @xmath138 runs over the excited states , @xmath139 .
@xmath140 .
the relaxation time is given by @xmath141 where @xmath142 @xmath143 $ ] is the rate for transitions from the higher - energy ( pseudo - spin down ) state to the lower - energy ( pseudo - spin up ) state@xmath144and @xmath145 is the rate for transitions from the lower - energy ( pseudo - spin up ) state to the higher - energy ( pseudo - spin down ) state@xmath144where the sum is over phonon modes @xmath146 with energies @xmath147 .
a thermodynamic average over the lattice states has been taken .
it yields the bose occupation factors @xmath148 for the phonons .
the matrix elements of @xmath149 are computed as follows .
the expectation value of the external field @xmath150 can be written as@xmath151 \cdot \vec{\sigma}_{s , s^{\prime } } \nonumber \\ & \equiv & \mu _ { b}\vec{b}\cdot \mathbf{d}_{nn^{\prime } } \cdot \vec{\sigma}% _ { s , s^{\prime } } \,\ , \label{eq : ddef}\end{aligned}\]]and the tensor @xmath152 is the effective @xmath10 factor at the @xmath153th valley .
this equation defines the tensor @xmath154 that characterizes the coupling of the various states by the external field .
the principal axes of the @xmath10 tensor are the same as that of the effective mass tensor at the @xmath153th valley .
it has the form@xmath155 there are only two independent constants .
a simple example of the diagonal part of the * d * tensor is that for the ground state of an impurity in the _ unstrained _ lattice when the central cell corrections are included .
then we have @xmath156 and @xmath157 is proportional to the unit matrix:@xmath158with@xmath159 the matrix elements between ground and excited states have the form @xmath160with @xmath161and the tensor @xmath162 is defined by@xmath163where @xmath164 is the
local anisotropy axis .
if the original @xmath10 were isotropic , then @xmath165 and there would be no coupling between different states and no spin relaxation . if the lattice is strained , then the @xmath98 coefficients become strain - dependent and the general expression for @xmath166 from eq .
eq : ddef must be used .
uniaxial strain lifts the degeneracy of the valleys .
we include this effect in the hamiltonian and it determines the proper combinations of the @xmath167 defined in eq .
[ eq : psi6 ] .
these then feed into @xmath168 as a function of strain the @xmath169 for the ground state cross over from the completely symmetric combination @xmath170 to the combination @xmath171 , @xmath172 in the limit of large strain.@xcite the phonons involved are just the acoustic ones , one longitudinal and two transverse - these are the only ones with low enough energy to play a role in relaxing the spins
. the matrix elements of the electron - phonon interaction are only nonzero within one valley and for a single phonon mode they are conventionally parametrized as:@xmath173near the @xmath153th valley , where @xmath174 and @xmath175 is the polarization vector .
@xmath176 destroys a phonon with wavevector @xmath177 and polarization @xmath178 . once again
, we see that the interaction can be characterized by just 2 parameters , in this case @xmath179 and @xmath180 , as already defined in eq .
[ eq : xi ] . performing the integration over the envelope function at wavector @xmath177 now gives@xmath181where @xmath182thus
the electron - phonon interaction involves a form factor for the bound states .
since @xmath1 is normalized , we have @xmath183 when the wavelength of the phonon is much longer than the spatial extent @xmath184 of the bound state : @xmath185 .
the calculations of [ sec : well ] indicate that this is the case .
it is also independent of @xmath186 in the golden rule calculation , the energy denoninator @xmath187 will suppress contributions from the excited states of @xmath188 .
thus we will keep only states that are split off from the ground state by corrections to the effective mass approximation .
this approach works very well in si : p and should be even better for the quantum dot .
this produces the golden - rule transition rate@xmath189 ^{2}\sum_{\vec{q}\lambda } \,a^{2}(\vec{q}% ) \,\delta ( e_{g\uparrow } -e_{g\downarrow } -\hbar \omega _ { \vec{q}\lambda } ) \left\langle a_{\vec{q}\lambda } a_{\vec{q}\lambda } ^{\ast } \right\rangle \times \nonumber \\ & & \left| \sum_{r=2}^{6}\frac{\vec{b}\cdot \mathbf{d}_{gr}\cdot \vec{\sigma}% _ { \uparrow \downarrow } \,\,\widehat{e}_{\lambda } ( \vec{q})\cdot \mathbf{d}% _ { gr}\cdot \vec{q}\,\delta _ { ss^{\prime } } } { e_{g}-e_{r}}\right| ^{2}.\end{aligned}\ ] ] we approximate the phonon dispersion as @xmath190 . setting @xmath191 , performing the integral over the magnitude of @xmath177 , and repeating the calculation for @xmath192
, we obtain a total spin relaxation time @xmath193\,g_{g}^{3}\mu _ { b}^{3}b^{3}\times \nonumber \\ & & \sum_{\lambda = 1}^{3}\frac{1}{v_{\lambda } ^{5}}\int_{0}^{2\pi } d\phi ^{\prime } \int_{0}^{\pi } \sin \theta ^{\prime } d\theta ^{\prime } \,\left| \sum_{r=2}^{6}\left [ \vec{b}(\theta , \phi ) \cdot \mathbf{d}_{gr}\cdot \vec{% \sigma}_{\uparrow \downarrow } \right ] \left [ \frac{\,\,\widehat{e}_{\lambda } ( \vec{q})\cdot \mathbf{d}_{gr}\cdot \widehat{q}\,}{e_{g}-e_{r}}(\theta ^{\prime } , \phi ^{\prime } ) \right ] \right| ^{2}.\end{aligned}\ ] ] here @xmath194 are the polar axes of the direction of @xmath195 measured from the @xmath196 $ ] direction of the crystal .
@xmath197 is the mass density .
this is our basic result , in a form very similar to that given by hasegawa @xcite . in fig .
[ fig : energies ] the effect of strain on the lowest energy levels is shown . here
the zero - strain splittings are those of a p impurity .
a similar plot was given in ref .
@xcite , but at that time the correct zero - strain splittings were not known . in the dot , the energy splittings for the unstrained case would be two to three orders of magnitude smaller ( see below for a discussion ) . in fig .
[ fig : strain ] , we show the effect of strain on the relaxation time of an electron in a p impurity potential .
there are two effects : the overall increase of the energy denominators and the change of the ground state to a less symmetric valley weighting .
this leads to a non - monotonic dependence of @xmath12 on strain , but in the region of interest , the effect of strain is to greatly increase @xmath12 , since most proposed structures have @xmath198 at large strain , only one energy denominator remains small , that between the ground state , symmetric in the @xmath199 valleys , and the first excited state , antisymmetric in the @xmath199 valleys .
the overlap matrix * d * is very small between these two state .
this reduction of the matrix element is the dominant effect . also of interest
is the dependence of @xmath12 on the angle of the external field , since this may serve as a diagnostic tool in experiments to verify that the relaxation process is really due to spin - phonon coupling . as a function of strain ,
this dependence becomes highly anisotropic , as seen in fig .
[ fig : anisotropy ] .
this is due to the elimination of all but the @xmath200 valleys from the problem at high strain .
we note that the limiting value @xmath201 when the external field is along a crystal axis is cut off by intervalley scattering effects not included in the present calculation @xcite .
the change in the confining potential reduces the corrections to the effective mass approximation , as discussed in sec .
[ sec : decoherence ] which in turn reduces the energy denominators , leading to a decrease in @xmath118 . in the friesen _
_ structure , we have estimated the splitttings @xmath202 using the method of sham and nakayama @xcite , and they range from 0.05 to 0.1 mev , depending on the gate voltages that produce the potential .
this may be incorporated into the calculation of the @xmath203 by introducing a variable coupling @xmath204 that mixes the @xmath205 and @xmath206 valley wavefunctions with the @xmath199 valley wavefunctions . for a precise definition of @xmath204 ,
approximately , however , @xmath207 .
the results of the calculations are shown in fig .
[ fig : deltac ] .
the graphs show the absolutely crucial role that strain plays in the determination of @xmath12 . because of the small energy denominators in the dot , the rate is extremely fast at small @xmath204 for the unstrained case : @xmath208 we also point out that in stuctures where @xmath12 appears to be too long for efficient preparation of the spins ( as discussed in sec .
sec : introduction ) the deficiency can be made up by increasing the temperature @xmath209 .
as seen in fig .
[ fig : temperature ] , @xmath12 decreases very rapidly as @xmath209 is increased .
actually , this calculation even underestimates the decrease , since multiphonon processes begin to contribute at about 3k .
however , it must be borne in mind that the temperature must be small enough that the initial state is essentially pure ( all spins up , for example ) .
a more practical method of decreasing @xmath27 if desired , would be to increase the magnetic field .
the effect of alloying with ge is to increase the soc and hence to increase @xmath210 .
first principles calculations of ge impurities in si have shown that there is little effect on the states near the bottom of the conduction band , though this is not necessarily the case for higher energy states in the band @xcite .
this is in accord with the isoelectronic character of the atoms in the alloy .
it is then reasonable to employ the virtual crystal approximation ( vca )
. the approximation should be quantitatively accurate for small , ( say @xmath211 ) concentrations of ge , but may be taken as a good qualitative guide also to higher concentrations .
the bloch functions in the absence of soc satisfy@xmath212where@xmath213where @xmath214 is the number of unit cells , @xmath215 labels the unit cells , @xmath216 labels the two positions in the unit cell , @xmath217 labels the 4 atomic states , @xmath218 are the atomic orbitals , and the coefficients @xmath219 give the proper linear combination of atomic orbitals ( lcao s ) for the state at momentum @xmath220 and band @xmath221 .
the normalization condition is@xmath222 in the vca , the @xmath219 are the same for the si and ge sites .
the soc hamiltonian for the alloy is written as @xmath223 where @xmath224 is the soc strength for si ( ge ) when @xmath225 is a si ( ge ) site .
the energy shift of an electron in the external field @xmath226 can be written as@xmath227where@xmath228 and @xmath229 is the completely antisymmetric symbol with @xmath230 .
our approximation now consists in regarding the dependence on the ge concentration as coming only in the term @xmath231 .
hence any component of the @xmath10-tensor is proportional to a weighted average of @xmath232 and @xmath233 in particular , consider the conduction band minimum at @xmath234 in the compound si@xmath53ge@xmath235 then we have @xmath236% \left [ ( 1-x)+x(\lambda _ { ge}/\lambda _ { si})\right ] , $ ] where @xmath237 is the value for pure silicon .
the numbers , known from atomic physics , are @xmath238 , so @xmath239 \left [ ( 1-x)+6.71x\right ] .$ ] similarly , we have @xmath240 \left [ ( 1-x)+6.71x\right ] $ ] , etc . only the overall scale of @xmath10 changes , not the anisotropy in the vca .
the presence of ge in the lattice breaks the translation symmetry , an effect that is neglected in the vca . taking this into account would lead to a @xmath10-factor that is averaged over momentum space rather than one that is evaluated only at the si conduction band minima .
if we take the l - minimum of pure ge as an example , we would expect that the corrections to @xmath241 would be greater if there is some averaging over momentum space .
thus a calculation going beyond the vca would give corrections that would reduce @xmath12 .
we show the results for differing ge concentrations in the active region of the well in fig .
[ fig : germanium ] . in the ge - poor regime @xmath242 ,
the relaxation rate @xmath114 decreases fairly slowly .
thus designs with a si - rich well will not suffer from very fast spin relaxation on the minority ge sites .
by contrast , if we cross the critical point around @xmath243 where the conduction band minima switch positions , then there is a very rapid decrease of @xmath12 .
this is due to the much greater g - factor anisotropy of the [ 111 ] minima : @xmath244 , ( as compared to @xmath245 in si ) .
the results for the ge - rich region are much simpler than in the si - rich region .
uniaxial strain does not affect the relative valley energies or the matrix elements parametrized by * d*. the only effect of going to the dot from the impurity is to reduce corrections to the effective mass approximation , which will strongly decrease the relaxation time . indeed , if we take the impurity result for ge : p from ref .
hasegawa , which is @xmath246 and estimate the decrease in @xmath119 as about a factor of 100 , then we obtain @xmath247 to @xmath248 which is vastly shorter than that of si - rich structures .
the physics is the same as that which governs the divergence of the rate in fig .
[ fig : deltac ] for the unstrained case , namely the much smaller energy denominators . to make contact with sec .
[ sec : well ] note that if there is ge in the barrier regions , as is the case in most sige designs , then we will have an envelope function that weights si and ge differently , then @xmath52 in the vca formulas must be replaced by @xmath249 , where @xmath250 .
this is the product of the ge concentration in the layer , times the probability that the electron is on the layer , summed over layers . referring to figs .
[ fig : donor ] and [ fig : qw ] , we see that @xmath84 is always less than @xmath251 for structures that have pure si active layers and ge in the barrier layers . the effect on @xmath12 is small so these structures are phonon - safe .
.materials parameters used in this work . [ cols="^,^,^",options="header " , ]
decoherence due to spin relaxation by emission and absorption of single phonons does not pose a major obstacle to quantum computation in quantum dot sige heterostructures using spin qubits , if these qubits are properly designed .
this spin relaxation mechanism is the dominant one in the case of electrons in donor bound states , such as in si : p . in order
that this source of decoherence be kept under control , however , certain conditions must be satisfied in the design of the structure .
unstrained sige alloys generally have relatively short spin relaxation times due to valley degeneracy .
this degeneracy produces strong spin mixing in the kramers - degenerate ground states of the quantum dot once the field is turned on .
this in turn increases the electron - phonon matrix element that connects the two states and decreases the relaxation time .
this effect , fundamentally due to spin - orbit coupling , is stronger in the quantum dot than in the impurity state .
the corrections to effective - mass theory , stronger for the impurity potential , tend to lift the degeneracy .
the crucial role that strain plays , in si - rich structures , is to lift the valley degeneracy . in this respect
compressive uniaxial strain is the best , since the residual degeneracy is reduced to two valleys .
the wavefunction for the ground state is then symmetric in the valley index , and the first excited state , the only one with a small energy denominator , is antisymmetric in this index .
this fact strongly suppresses the spin mixing effect of the external field
. thus a workable design should include uniaxial strain that exceeds a critical value .
this value is small enough that there will be sufficient strain in most practical structures in which the active layer is si rich and sandwiched between relatively ge - rich layers
. the active layer must be thin enough to avoid dislocation - mediated relaxation .
spin - orbit coupling is stronger in ge than in si due to the higher atomic number .
this suggests that si - rich structures are to be preferred , and the calculations bear this out .
the relaxation time decreases with increasing @xmath252 in a si@xmath53ge@xmath54 alloy .
in fact , if @xmath52 increases beyond @xmath243 , decoherence becomes very strong , and quantum dots in these ge - rich structures , such as those proposed by vrijen _
_ et al .
@xcite , _ _ may well run into difficulties for this reason .
a key point for these structures is that uniaxial stress does not lift the valley degeneracy , since the valleys have moved off the crystal axes .
strain - induced suppression of spin relaxation can not occur . beyond design issues ,
fabrication quality is also important .
even a small concentration of magnetic impurities will negatively impact the spin relaxation . in modern semiconductor technology , however , concentrations of magnetic impurities much less than 0.1 ppm are routine .
the impact of lattice imperfections is less clear .
these will generally act to lower the symmetry of the system , and to lessen the accuracy of the effective - mass approximation . as we have seen in the context of the impurity calculations
, these effects generally increase the spin relaxation time because of reduced degeneracy .
on the other hand , localized bound states can form at such imperfections .
if this results in free electron spins , it could have seriously negative effects on the relaxation time .
similar effects would result from lattice defects that produce localized phonon modes .
finally , it appears that quantum dot designs will need rather low temperatures in order to operate .
the impurity experiments clearly indicate that multi - phonon relaxation increase rapidly with temperature tgcastner , with a crossover to this regime at about 3 k. we speculate that this crossover temperature is pushed up when the effects of strain are included , but we have not yet done the requisite calculations .
we would like to acknowledge useful discussions with our collaborators at the university of wisconsin - madison and with xuedong hu .
this work was supported by the national science foundation under the materials theory program , grant no .
dmr-0081039 , the materials research science and engineering center program , grant no .
dmr-0079983 , and the army research office . , for finding a donor - bound electron at a ge site , as a function of ge content , @xmath52 .
the simulated structure , design ( 1 ) , is shown in the inset .
a strained si quantum well of thickness 6 nm is sandwiched between relaxed si@xmath53ge@xmath253 barrier regions of thickness 20 nm .
the electron is bound to a p@xmath254 ion embedded in the center of the quantum well . ] , for finding an electrostatically bound electron at a ge site , as a function of quantum well thickness , @xmath93 .
the inset shows the heterostructure layers for design ( 2 ) , beginning at bottom : a thick , doped semiconductor back - gate , a relaxed si@xmath53ge@xmath253 barrier layer , a strained si quantum well , a thick , relaxed si@xmath53ge@xmath253 barrier layer , and lithographically patterned metallic top - gates .
the distance between back- and top - gates is held fixed at 40 nm , while the quantum well , of variable thickness , @xmath93 , is centered 15 nm above the back - gate . ] with an uniaxial stress applied in the [ 100 ] direction .
( it is important to note for clarity that what we call compression relative to the [ 100 ] or growth direction is equivalent to tension in the plane of the quantum well .
this label convention differs across fields . ) for reference , @xmath255 corresponds to the compressive ( s negative ) strain caused in a pure silicon layer by a si@xmath256ge@xmath257 sublayer .
the energies are expressed in ev and the numbers in parenthesis indicate the degeneracy of the level .
the a@xmath258 ( ground state , solid line ) and t@xmath258 level wave functions are mixed by strain ( causing valley population intermixing ) and the relaxation rate is proportional to @xmath259 .
thus , with all else equal , an increase in strain causes the relaxation rate to decrease . for a detailed analysis
see wilson and feher @xcite , where @xmath3 here is their @xmath260 times 100 or koiller _ et al . _
@xcite whose notation is @xmath261 . ]
uniaxially strained silicon versus strain parameter @xmath3 for a temperature of 3 kelvin and a magnetic field @xmath262 of 1 tesla in the [ 111 ] ( @xmath263 ) and [ 110 ] ( @xmath264 , @xmath265 ) directions respectively .
@xmath255 corresponds to the strain caused in the pure silicon layer by a si@xmath256ge@xmath257 sublayer . ]
uniaxial strain @xmath255 , corresponding to the strain caused in the pure silicon layer by a si@xmath256ge@xmath257 sublayer .
the dashed line is the unstrained case .
the rate is plotted for varying the intervalley coupling constant @xmath267 .
@xmath267 corresponds to 1/6th the energy splitting between the singlet ( ground state ) and triplet ( next excited state ) valley energies _ in unstrained si_. as a result , it controls the mixing between the @xmath205 and @xmath206 valley wavefunctions with the @xmath199 valley wavefunctions .
the magnetic field is set at @xmath266 t along the [ 111 ] direction and the temperature is 3 k. ] uniaxially strained silicon ( @xmath255 , corresponding to the strain caused in the pure silicon layer by a si@xmath256ge@xmath257 sublayer ) versus ( a ) temperature with the magnetic field set at @xmath266 t and ( b ) magnetic field strength with the temperature set at 3 k. ] uniaxially strained silicon versus concentration of germanium within unstrained ( left vertical axis ) and compressively strained ( right vertical axis , corresponding to a si@xmath256ge@xmath257 sublayer ) pure silicon . the magnetic field is set at @xmath266 t along the [ 111 ] direction and the temperature is 3 k. here we assume that the addition of germanium does not affect the strain within the bulk and only acts to increase the @xmath10-factor and spin orbit coupling .
we expect this approximation to hold for small concentrations of germanium and to break down with increasing concentration as the conduction band valley minima switch from the x type of silicon to the l type of germanium . ] | direct phonon spin - lattice relaxation of an electron qubit bound by a donor impurity or quantum dot in sige heterostructures is investigated .
the aim is to evaluate the importance of decoherence from this mechanism in several important solid - state quantum computer designs operating at low temperatures .
we calculate the relaxation rate @xmath0 as a function of [ 100 ] uniaxial strain , temperature , magnetic field , and silicon / germanium content for si : p bound electrons .
the quantum dot potential is much smoother , leading to smaller splittings of the valley degeneracies .
we have estimated these splittings in order to obtain upper bounds for the relaxation rate . in general , we find that the relaxation rate is strongly decreased by uniaxial compressive strain in a sige - si - sige quantum well , making this strain an important positive design feature .
ge in high concentrations ( particularly over 85% ) increases the rate , making si - rich materials preferable .
we conclude that sige bound electron qubits must meet certain conditions to minimize decoherence but that spin - phonon relaxation does not rule out the solid - state implementation of error - tolerant quantum computing . |
a plane partition is a rectangular array of non - negative integers with weakly decreasing rows and columns .
the number of plane partitions contained in a @xmath0 rectangle with entries at most @xmath1 is given by macmahon s formula @xmath2 ( see @xcite ) .
as a variation of this , proctor proved a simple product formula for the number of plane partitions with entries at most @xmath1 which are contained in a shape with row lengths @xmath3 ( see corollary 4.1 in @xcite ) . from the hexagon @xmath4 by removing a maximal staircase from the west corner.,width=188 ] a _ lozenge tiling _ of a region on the triangular lattice is a covering of the region by unit rhombi ( or lozenges ) so that there are no gaps or overlaps .
we use notation @xmath5 for the number of lozenge tilings of a region @xmath6 ( @xmath7 ) .
the plane partitions contained in a @xmath8 rectangle with entries at most @xmath1 are in bijection with lozenge tilings of the hexagon @xmath4 of sides @xmath9 ( in cyclic order , starting from the north side ) .
in the view of this we have an equivalent form of proctor s result as follows .
[ proctor ] assume that @xmath10 are non - negative integer so that @xmath11 .
let @xmath12 be the region obtained from the hexagon @xmath4 by removing the maximal staircase " from its east corner ( see figure [ kuoqar5 ] for @xmath13 ) .
then @xmath14,\ ] ] where empty products are equal to @xmath15 by convention .
one can find more variations and generalizations of the proctor s result in @xcite .
we consider next a different generalization of theorem [ proctor ] . from a hexagon ( ( a ) and ( b ) ) .
two examples of the region @xmath16 ( ( c ) and ( d)).,width=491 ] let @xmath17 be the region described as in figure [ kuoqar6 ] . to precise , @xmath17 consists of all unit triangles on the right of the vertical symmetry axis of the hexagon of sides @xmath18 ( in cyclic order , starting from the north side ) .
figure [ kuoqar6](a ) illustrates the region @xmath19 and figure [ kuoqar6](b ) shows the region @xmath20 ( see the ones restricted by the bold contours ) .
we are interested in the region @xmath17 with @xmath21 up - pointing unit triangles removed from the base ( @xmath22 ) .
if the positions of the triangles removed are @xmath23 , then we denote by @xmath16 the resulting region ( see figures [ kuoqar6](b ) and ( c ) for @xmath24 and @xmath25 , respectively ) .
the number of lozenge tilings of the region @xmath26 is given by the theorem stated below .
[ main ] assume @xmath10 are non - negative integers . if @xmath27 for some non - negative integer @xmath21 , then @xmath28 where @xmath29 , for @xmath30 .
if @xmath31 for some non - negative integer @xmath21 , then @xmath32 where @xmath33 , for @xmath30 .
we note that by specializing @xmath34 and @xmath35 , for @xmath36 , the region @xmath12 is obtained from @xmath37 by removing forced lozenges on the lower - left corner ( see figure [ kuoqar7 ] ) .
thus @xmath38 and the proctor s theorem [ proctor ] follows from theorem [ main ] .
we enumerated the lozenge tilings of the region @xmath16 in @xcite under the name _ quartered hexagon _ ( see theorem 3.1 , equations ( 3.1 ) and ( 3.2 ) ) .
in particular , the region @xmath16 is obtained from the quartered hexagon + @xmath39 ( @xmath40 ) by removing several forced lozenges ( see figures 2.8 and 2.9(a ) in @xcite ) .
thus , we still call our @xmath17-type regions _ quartered hexagons_. in @xcite , we identified the lozenge tilings of @xmath16 with certain families of non - intersecting paths on @xmath41 , then used lindstrm - gessel - viennot theorem ( see e.g. @xcite , lemma 1 ; or @xcite , theorem 1.2 ) to turn the number of path families to the determinant of a matrix whose entries are binomial coefficients , and evaluated the determinant . from the region @xmath42.,width=302 ] a _ perfect matching _ of a graph @xmath43 is collection of edges so that each vertex of @xmath43 is incident to exactly one selected edge . the _ dual graph _
@xmath43 of a region @xmath6 on the triangular lattice is the graph whose vertices are unit triangle in @xmath6 and whose edges connect precisely two unit triangles sharing an edge .
we have a bijection between the tilings of a region @xmath6 and the perfect matchings of its dual graph @xmath43 .
we use notation @xmath44 for the number of perfect matching of a graph @xmath43 . in this paper
, we give a new inductive proof of theorem [ main ] by using proctor s theorem [ proctor ] as a base case .
the method that we use in the proof is the _ graphical condensation method _ first introduced by eric kuo @xcite . in particular
, we will employ the following theorem in our proof .
let @xmath45 be a planar bipartite graph , and @xmath46 and @xmath47 its vertex classes .
assume that @xmath48 are four vertices appearing on a face of @xmath43 in a cyclic order .
assume in addition that @xmath49 , @xmath50 , and @xmath51
. then @xmath52
we only prove ( [ eq1 ] ) , as ( [ eq2 ] ) can be obtained by a perfectly analogous manner .
it is easy to see that if @xmath53 then the region @xmath16 has only one tiling ( see figure [ kuoqar8](a ) ) . on the other hand , since now @xmath54$]$ ] for the set @xmath55 of all positive integers not exceed @xmath21 . ] , the right hand side of the equality ( [ eq1 ] ) is also equal to @xmath15 , then ( [ eq1 ] ) holds for @xmath53 .
moreover , if @xmath56 , then @xmath57 , @xmath58 , and the region has the form as in figure [ kuoqar8](b ) . in this case , the region has also a unique tiling ; and it is easy to verify that the right hand side of ( [ eq1 ] ) equals 1 .
thus , we can assume in the remaining of the proof that @xmath59 .
we will prove ( [ eq1 ] ) by induction on @xmath60 .
.,width=491 ] if @xmath61 , then we have @xmath62 .
it is easy to see that there are only 2 possible shapes for the region @xmath17
( i.e. the region before removed triangles from the base ) as in figures [ kuoqar8](c ) and ( d )
. then it is routine to verify ( [ eq1 ] ) for @xmath62 . for the induction step ,
we assume that ( [ eq1 ] ) is true for any region with @xmath63 , for some @xmath64 , then we need to show that it is also true for any region @xmath26 with @xmath65 .
let @xmath66 be a set of positive integers , we define the operators @xmath67 and @xmath68 by setting @xmath69 and @xmath70 then one can re - write ( [ eq1 ] ) in terms of the above operators as : @xmath71)}\frac{\bigstar(s)}{\bigstar([k+c])},\ ] ] where @xmath72 and @xmath73:=\{1,2,\dotsc , k+c\}$ ] . from this stage we will work on this new form of the equality ( [ eq1 ] ) .
we first consider three special cases as follows : 1 . if @xmath74 , then by considering forced lozenges as in figure [ kuoqar8](f ) , we get @xmath75 where @xmath76
. then ( [ eq1a ] ) follows from the induction hypothesis for the region on the right hand side of ( [ eq3 ] ) .
2 . if @xmath58 , then @xmath77 ; and we get the region @xmath78 is obtained from the region @xmath79 by removing forced lozenges along its base . thus ,
( [ eq1a ] ) follows from proctor s theorem [ proctor ] .
3 . let @xmath80 ( so @xmath81 ) .
we consider one more special case when @xmath82 . by removing forced lozenges again
, one can transform our region into the region @xmath83 , then we get again ( [ eq1a ] ) by induction hypothesis for the latter region ( see figure [ kuoqar8](e ) ) . from now on , we assume that our region @xmath26 has the two parameter @xmath21 and @xmath84 positive ( so @xmath85 ) , and that @xmath86 .
now we consider the region @xmath6 obtained from @xmath26 by recovering the unit triangle at the position @xmath87 on the base .
we now apply kuo s theorem [ kuothm ] to the dual graph @xmath43 of @xmath6 , where the unit triangles corresponding to the four vertices @xmath48 are chosen as in figure [ kuoqar4 ] ( see the shaded triangles ) . in particular , the triangles corresponding to @xmath88 and @xmath89 are at the positions @xmath87 and @xmath90 on the base ; and the ones corresponding to @xmath91 are on the upper - right corner of the region .
.,width=377 ] one readily sees that the six regions that have dual graphs appearing in the equation ( [ kuoeq ] ) of kuo s theorem have some lozenges , which are forced into any tilings .
luckily , by removing such forced lozenges , we still get new regions of @xmath17-type . in particular , after removing forced lozenges from the region corresponding to @xmath92 , we get the region @xmath93 ( see the region restricted by bold contour in figure [ kuoqar9](a ) ) .
this implies that @xmath94 similarly , we have five more equalities corresponding to other graphs in ( [ kuoeq ] ) : @xmath95 @xmath96 @xmath97 @xmath98 @xmath99 plugging the above six equalities ( [ eqn1 ] ) ( [ eqn6 ] ) in ( [ kuoeq ] ) , we have the following recurrence @xmath100 the five regions other than @xmath101 in the above recurrence ( [ recu1 ] ) have their @xmath102-parameter less than @xmath103 .
therefore , by induction hypothesis , we get @xmath104)}\frac{\bigstar(s\cup\{d\}-\{s_1\})}{\bigstar([k+c])},\ ] ] @xmath105)}\frac{\bigstar(s-\{s_{k+c}\})}{\bigstar([k+c-1])},\ ] ] @xmath106)}\frac{\bigstar(s-\{s_1\})}{\bigstar([k+c-1])},\ ] ] @xmath107)}\frac{\bigstar(s\cup\{d\}-\{s_{k+c}\})}{\bigstar([k+c])},\ ] ] @xmath108)}\frac{\bigstar(s\cup\{d\}-\{s_1,s_{k+c}\})}{\bigstar([k+c-1])}.\ ] ] by the above five equalities ( [ eqm1 ] ) ( [ eqm5 ] ) and the recurrence ( [ recu1 ] ) , we only need to show that @xmath109 and ( [ eq1a ] ) follows .
first , we want to simplify the first ratio in the first term on the right - hand side of ( [ eqfinal ] ) .
we re - write it as @xmath110 the ratio @xmath111 has its numerator and denominator almost the same , except for some terms involving @xmath87 or @xmath90 .
canceling out all common terms of the numerator and the denominator , we have @xmath112 similarly , we get @xmath113 and @xmath114 thus , we obtain @xmath115 by the same trick , we can simply the first ratio in the second term on the right - hand side of ( [ eqfinal ] ) as @xmath116 next , we simply the second ratio in each term on the right - hand side of ( [ eq1a ] ) . by replacing the operator @xmath67 by the operator @xmath68 , the whole simplifying - process works in the same way with each term @xmath117 replaced by @xmath118 .
thus , we get @xmath119 and @xmath120 finally , by ( [ eqsimple1])([eqsimple4 ] ) , we can simplify the equation ( [ eqfinal ] ) to @xmath121 which is obviously true with @xmath122 .
this means that ( [ eqfinal ] ) holds , and so does ( [ eq1 ] ) .
the author wants to thank christian krattenthaler for helpful comments , and ranjan rohatgi for useful discussion . | it has been proven that the lozenge tilings of a quartered hexagon on the triangular lattice are enumerated by a simple product formula . in this paper
we give a new proof for the tiling formula by using kuo s graphical condensation .
our result generalizes a proctor s theorem on enumeration of plane partitions contained in a maximal staircase " .
tilings , perfect matchings , plane partitions , graphical condensation .
05a15,05c70 , 05e99 |
understanding the distribution of mass on large scales is a fundamental quest in cosmology .
how does it compare with the distribution of stars , light , and gas ?
we know that individual galaxies are surrounded by large dark matter halos , and that groups and clusters of galaxies are dominated by dark matter comprising five to ten times more mass than baryonic matter ( stars and gas ) .
it is generally believed that the relative contribution of dark matter increases with scale from galaxies , to groups , to clusters , and to large - scale structure ; larger scale systems are believed to contain more dark matter , relative to light or stars , than do galaxies ( @xcite ; and more recently @xcite ) . in this paper
we investigate the distribution of mass , light , and stars and explore how light and stars trace mass as a function of environment and scale .
we discuss the implications for the disribution of dark matter and for the stellar mass fraction .
one of the classical methods to investigate the distribution of mass in the universe is to compare it directly with the distribution of stellar light : how does mass follow light ?
this method was first used on large scales by @xcite ; this was followed by many additional investigations ( see @xcite , and references therein ) .
the comparison between the distribution of mass and light is represented by the mass - to - light ratio , @xmath3 , which can be studied as a function of scale and environment .
@xcite , @xcite , and others showed that @xmath3 increases systematically with scale , from the small scale of galaxies to the larger scale of groups , clusters , and larger scale structure , indicating a growing dominance of dark matter with scale ( see also @xcite , and references therein ) .
@xcite investigated the overall @xmath2 function from galaxies to large scales , separating galaxies into ellipticals ( older ) and spirals ( younger ) .
they showed that @xmath2 rises from the small scales of galaxies up to a few hundred kpc , reflecting the large dark matter halos around galaxies , then flattens to a constant value , where light approximately traces mass , with comparable relative contribution of dark matter on all scales .
@xcite thus suggest that most of the dark matter in the universe may be located in the large dark matter halos around individual galaxies and that the dark matter in groups and clusters may simply be the sum of their individual galaxy members ( plus gas ) .
more recently , @xcite used gravitational lensing observations of clusters from the sloan digital sky survey ( sdss ) and found a similar trend of increasing @xmath3 with radius that flattens on scales of several mpc .
the flattening of @xmath3 on large scales indicates that light follows mass on these scales and the @xmath3 ratio approaches a mean cosmic value @xcite .
in this paper we use the sdss observations of weak lensing mass and the observed distribution of light around 132,473 stacked groups and clusters of galaxies at @xmath11 @xcite to analyze the mass - to - light profile as a function of environment , from small groups of a few galaxies to the richest clusters , and as a function of scale , from the small scale of 25 kpc inside clusters to large cosmic scales of @xmath12 mpc .
these scales reach to @xmath13 virial radii of the systems , well into the large - scale cosmic environment .
we investigate how light traces mass on these scales , how the @xmath3 profile is affected by the varying stellar population , and how the stars trace the underlying mass distribution .
we estimate the approximate stellar mass fraction as a function of environment and scale . we show that light and stars trace mass on scales above several hundred kpc , and that the mean stellar mass fraction is nearly constant on all these scales in all environments , @xmath7 ( [ d4 ] ) .
we further show that most of the dark matter may be located in the large halos around individual galaxies ( [ d3 ] ) .
we discuss the data in [ data ] and the analysis and results in [ results ] .
we investigate the contribution of individual galaxies to the total @xmath14 function in
[ d3 ] , determine the distribution of the stellar mass fraction in
[ d4 ] , and calculate @xmath15 in [ d5 ] .
we summarize our conclusions in [ conclusions ] .
we use a flat @xmath16cdm cosmology with @xmath17 , @xmath18 , and @xmath19 km / s / mpc ( where @xmath20 should be inserted for @xmath16cdm ; @xcite ) .
the maxbcg cluster catalog @xcite was obtained from the sloan digital sky survey ( sdss ; @xcite ) data release 4 @xcite .
the cluster finder is based on the red - sequence method ; it maximizes the likelihood that a galaxy is a brightest cluster galaxy ( bcg ) at the center of an overdensity of red - sequence galaxies @xcite .
all clusters are selected from a 7500 deg@xmath21 region on the sky and are in the photometric redshift range @xmath11 .
the cluster richness @xmath22 is defined by the number of galaxies on the red sequence with rest - frame @xmath23-band luminosity @xmath24 located within a radius @xmath25 from the bcg ( where the @xmath23-band @xmath9 is the @xmath26 value from @xcite , corresponding to @xmath27 , and @xmath25 is the radius within which the local galaxy overdensity is 200 ; see @xcite ) .
the radius @xmath25 , which is used only to define the cluster richness @xmath28 , differs from the @xmath29 we use throughout this paper , which defines a mass overdensity of 200 times the critical density .
the published catalog contains 13,823 clusters with @xmath30 ; @xcite ( hereafter ) augment the catalog to include small groups of galaxies with @xmath31 , resulting in a sample of 132,473 groups and clusters with @xmath32 .
the cluster photometric redshifts are accurate to @xmath33 , with a scatter of @xmath34 for @xmath30 , degrading to @xmath35 for the poorest systems , with the same accuracy .
the cluster sample was partitioned into 12 richness bins ranging from small groups with @xmath36 to the richest clusters with @xmath37 ( @xcite ; hereafter ) .
the number of clusters per bin decreases from 58,788 for the poorest bin to 47 for the richest .
the weak lensing mass measurements were carried out by and on the stacked clusters in each richness bin , all centered on the cluster bcg . because lensing is not sensitive to a uniform mass distribution ( `` mass sheet '' ) , the measured mass reflects the mean mass density of the lens sample above the mean density of the universe .
the average tangential shear of background galaxies due to lensing by the foreground stacked clusters was measured by and used to calculate the mass density distribution around the clusters as a function of projected radius from the center out to scales of @xmath38 mpc . made corrections to the density profile for contamination of the lensed sample by cluster members and residual additive biases in the tangential shear .
the mass density contrast of the lensing clusters in each richness bin was measured in logarithmically spaced bins of projected clustercentric radius from @xmath39kpc to @xmath40mpc .
the 2d density contrast profile was then deprojected to a 3d density excess profile @xmath41 using an abel inversion ( see for details ) .
this measurement reflects the cluster - mass cross - correlation function times the mean density of the universe , @xmath42 .
this was used to calculate the cumulative excess mass @xmath43 within each radius @xmath44 , as well as to find @xmath29 , the radius within which the average mass density is 200 times the critical density .
we correct the mass profiles presented by @xcite ( hereafter ) for an improved photometric redshift distribution of the lensed background galaxies ; a full description is given in [ data5 ] .
the light distribution around the stacked clusters , using galaxy luminosities as defined below , was measured ( by ) as a function of radius for the same scales as above , from @xmath45 kpc to @xmath46 mpc .
a uniform background was subtracted using similar measurements around random points .
the luminosity measured reflects the luminosity density of the systems above the mean , @xmath47 , thus representing the cluster - light cross - correlation function as a function of radius , @xmath48 , comparable to the mass measurements described above .
all galaxy luminosities used are in the @xmath23-band , with k - correction applied to bring them to the mean cluster redshift of @xmath49 ; these are denoted @xmath50 .
all galaxy magnitudes are sdss model magnitudes .
a volume and magnitude limited sample of galaxies was chosen with @xmath51 and @xmath52 .
the projected luminosity profiles discussed above were inverted to obtain the 3d luminosity density profile , similar to the mass inversion .
this was then integrated to obtain the excess cumulative luminosity @xmath53 within the same clustercentric radial bins as above , from @xmath54kpc to @xmath1mpc . because the luminosity density was not measured within the innermost @xmath45kpc where the bcg is located , the average luminosity of the bcg in each stacked richness bin was added to the cumulative luminosity , so that the central @xmath55 is included in the light ( see for more details ) . the excess mass within radius @xmath44 from the center of the stacked clusters , @xmath56 , divided by the excess luminosity within radius @xmath44 , @xmath57 , gives the mass - to - light ratio @xmath58 \
, r^{2}}{\int_{0}^{r } { \mathrm{d}}r \ , [ \ell(r ) - \bar{\ell } ] \ , r^{2 } } = \frac{\int_{0}^{r}{\mathrm{d}}r \ , \bar{\rho } \,\xi_{cm } \,r^{2}}{\int_{0}^{r}{\mathrm{d}}r \ , \bar{\ell } \,\xi_{c\ell } \,r^{2}}.\ ] ] on small scales of virialized systems , this represents the mean cluster mass - to - light ratio . on large scales , as the density approaches the mean , this measures approximately the mean mass - to - light ratio of the universe and reflects how mass traces light on large scales . a small correction factor reflecting the bias of the galaxy light tracers relative to the mass relates the observed large - scale asymptotic @xmath59 to the mean cosmic value @xmath60 : ( ) _ asym = & b_m / l^-2 + b_m / l^-2 = & .
the bias factor @xmath61 depends on the ratio of the bias of clusters relative to mass and light , @xmath62 , which is near unity since the cluster bias essentially cancels out in this ratio .
the bias of the galaxy tracers relative to the mass , @xmath63 , is @xmath64 for galaxies near @xmath9 @xcite and varies very slowly for galaxies below @xmath9 @xcite .
since the luminosity threshold of the galaxy tracers is only @xmath65 ( and the mean luminosity of all the galaxies within @xmath66mpc is @xmath67 ) , this implies a bias of 1.05 ( for @xmath68 ; @xcite ) .
thus , on large scales , @xmath69 should approach a constant value , representing the cosmic mass - to - light ratio ( with only minor bias correction ) .
we correct the lensing masses of and using the correction from @xcite that reflects an improved treatment of the photometric redshift distribution of the background source galaxies based on a detailed analysis by @xcite .
this correction increases the lensing mass by @xmath70 ( _ stat . _ ) @xmath71 ( _ sys . _ ) .
we also correct the @xmath29 values by the small correction needed ( @xmath72 ) due to this mass increase . the mass and luminosity determinations discussed above use bcgs as the center for the stacked clusters .
if the bcg is slightly offset from the center of the cluster ( e.g. , @xcite ) , both the lensing mass and luminosity in the central regions will be slightly underestimated ( relative to the central mass and light of the cluster dark matter halo ) although the effect will partially cancel in the mass - to - light ratio .
the effect is strongest near the center of the clusters , as well as in poor groups for which miscentering is more likely to occur .
however , the effect is negligible on large scales .
discussion of the effect can be found in , @xcite and @xcite .
we do not correct for this effect as the corrections are uncertain and are typically small ( except in the poorest groups ) ; they are not important for our main goal of understanding the behavior of the mass - to - light profile on larger scales .
our results thus reflect the mass - to - light ratio around bcg galaxies , not necessarily around the center of the dark matter halos , especially for small groups . we further note that unless otherwise stated ( e.g. , [ d2 ] - [ d4 ] ) the luminosity accounts only for galaxies above the luminosity threshold of @xmath73 , as discussed above
. it does not include the entire luminosity of the systems .
this is discussed in [ d2 ] .
we combine the twelve stacked bins of groups and clusters ( [ data ] ) into three broader richness bins to study the overall @xmath74 as a function of radius around poor ( @xmath75 ) , intermediate ( @xmath76 ) , and rich ( @xmath77 ) systems .
we use the same radial bins as in .
we investigate both the cumulative @xmath78 , and the local , differential mass - to - light profile , @xmath79 . for simplicity
, we refer to these as @xmath80 and @xmath81 , respectively .
the integrated mass - to - light profile @xmath80 is presented in figure [ graph : graph100 ] as a function of radius from @xmath82kpc to @xmath83mpc for each of the three richness bins .
the observed increase of @xmath80 with radius on small scales is caused by the central bcg galaxy which dominates the cluster luminosity in the central regions ( typically @xmath84 kpc ) : these regions reflect primarily @xmath80 of the bcg galaxy .
the three top curves in figure [ graph : graph100 ] present @xmath80 when the mean bcg luminosity is excluded from each richness bin , revealing the impact of the bcg .
a comparison of the two sets of curves shows that the bcg luminosity accounts for the increasing @xmath80 in the central regions , and has negligible effect on the mass - to - light ratio on large scales . when the bcg light is excluded , the integrated mass - to - light profile is nearly flat on all scales , except for the rise on small scales that results from excluding the light ( i.e. , @xmath85 ) but not the mass .
if we further exclude the mass within the central region ( see below ) , @xmath86 flattens on all scales . on large scales
@xmath80 is essentially independent of cluster richness ; all environments , from the smallest groups to the richest clusters , exhibit the same overall mass - to - light ratio , reaching a universal value on scales of a few mpc . on intermediate scales of @xmath87mpc , while @xmath80 is nearly flat , small differences are observed as a function of richness and radius .
we discuss these trends below and show that they are consistent with the varying mean stellar population age as a function of richness and radius .
the rise in @xmath88 on small scales is considerably faster for rich clusters than for poor groups ( figure [ graph : graph100 ] ) .
rich clusters reach a nearly flat @xmath80 distribution at only @xmath89mpc ( typically near the edge of the bright bcg ) .
poor groups have a slow rise of @xmath80 with radius : the bcg luminosity dominates the group s luminosity and thus @xmath86 to scales of a few mpc .
this is because the bcg luminosity , while known to increase with cluster richness ( @xmath90 ; @xcite ) , is considerably more dominant in poor groups ( @xmath91 ; @xcite ) .
this is easily understood : in a poor group of a few galaxies , the bcg typically contains most of the group luminosity ; as the group s luminosity @xmath92 is integrated over radius , the bcg light dominates the group luminosity to large scales . in rich clusters , on the other hand , the bcg is brighter but constitutes only a small fraction of the total cluster luminosity at nearly all scales outside the innermost region . , for poor ( @xmath75 ) , intermediate ( @xmath93 ) , and rich ( @xmath77 ) clusters .
the dashed curves and related points show @xmath3 including both the light and mass of the central bcg , whereas the solid curves and related points show @xmath94 excluding the bcg light . ] , outside the central regions , with the bcg light and central mass excluded ( see [ results ] ) . ] because the effect of the bcg is dominant in the innermost regions , and our interest is mostly in understanding the general behavior of @xmath86 on larger scales , we present in figure [ graph : graph200 ] the cumulative @xmath95 profile excluding both the central bcg luminosity ( as in fig . [
graph : graph100 ] ) as well as the central mass . for the latter ,
we exclude all the mass within a central radius of 50 @xmath96 kpc for the poorest systems ( which have the smallest bcgs ) up to 150 @xmath96 kpc for the richest clusters . selecting somewhat different radii for the central mass exclusion
has only a small effect on the overall @xmath86 profiles ( e.g. , compare with figure [ graph : graph100 ] where no central mass has been excluded ) ; it does not affect the main results discussed below .
the @xmath86 profile outside the innermost regions ( figure [ graph : graph200 ] ) shows a nearly flat distribution on all scales , for all systems rich and poor , from the surprisingly small scale of few hundred kpc to the largest cosmic scales at 22 @xmath96 mpc .
the fact that the distribution is nearly flat on all scales indicates that stars , which make up only a few percent of the total mass , trace the distribution of mass well . while @xmath86 is nearly flat
, we observe a small trend with richness and with radius : @xmath86 increases slightly with richness at a given radius ( on scales of @xmath97 mpc ) , and decreases slightly with radius ( up to few mpc ) for all richnesses . as we show below
, these trends are consistent with the different mix of stellar populations old in e / s0 galaxies and younger in spiral galaxies as a function of radius and richness . following the density - morphology relation @xcite , the fraction of e / s0 galaxies is high in high - density regions ( rich clusters ) , and decreases with radius and richness to the lower density regions on larger scales and in poorer groups . since
early - type galaxies are dominated by an old stellar population with negligible recent star formation , their @xmath98 ratio is larger ( by a factor of @xmath99 ; see below ) than that of the younger stellar population of spiral galaxies , whose luminosities are dominated by bright young stars . as the fraction of spirals increases with radius , the integrated @xmath74 of the cluster decreases slowly with radius , as observed .
the small increase of @xmath74 from poor to rich clusters at a given radius ( figure [ graph : graph200 ] ) reflects partly the same effect : at a given radius a more massive cluster is generally at a smaller multiple of its virial radius , and thus a higher density , so it is dominated by a larger fraction of old e / s0s with higher @xmath3 . , outside the innermost cluster regions ( bcg mass and light excluded ) .
same as figure [ graph : graph200 ] , but plotted against the clustercentric radius in units of @xmath29 .
the dashed line shows the expected trend of the varying stellar population age as a function of scale ( [ results ] ) . the vertical band shows the location of the group and cluster correlation scale . ] , plotted against the clustercentric radius in units of @xmath29 .
the dashed line shows the expected trend of the varying stellar population ( see text ) . for clarity ,
only a representative error bar is shown . ]
figures [ graph : graph300 ] and [ graph : graph400 ] present the cumulative and local ( differential ) mass - to - light profiles , respectively , as a function of radius in units of the physical scale of the system , @xmath100 , from small scales to nearly 40@xmath29 . in figure
[ graph : graph300 ] , we again exclude the bcg luminosity and the innermost mass in order to better understand the overall trend of @xmath86 without the dominant impact of the bcg .
the differential analysis ( figure [ graph : graph400 ] ) is independent of the bcg , except in the very central region , since it reflects the local @xmath101 ratio at any radius .
the results in both figures highlight the similarity of @xmath102 for all systems and on all scales .
mass and light appear to follow each other especially when accounting for the stellar population age from deep inside clusters ( outside their bcg cores ) out to large cosmic scales .
figure [ graph : graph450 ] presents the observed profiles of the mass density , luminosity density , and galaxy number density for all environments as a function of @xmath100 .
the profiles are nearly identical , independent of richness .
this similarity of profiles is equivalent to the similar @xmath101 profile observed for all systems as shown above .
since @xmath100 corresponds to a similar overdensity for all systems , no significant stellar population effect is expected between the different richness bins , and none is observed .
the richest clusters show a slightly higher cumulative @xmath86 at @xmath103 ( figure [ graph : graph300 ] ) .
this is caused by a somewhat higher @xmath3 at the very center of the clusters , which may reflect a higher central mass concentration relative to luminosity in the innermost regions of the richest systems ( e.g. , more massive or extended bcgs than our central mass exclusion and/or loss of luminosity due to merging ) .
the @xmath101 profile ( figure [ graph : graph400 ] ) is similar for all richness groups and on all scales starting from deep inside clusters ( @xmath104 ) ; the profile decreases slowly with radius , by a factor of @xmath105 , from small scales up to several @xmath29 and remains flat thereafter to nearly @xmath106 .
this trend is consistent with the stellar population mix , which we quantify below .
we use the mean observed relative stellar @xmath107 ratio of early - type versus spiral galaxies of @xmath108 @xcite .
this is consistent with @xcite single stellar population synthesis models with ages 10 gyr versus 4 gyr , the approximate mean ages of stellar populations in e / s0 and spiral galaxies ( albeit with much scatter ; see @xcite ) .
we combine this ratio with the mean fraction of early and late type galaxies using the density - morphology relation discussed above . here
we use a typical e / s0 fraction of 40% on large - scales ( @xmath109 ) , increasing to 90% in the central high - density regions of groups and clusters ( @xmath110 ) .
the resulting mass - to - light ratio then decreases by a factor of @xmath111 from the centers of clusters to the lower densities on large scales .
this is presented by the dashed lines in figures [ graph : graph300 ] and [ graph : graph400 ] ( for the cumulative and differential functions , respectively ) .
this expected trend agrees well with the observed @xmath74 profile for all systems , suggesting that the small decline in @xmath74 as a function of scale can be accounted for by the different stellar population .
clusters have a strong spatial auto - correlation function and are located in high - density regions with a higher density of older galaxies that extends to large scales ; this is likely the reason that the high @xmath101 ratio persist to several @xmath29 before dropping to a lower more constant field value on large scales .
the typical cluster correlation scale is @xmath112 @xcite for both poor and rich clusters ; this is shown by the vertical band in figures [ graph : graph300 ] and [ graph : graph400 ] . while @xmath3 reaches the cosmic value on these large scales , we emphasize that when the stellar population age is accounted for , all systems exhibit the same mean mass - to - light ratio on all scales .
light , or more precisely stellar mass , thus traces the total mass remarkably well on nearly all scales .
this is discussed further in [ d4 ] .
, as a function of the mean cluster luminosity within @xmath29 , @xmath113 ( a proxy for cluster richness or mass ) .
the mass - to - light ratio is shown within two radii , @xmath114 mpc and @xmath115 mpc .
the dashed line represents the expected trend of the varying stellar age ( for @xmath116 mpc ; see text ) . the bcg mass and light are excluded . ] , including the bcg mass and light , as a function of richness at large scales ( @xmath117mpc and @xmath118mpc ) .
the horizontal axis is @xmath119 , the cumulative light within @xmath29 , a proxy for the cluster richness or mass .
@xmath3 is constant on these scales , independent of the central environment .
the gray band shows the mean @xmath120 .
this corresponds to a cosmic mass - density of @xmath121 ( see [ d5 ] ) . ]
an alternate way to view the behavior of @xmath86 for different environments and scales is presented in figure [ graph : graph500 ] . here
we show @xmath86 as a function of mean cluster luminosity within @xmath29 ( a proxy for cluster richness or mass ) , for several radii ; this is an inverse of the previous plots of @xmath86 presented for a few richness groups .
figure [ graph : graph500 ] shows the direct dependence of @xmath86 on the richness ( mass ) of the central environment .
the cumulative @xmath80 is presented within two radii , @xmath114 mpc and @xmath122 mpc .
the mass and light of the central bcg have been excluded in figure [ graph : graph500 ] as they were in figures [ graph : graph200 ] and [ graph : graph300 ] , although the effect of the bcg is negligible on the larger scale ( @xmath123 mpc ) .
each point in the cluster luminosity @xmath124 is an average of two of the twelve original richness bins .
the stellar population effect as a function of @xmath124 for the 3.1 mpc radius is shown by the dotted line ; it agrees well with the observed @xmath3 function , except for the richest clusters , which as described above contain an increased amount of mass compared to light at their very centers . here
, the slow increase of @xmath3 with richness due to the stellar population trend is caused by the fact that the same physical radius corresponds to a different galaxy overdensity in environments of different richness .
figure [ graph : graph600 ] presents the integrated @xmath86 versus luminosity ( richness ) for all systems within the large scales of 9 and 18.4 @xmath96 mpc .
the @xmath3 contains all the mass and light including that of the bcg .
the integrated @xmath86 ratio on these scales is constant , independent of the central environment ; the mass - to - light ratio reaches a mean cosmic value of @xmath125 . as seen in the previous figures ,
the same constant value is reached even within the inner parts of clusters when the stellar population mix is accounted for .
all the measurements discussed above are centered on the bcg , which may be slightly offset from the center of the dark matter halo of the cluster .
our results therefore reflect the distribution of @xmath74 around bcgs out to large cosmic scales and should be so interpreted .
furthermore , for scales outside the innermost regions of clusters , the possible mis - centering has negligible impact on @xmath3 .
the poorest groups have the highest uncertainty in terms of cluster centering , luminosity uncertainties , and mass uncertainties because of the weaker lensing signal .
therefore , the poorest groups ( @xmath126 ) are prone to larger uncertainties .
previous results found that @xmath127 increases with cluster richness ( mass ) , approximately as @xmath128 ( e.g. , @xcite ) .
here we show that this is mostly the effect of the bcg luminosity : the @xmath29 scale of the poorest groups is nearly the same size as the bcg galaxy ( a few hundred kpc ) , therefore the group @xmath127 is dominated by the lower @xmath86 of the bcg , while the bcg luminosity of rich clusters is negligible at the cluster scale of @xmath129 mpc . using the observed relations ( @xmath130 and @xmath131 ; [ results ] ; @xcite ) we find @xmath132 , as is indeed observed . comparing the @xmath74 profile outside the innermost region ( at @xmath133 ) , we show that @xmath134 is nearly independent of richness ( figures [ graph : graph300 ] and [ graph : graph400 ] ) .
the nearly flat distribution of @xmath14 persists on all scales , from @xmath104 ( @xmath135 kpc in small groups and @xmath8 kpc in large clusters ) out to nearly @xmath136 mpc .
the small decrease in the mass - to - light ratio as a function of @xmath100 is consistent with a decreasing stellar population age as a function of local galaxy density ( [ results ] ) .
this effect , combined with the fact that the physical radius @xmath44 corresponds to a different galaxy density in clusters of different richnesses , also accounts for the observed trend in @xmath80 as a function of cluster richness at fixed @xmath44 ( figure [ graph : graph500 ] ) .
this agreement implies that the underlying mass and light distribution , when accounting for the stellar population age , follow each other remarkably well on all scales .
it further suggests that the stellar mass fraction , @xmath137 , is nearly constant on these scales ( see [ d4 ] ) .
the luminosities discussed above do not represent the total luminosities since they include only galaxies above the threshold of @xmath138 ( [ data ] ) .
the luminosity contributed by fainter galaxies is not included , and neither is the diffuse intracluster light ( icl ) integrated light from individual stars in the cluster potential @xcite .
while this does not affect the self - consistent results discussed above , the total luminosity should be accounted for when comparing our results with other measurements .
we estimate these additional luminosity contributions below .
galaxies below the threshold of @xmath139 are estimated to contribute an additional @xmath140 to the current luminosity , using the observed schechter luminosity function of sdss galaxies which has a faint - end slope of @xmath141 in the @xmath23-band (; consistent with @xcite and @xcite ) .
the radial distribution of faint galaxies is assumed to be similar to that of the other galaxies as suggested by the sdss observations of @xcite showing that the galaxy luminosity function in groups and clusters is independent of radial scale .
the diffuse intracluster light ( icl ) from individual stars , thought to be stripped from galaxies as a result of gravitational interactions within groups and clusters , has been measured in detail for sdss clusters at @xmath142 .
@xcite used stacked images of 683 clusters to measure the icl in the @xmath23-band out to 700 kpc . they find that the icl contributes @xmath143 of the luminosity in groups and clusters , nearly independent of cluster richness and bcg luminosity .
the icl is more concentrated than the galaxy distribution .
@xcite measure the icl around 42,000 stacked sdss luminous red galaxies ( lrgs ) that are located in the centers of groups and clusters .
they find an icl contribution of @xmath144 .
we adopt an icl correction of @xmath145 additional luminosity ( i.e. , @xmath146 of the total cluster luminosity ) .
we add the above luminosity of faint galaxies ( @xmath140 ) and the intracluster light ( @xmath145 ) ; they decrease the @xmath3 ratios by these factors ( 1.36 and 1.15 ) on the group and cluster scales ( i.e. , @xmath147 ) , and by @xmath140 ( a factor of 1.36 ) everywhere else .
this corrected total @xmath148 profile is presented in figure [ graph : graph800 ] .
it enables us to compare the results with other measurements and with the contribution from individual galaxies , as discussed in [ d3 ] .
it is generally believed that groups and clusters have significantly more dark matter ( relative to light ) than individual @xmath149 galaxies ( e.g. , @xcite ) . @xcite and @xcite suggested , however , that clusters and groups do not contain significantly more dark matter per unit light than do galaxies , but instead most of the dark matter resides in the large halos of individual galaxies ; these make up the dark matter observed in groups , clusters , and large - scale structure .
we investigate this further below : how much of the observed mass , and thus @xmath3 , is contributed by individual galaxies ( plus gas ) .
we use the mean observed mass - to - light ratio of typical isolated @xmath149 elliptical and spiral galaxies , combined with the density - morphology relation ( [ results ] ) to estimate the amount of mass and @xmath74 contributed by individual galaxies to the observed @xmath74 profile on all scales and in all environments . as discussed in [ results ] , the @xmath94 function flattens to a constant value at approximately @xmath150 kpc .
this scale is comparable to the virial radius of bright @xmath9 galaxies ( @xmath151 kpc at the virial overdensity of @xmath152 relevant for @xmath16cdm ) .
if galaxy halos extend to these scales , as suggested by observations ( below ) , could the dark matter of individual galaxies ( including these large halos ) , plus the known gas component , account for all or most of the mass observed in the @xmath153 function , from small scales inside clusters to the large cosmic scales of nearly @xmath136 mpc ?
to test this we use , for simplicity , a @xmath150 kpc halo radius around @xmath9 galaxies , with @xmath154 values that are consistent with observations : @xmath155 for spirals , and , with a factor of two lower @xmath156 for older galaxies ( [ results ] ) , an @xmath157 for early - type @xmath9 galaxies ( e / s0 ) .
these values are motivated by observations of individual @xmath9 galaxies .
for example , the milky way ( mw ) and m31 , both nearly @xmath9 spiral galaxies , have been measured in detail to dermine their extended masses ( see below ) .
their luminosities are @xmath158 and @xmath159 ( @xcite observe @xmath160 ; @xcite infer @xmath161 and @xmath162 ; @xcite report @xmath163 in the range @xmath164 ; @xcite find a total m31 luminosity of @xmath165 ) .
this luminosity is consistent with the sdss @xmath166 used above ( @xmath167 when converted to @xmath20 ) .
the extended mass of the milky way and m31 have been measured recently using proper motions observed with the hubble space telescope ( hst ) combined with the timing argument ( ta ) ; @xcite determine the sum of the two virial masses to be @xmath168 , consistent with the @xcite revised ta method that gives @xmath169 . @xcite
combined the leo - i proper motion with the ta method and other measurements to obtain @xmath170 , and @xcite obtain an estimated virial mass of @xmath171 .
more recently , @xcite applied the least action method @xcite to the local group galaxies to derive masses that are consistent with the high end of the above values : they find @xmath172 and @xmath173 ; their best value for the mw increases to @xmath174 when data of four external groups is included . considering the above masses and luminosities , while uncertain
, we use as approximate values @xmath175 for @xmath176 spiral galaxies ( within radius @xmath177 kpc ) .
early - type galaxies , with their fainter @xmath156 luminosities ( by a factor of about 2 ) , thus imply @xmath178 kpc ) @xmath179 for @xmath180 e / s0 galaxies .
this is consistent with recent observations of isolated early - type galaxies using weak and strong gravitational lensing .
@xcite use hst data to study strong and weak lensing by isolated elliptical galaxies and find a mean @xmath181 kpc ) = @xmath182 , consistent with the value above .
@xcite conduct a detailed weak - lensing analysis of galaxies from the cfht legacy survey , finding a mean @xmath183 for red @xmath9 galaxies in low - density regions ( using a truncated isothermal sphere model with an observed best - fit truncation radius of @xmath184 kpc ) .
they also find a mean @xmath185 for all the red and blue galaxies combined ( with a best - fit truncation radius of @xmath186 kpc ) .
we adopt the above @xmath187 and @xmath188 within @xmath189 kpc of individual @xmath180 galaxies as approximate values to illustrate the contribution of galaxies to the dark matter and @xmath94 function on all scales . to determine how much galaxies with these @xmath154 ratio contribute we combine these galaxy @xmath154 values with the density - morphology relation that describes the mean observed fraction of spiral and e / s0 galaxies as a function of density ( [ results ] ) .
the e / s0 fraction decreases from nearly @xmath190 at the high - density regions of clusters to @xmath191 in the low - density field on large scales .
this yields an estimate of the mean @xmath154 contributed by the galaxies .
( we note that while the large dark matter halos of individual galaxies may be stripped off in the dense regions of clusters , and halos may overlap , the total mass remains within the cluster potential . )
the contribution from galaxies therefore ranges from @xmath192 in the dense regions of clusters , decreasing to @xmath193 on large scale ( where @xmath191 of galaxies are e / s0s ) .
we add the additional gas component that exists on all scales ; for this , we use a gas component that is @xmath194 of the total mass , since the cosmic baryon fraction is @xmath195 @xcite , and the stellar fraction is @xmath196 ( [ d4 ] ) .
this is consistent with the extended gas distribution observed in groups and clusters within their virial radii @xcite . using a lower gas fraction , as observed in the central parts of small groups @xmath197 , or a slightly varying gas fraction , only affects the results on the smallest scales ( @xmath198 ) of small groups by @xmath199 , well consistent with the observed @xmath14 function . on larger scales , the intergalactic medium ( igm )
is expected to contain a similar gas fraction ( @xcite ; see also @xcite for the extended gas component observed in the circumgalactic medium around galaxies ) .
this sum of galaxy and gas mass represents the mass contributed by individual galaxies ( including their dark matter halos ) plus the intracluster / intergalactic / circumgalactic gas .
this contribution ranges from @xmath200 inside rich clusters , decreasing to @xmath201 on large scales .
this @xmath154 of individual galaxies plus gas is plotted as the declining blue band in figure [ graph : graph800 ] .
it agrees remarkably well with the entire observed @xmath202 profile from the small scales inside groups and clusters to the largest cosmic scales of nearly @xmath203 mpc .
the comparison in figure [ graph : graph800 ] shows that individual galaxies and their large dark matter halos , plus the intracluster / intergalactic gas , can account for all or most of the mass in groups , clusters , and on large scales , reproducing both the total amount of dark matter and the overall distribution of mass and light on all scales .
this suggests that most of the dark matter in the universe may be located in the large halos of individual galaxies ; groups , clusters , and large - scale structure do not appear to contain significantly more dark matter ( relative to light ) than do their @xmath9 galaxy members .
this is consistent with the earlier results of @xcite .
the values used above for individual galaxies are approximate , with large uncertainties .
using lower @xmath154 values for galaxies will reduce their contribution , and vice versa .
faint galaxies , which are known to have higher @xmath3 ratios , will further increase the dark matter contribution from individual galaxies .
is the total luminosity ( including light from faint galaxies below our luminosity threshold plus the diffuse icl ; see [ d2 ] ) .
the declining blue band represents the @xmath3 of individual galaxies plus gas ( [ d3 ] ) .
the galaxies plus gas appear to contribute all or most of the total observed @xmath14 function on all scales and environments . ] as function of @xmath100 , for the three richness bins .
the horizontal band shows the cosmic stellar mass fraction ( [ d4 ] ) .
_ b _ ) as in panel _ a _ , but for the cumulative stellar mass fraction @xmath204 .
_ c _ ) the stellar fraction @xmath205 within @xmath206 ( the radius within which the mean density is 200 the matter density ) plotted as a function of @xmath207 ( the mass within @xmath206 ) . ]
the nearly flat distribution of @xmath154 with radius , especially when accounting for the age of the stellar population , suggests that the stellar mass fraction may be similar in all environments , thus causing light to trace mass .
we estimate below the implied stellar mass fraction as a function of scale and environment .
we divide the mean observed stellar mass - to - light ratio @xmath107 of early and late - type galaxies , coupled with their relevant fraction as given by the density - morphology relation ( [ results ] ) , by our @xmath154 , including the bcg .
this yields the approximate stellar mass fraction @xmath208 : [ stellarfraction ] f _ * ( r / r_200 ) = & m_*/m ( r / r_200 ) + & ( r / r_200 ) estimates of the @xmath107 ratios of e / s0s and spirals vary significantly throughout the literature , although most stellar and dynamical mass - to - light estimates yield @xmath209 @xcite .
stellar mass - to - light ratio determinations can differ as a result of the precise stellar population synthesis model and data used to obtain them .
@xcite compare several @xmath210 estimates for early - type galaxies from different sources , all corrected to @xmath211 and a chabrier initial mass function ( imf ) for consistency .
a similar comparison for @xmath212 ( g. graves , private communication ) reveals that models based on fits to observational spectra ( @xcite , using sdss data ; @xcite ) give @xmath213(e / s0)@xmath214 ( and @xmath213(sp)@xmath215 ) , as does a single burst star formation model .
models based on fits to photometric seds @xcite find lower values of @xmath213(e / s0)@xmath216 ( @xmath213(sp)@xmath217 ) .
similarly , @xcite , using the cosmos data , find @xmath107(e / s0)@xmath218 and @xmath107(sp)@xmath219 .
all the above use the chabrier imf ( or are scaled to it ) . adopting the salpeter
imf increases these mass - to - light ratios by a factor of @xmath99 .
measurements of @xmath107 for the milky way and m31 , `` typical '' @xmath9 spiral galaxies , find @xmath220 @xcite or @xmath221 @xcite , and @xcite find @xmath222 for all components of the galaxy combined .
the value for m31 is significantly higher than that found for the milky way and other spirals above ; this may be caused by m31 s relatively red color for a spiral @xcite .
the @xmath107 values for the milky way and m31 seem to indicate a higher @xmath107 for spirals than found by some of the models above ; as a result , we adopt the higher range of values from the sources described , using @xmath107(e / s0)@xmath223 and @xmath107(sp)@xmath224 .
the results for the stellar fraction discussed below are proportional to the assumed value of @xmath107 ; using different values , the resulting stellar mass density and stellar mass fraction will scale proportionally . since @xmath107 is independent of the hubble parameter @xmath225
, we convert the @xmath154 ratios presented in the previous section ( which were presented in units of @xmath225 ) to the @xmath16cdm @xmath20 ( thus reducing all @xmath154 by 0.7 ) .
the total observed mass - to - light ratio including the bcg and accounting for the total luminosity of the system ( [ d2 ] ) is then used to determine @xmath205 as described above . in the top panel of figure
[ graph : graph900 ] we present the local stellar fraction @xmath205 as a function of radius in units of @xmath100 .
the stellar mass fraction is essentially constant with radius from 0.2 to @xmath106 , with a value @xmath226 for all richnesses .
also shown is a horizontal band representing the average stellar mass fraction of the universe .
for this , we use the measured luminosity density of @xmath227 ( [ d5 ] ) , increased by a factor of 1.36 to account for the light from galaxies below our luminosity threshold ( [ d2 ] ) .
we combine this with the @xmath107 for spirals and e / s0s as discussed above to obtain a cosmic mean stellar mass density of @xmath228 ( for @xmath229 ) , implying @xmath230 .
this is consistent with the mean stellar mass density of @xmath231 ( @xmath232 for @xmath233 ; [ d5 ] ) found by @xcite using the data of @xcite .
the middle panel of figure [ graph : graph900 ] shows the cumulative stellar fraction @xmath205 within @xmath100 .
the strong influence of the central bcg is clearly seen in the poor groups , as discussed in [ results ] , causing the increase in stellar fraction near the cluster center . at small @xmath100 , especially in poor groups ,
the stellar fraction increases as the bcg becomes dominant . on large scales , systems of all richnesses
reach the same cumulative stellar fraction of @xmath234 . in the lower panel of figure
[ graph : graph900 ] , we present the stellar fraction within @xmath206 , the radius at which the interior mean density is 200 times the matter density of the universe , versus @xmath207 , the mass within @xmath206 .
the increasing stellar fraction at lower mass ( smaller @xmath206 ) reflects the dominance of the central bcg on this scale ( [ results ] ) .
the observed mass - to - light ratio on large scales can be used to determine the mass density of the universe , @xmath235 ( e.g. , @xcite , and references therein ) . figure [ graph : graph600 ] presents the observed @xmath154 within two large scales ( 9 and 18.4 @xmath236 mpc ) around all systems , from the poorest groups to the richest clusters ( [ results ] ) .
the observed @xmath154 within these large scales is independent of the central environment , reaching a representative cosmic value ( see also ) .
this mean value , shown by the horizontal band in figure [ graph : graph600 ] , is @xmath237 .
combined with the luminosity density of the universe in the @xmath238 band for galaxies above @xmath239 , @xmath240mpc@xmath241 ( comoving ) , we find @xmath242 , independent of @xmath225 . the small bias factor , @xmath243 , depends primarily on the bias of the galaxy tracers relative to the underlying mass distribution ( see [ data ] ) .
the galaxies used have a low threshold luminosity of @xmath244 .
the bias of such sub-@xmath9 threshold galaxies is @xmath245 ( for @xmath68 ; increasing to 1.09 for @xmath246 ; @xcite ) .
we thus find @xmath247
we compare the distribution of mass and light around 132,473 bcg - centered sdss groups and clusters as a function of scale , from small scales inside clusters to large cosmic scales of nearly @xmath203 mpc , and for different richness environments .
the masses are determined from stacked weak gravitational lensing observations and are used to derive the mass - to - light profile , @xmath248 ; this profile indicates how light traces mass on all scales and in all environments .
we summarize our main conclusions below .
the @xmath249 profile rises with radius on small scales inside groups and clusters , reflecting the increasing mass - to - light ratio of the central bright bcg galaxy .
@xmath249 then flattens to a nearly constant value , showing that light follows mass on large scales .
this flattening to a nearly constant @xmath154 ratio begins at relatively small scales of only a few hundred kpc inside clusters and remains nearly constant to the largest cosmic scales of @xmath250 mpc ; this is especially so when accounting for the varying stellar population age as a function of local density ( see @xmath251 below ) .
this indicates that light follows mass on all scales ( above @xmath189 kpc ) and in all environments , even inside clusters .
the rise to the flat constant value in the cumulative @xmath94 function is considerably slower in poor groups because they are dominated by light from the central bcg galaxy ( see @xmath252 below ) .
the luminosity of the bcg galaxy increases with cluster richness , but its dominance relative to the total cluster luminosity decreases with richness .
this decreasing trend of @xmath253 with cluster richness affects the cumulative @xmath154 profile in the central regions of groups and clusters , causing the slower increase of @xmath249 with radius for poor groups , but reaching a constant @xmath154 ratio on very small scales of @xmath189 kpc in rich clusters .
the decreasing dominance of the bcg with richness suggests that the growth of the central bcg is less efficient than the growth of the cluster as a whole .
clusters , which evolve by merging and accretion of poorer systems , grow faster than do their central bcg galaxies ( also by merging and accretion ) .
the small trends observed in the @xmath249 profile outside the central bcg regions a slowly decreasing @xmath154 with radius and richness are shown to be consistent with the varying stellar population as a function of density , following the density - morphology relation .
this stellar population trend results in a slowly decreasing @xmath154 with radius due to the increasing population of young spiral galaxies , which have lower @xmath154 than ellipticals , before flattening to a constant cosmic value on larger scales .
this indicates that stars , which account for only a few percent of the total mass , trace the total mass remarkably well .
the dark matter in the universe thus follows light , and especially stellar mass , on all scales above a few hundred kpc .
we determine the stellar mass fraction as a function of environment and scale , @xmath254 .
we find that the stellar mass fraction is nearly constant on all scales and all environments above few hundred kpc , with @xmath226 .
this fraction is consistent with the cosmic stellar mass fraction .
the fact that stars follow mass so well is the main reason why light traces mass on all these scales .
we show that most of the dark matter in the universe may be located in large dark matter halos around individual galaxies ( @xmath189 kpc for @xmath9 galaxies )
. the mass and mass - to - light ratio of groups , clusters , and large scale structure is consistent with being contributed by the mass of individual galaxies , including their large dark matter halos ( which may be stripped off inside the dense regions of clusters ) , plus the additional intergalactic / intracluster gas .
the mass from individual galaxies , plus gas , appear to be consistent with the entire observed @xmath3 profile on all scales and in all environments ( fig .
[ graph : graph800 ] ) .
this suggests that most of the dark matter in the universe may be located in the large halos of individual galaxies ; groups , clusters , and large scale structure are simply made - up by this dark matter ; they do not contain significantly more dark matter ( relative to light ) than do the individual galaxies .
the constant @xmath154 ratio on large scales represents the universal mass - to - light ratio .
this corresponds to a mass - density parameter of @xmath255 ( where the small galaxy bias factor is @xmath256 ) .
we thank the referee for the helpful comments on the paper . a.
k. acknowledges support from an nsf graduate research fellowship . | how is mass distributed in the universe ?
how does it compare with the distribution of light and stars ?
we address these questions by examining the distribution of mass , determined from weak lensing observations , and starlight , around @xmath0 sdss maxbcg groups and clusters as a function of environment and scale , from deep inside clusters to large cosmic scales of @xmath1 mpc .
the observed cumulative mass - to - light profile , @xmath2 , rises on small scales , reflecting the increasing @xmath3 of the central bright galaxy of the cluster , then flattens to a nearly constant ratio on scales above @xmath4 kpc , where light follows mass on all scales and in all environments .
a trend of slightly decreasing @xmath5 with scale is shown to be consistent with the varying stellar population following the morphology - density relation .
this suggests that stars trace mass remarkably well even though they represent only a few percent of the total mass .
we determine the stellar mass fraction and find it to be nearly constant on all scales above @xmath6 kpc , with @xmath7 .
we further suggest that most of the dark matter in the universe is located in the large halos of individual galaxies ( @xmath8 kpc for @xmath9 galaxies ) ; we show that the entire @xmath5 profile from groups and clusters to large - scale structure can be accounted for by the aggregate masses of the individual galaxies ( whose halos may be stripped off but still remain in the clusters ) , plus gas .
we use the observed mass - to - light ratio on large scales to determine the mass density of the universe : @xmath10 [ firstpage ] galaxies : clusters : general galaxies : groups : general cosmology : observations cosmological parameters dark matter large - scale structure of universe |
the concept of entanglement and non - locality @xcite , one of the most striking features of quantum mechanics , has been heavily debated since the early days of the field @xcite . by now , there is abundant experimental evidence @xcite that nature indeed does possess nonlocal aspects , in stark contrast to our everyday - life experience .
but it was only relatively recently , when richard feynman @xcite , david deutsch @xcite , and other researchers in the 1980s , envisioned the idea of exploiting the quantum degrees of freedom for a novel way of information processing .
the central question at the time was whether and how it is possible to efficiently simulate any finite physical system with a man - made machine .
deutsch argued that such a simulation is not possible perfectly within the classical computational framework that had been developed for decades @xcite .
instead , the universal computing machine should be of a quantum nature , i.e. , a `` quantum computer '' . since then , progress in different areas of research and industry tremendously influenced the advent of quantum computing .
first , the booming computer industry led to major progress in semiconductor- , nano- , and laser - technology , a prerequisite for the fabrication , addressing , and manipulation of single quantum systems needed in an experimental realization .
second , several algorithms have been developed by now , such as those by deutsch @xcite , grover @xcite , and shor @xcite , which clearly illustrate that quantum computers , exploiting the existence of entanglement , can solve problems much faster than classical computers .
a recent review on using quantum computers for quantum simulations can be found in ref .
in addition , the theories of quantum complexity and entanglement are currently being established , a process which is still far from being complete .
the emerging fields of nano- and quantum information science have inspired and motivated each other in various ways , today more than ever . shortly after the first quantum algorithms had been developed , setups
have been suggested to turn quantum computing into reality .
these ideas , among others , are based on quantum dots @xcite , cold trapped ions @xcite , cavity quantum electrodynamics @xcite , bulk nuclear magnetic resonance @xcite , low - capacitance josephson junctions @xcite , donor atoms @xcite , linear optics @xcite , molecular magnets @xcite , spin clusters @xcite , or color centers in diamond @xcite .
a long list of interesting results has followed , some of which will be reviewed in this article . in 1997 , it was proposed to encode the quantum information in the spin states of single - electron quantum dots @xcite .
the tunnel barrier between neighboring dots , which can be varied via gates , fig .
[ schemeldivqc ] , induces time - dependent electron - electron interactions and affects the spin states via the heisenberg exchange coupling .
the proposal demonstrates theoretically that such a setup allows for universal and scalable quantum computing , controllable by purely electrical means . here
, we will particularly focus on the experimental and theoretical achievements following ref .
@xcite , since substantial progress in this field has been made within the past few years .
an overview on recent results in other setups can , e.g. , be found in ref .
@xcite . , where @xmath0 is a function of gate voltage and corresponds to the energy splitting of the spin singlet and triplet states .
this allows for electrically controlled two - qubit gates with fast operation times @xcite .
( b ) a qd array based on ( a ) , where the qubits in the two right dots are coupled while the others are decoupled .
rotations of individual spins may be achieved by pulling the desired electron down into a region of high magnetization or high @xmath1 factor via back gates , such that the zeeman splitting and hence the resonance condition changes for this electron in the presence of a static magnetic field @xmath2 .
a resonantly applied oscillating magnetic field pulse @xmath3 hence rotates the addressed qubit ( electron spin resonance , esr ) , while all others remain unaffected due to off - resonance .
exploiting spin - orbit interaction , the rotations may also be driven fully electrically via electric - dipole - induced spin resonance ( edsr ) . for details on esr and edsr , see sec .
[ sec : soiandnuclearspins ] .
alternatively , fast single - qubit gates may be implemented via exchange - controlled spin rotations @xcite .
combination of single- and two - qubit gates results in a universal set of quantum gates , so that the proposed schemes allow for fast and purely electrically controlled quantum computation with electron spins in qds @xcite . ]
the review is organized as follows . in sec .
[ sec : systemsanddefinitions ] we introduce the three quantum dot ( qd ) systems discussed in this report : self - assembled qds , lateral qds , and nanowire - based qds
. we also comment on basic requirements for quantum computation and define the spin lifetimes @xmath4 , @xmath5 , and @xmath6 .
spin - orbit interaction and nuclear spins are covered in sec .
[ sec : soiandnuclearspins ] .
these present an undesired ( noise ) source of relaxation and decoherence for the spin qubits on the one hand , but on the other hand allow for all - electrical spin manipulation via electric - dipole - induced spin resonance , or for strong ( effective ) magnetic field gradients . in the main part , sec .
[ sec : recentprogress ] , recent progress in these qd systems is summarized , compared , and discussed .
newly proposed architectures for long - distance qubit - qubit coupling are reviewed in sec .
[ sec : newproposals ] , followed by our summary and final remarks , sec .
[ sec : outlook ] .
quantum dots ( qds ) confine electrons or holes ( missing valence band electrons ) in all three dimensions , on length scales which are comparable to the wavelengths of the particles , i.e. , typically @xmath7 in each spatial direction .
there are many possibilities to realize such confinement , which is evident from the variety of systems under study . in this report , we will mainly focus on three of them .
the first category is self - assembled qds .
these form naturally during growth , where ingaas dots within a gaas matrix are commonly used examples .
when ingaas is grown on gaas ( stranski - krastanov mode ) , islands form spontaneously after a small critical thickness of only a few monolayers due to the mismatch in the lattice constants .
these may then be covered with further layers of gaas .
such qds are typically lens - shaped , with heights of @xmath8 ( growth direction ) and diameters @xmath9 , and confinement results from the difference in the conduction and valence band edges of the involved materials .
alternatively , interface fluctuation qds arise from monolayer fluctuations in thin quantum wells , typically resulting in gaas dots within algaas @xcite .
the second category , lateral qds , is based on two - dimensional electron and hole gases ( 2degs , 2dhgs ) , which exist in heterostructures from materials with suitable band properties and additional dopants .
for instance , algaas / gaas heterostructures are routinely used to form 2degs and 2dhgs within gaas , strongly confined along the growth direction .
lithographically defined gate electrodes on the sample surface allow to create confinement also in the transverse directions , leading to quasi-2d qds of @xmath10 in diameter . finally , semiconductor nanowires naturally provide confinement in two dimensions , due to their small diameters of @xmath11 , and repulsive forces along the wire may again be added via nearby gates or via additional layers of barrier material .
we note that several other qd implementations exist , which , however , for space reasons will not be discussed here .
prominent examples are qds in carbon - based systems , like graphene @xcite or carbon nanotubes @xcite , which are highly attractive for implementing spin qubits .
any setup considered for quantum computation should fulfill a list of essential criteria , such as scalability and the ability to initialize the system in a fiducial state @xcite . for quantum error correction schemes to be applicable , it is important that the lifetimes are much longer than the gate operation times .
a decade ago , this had been considered a very serious challenge , by now this problem seems to have been overcome in qds as shown in sec .
[ sec : recentprogress ] .
three time scales are of interest in this context , which we illustrate in terms of the electron spin qubit states @xmath12 and @xmath13 , assuming that these are eigenstates of the pauli operator @xmath14 with energy difference @xmath15 .
first , the relaxation time @xmath4 describes transitions @xmath16 due to interactions with the environment , such as the lattice , which leads to relaxation from the excited @xmath12 to the ground state @xmath13 .
a typical measure for @xmath4 is @xmath17 with initial state @xmath12 .
second , the decoherence time @xmath5 quantifies the decay of quantum mechanical superpositions , and accounts for transitions of type @xmath18 induced by the environment ( see inset to fig .
[ figuredecoherencescience ] ) . when the state is initially @xmath19 , an eigenstate of @xmath20 , a typical measure for @xmath5 is the envelope function of @xmath21 .
we note that @xmath21 oscillates between @xmath22 at angular frequency @xmath23 for a perfectly isolated system , but decays to 0 as the state turns into either @xmath12 or @xmath13 .
the envelope function itself may be referred to as @xmath24 .
finally , in practice it is generally required to average over an ensemble , rather than to measure a single system only .
the averaged @xmath25 often decays faster than in each individual case , because the oscillation frequencies may be slightly different from system to system ( e.g. , small deviations in @xmath15 ) , which leads to destructive interference and additional damping . the so - called dephasing time obtained from an ensemble measurement is therefore labeled @xmath26 .
the three time scales @xmath4 , @xmath5 , and @xmath26 are not completely unrelated .
one finds that @xmath27 , and usually @xmath28 and @xmath29 in practice . in qds
, decoherence typically results from the nuclear spins and also the spin - orbit interaction . commenting on the terminology ,
the relaxation ( @xmath4 ) , decoherence ( @xmath5 ) , and dephasing ( @xmath26 ) times are only well defined when @xmath30 or @xmath31 , respectively , decay exponentially , which is the assumed behavior in most quantum error correction schemes .
strictly speaking , one should therefore avoid these terms when the longitudinal or transverse decay is of a non - exponential form .
we note , however , that the introduced nomenclature is often being used to characterize any decaying behavior for convenience . a key criterion for building quantum computers , the one which actually justifies their name , is the presence of a universal set of quantum gates .
this may fortunately be realized with one- and two - body interactions only , since any operation can be carried out as a sequence of one- and two - qubit gates .
in fact , the implementation of single - qubit rotations for each element , along with only one type of entangling two - qubit gates , e.g. , @xmath32 or cnot , between neighboring qubits would be sufficient for universal quantum computation @xcite .
we note in passing that entanglement of spin qubits can be created in many different ways and over long distances , for instance , by extracting and separating cooper pairs from an s - wave superconductor as proposed in ref .
@xcite and experimentally investigated recently in refs .
@xcite . nuclear spins and spin - orbit interaction , which present an undesired source of decoherence on the one hand ,
may prove useful for implementing qubit gates on the other hand , and both mechanisms will therefore be analyzed in more detail in the next section .
spin - orbit interaction ( soi ) couples the orbital motion of a charge to its spin and arises in the presence of inversion asymmetry , where two types can be distinguished .
rashba soi results from structural inversion asymmetry and , for electrons , is typically of the form @xmath33 , where the components @xmath34 are the standard pauli matrices for spin 1/2 , @xmath35 is the momentum operator , and @xmath36 is an effective electric field determined by the system structure @xcite .
dresselhaus soi is present in materials that lack bulk inversion symmetry , such as inas or gaas , and is of the form @xmath37 , where @xmath38 , @xmath39 , and @xmath40 correspond to the crystallographic directions @xmath41 $ ] , @xmath42 $ ] , and @xmath43 $ ] , respectively @xcite . for quasi-2d systems
, these hamiltonians can be reduced further . for strong confinement along the @xmath44 direction
, the rashba term simplifies to @xmath45 with rashba parameter @xmath46 .
the resulting form of the dresselhaus term strongly depends on the growth direction . for the @xmath44 axis chosen along the confinement direction
, one can substitute @xmath47 , @xmath48 , and all other terms can be neglected because of their smallness compared to terms @xmath49 @xcite .
for example , for @xmath50 $ ] one obtains @xmath51 , while for @xmath52 $ ] the spin projection along the confinement direction is conserved , @xmath53 .
both these hamiltonians vary under rotations of the coordinate system about the @xmath44 axis , so that their exact form is determined by the relative orientation of coordinate and crystal axes .
this is different for @xmath54 $ ] , where the effective dresselhaus term is @xmath55 , which moreover corresponds exactly to the form of the rashba term .
therefore , rashba and dresselhaus soi can cancel in lowest order for growth along the @xmath56 $ ] direction @xcite .
we note that the presence of soi results in small , but finite , anisotropic corrections to the heisenberg exchange interaction of electron spins , thus affecting the fidelity of quantum gates based on isotropic exchange .
fortunately , strategies have been developed with which the soi - induced gate errors can be strongly suppressed @xcite . in general , gate errors can be reduced from first to second ( or higher ) order in soi when the coupling strength @xmath0 is varied symmetrically in time , followed by additional qubit rotations @xcite . in particular , the anisotropic corrections can be cancelled completely in the cnot gate construction of ref .
@xcite when the system is pulsed such that the anisotropic terms are linear in @xmath0 @xcite .
additional errors caused by dipole - dipole interactions were found to be negligible for cnot ( in typical situations ) @xcite .
a rather useful technique for electrically controlled qubit rotations is the electric - dipole - induced spin resonance ( edsr ) .
it is closely related to the well - known electron spin resonance ( esr ) , which we therefore review first . for this , let us consider an electron in a qd in the presence of magnetic fields .
the hamiltonian @xmath57 consists of a spin - independent part @xmath58 , where the first ( second ) term corresponds to the kinetic ( potential ) energy , and the zeeman part @xmath59 which couples the magnetic field @xmath60 to the spin . in the following ,
we assume that a constant magnetic field @xmath61 is applied along the @xmath44 axis , while a small oscillating field @xmath62 , @xmath63 , is applied along the @xmath64 axis . for any fixed orbital state @xmath65 , with @xmath66 , the time evolution of the spin
is described by the ( von neumann ) master equation for the density matrix @xmath67 , @xmath68 = -\frac{i}{\hbar}[h_z,\rho ] , \\
h_z = \frac{\hbar \omega_z}{2 } \left ( \begin{array}{cc } 1 & 0 \\ 0 & -1 \end{array } \right ) + \frac{\hbar \omega_\perp}{4 } \left ( \begin{array}{cc } 0 &
e^{- i \omega t } \\
e^{i \omega t } & 0 \end{array } \right ) + \frac{\hbar \omega_\perp}{4 } \left ( \begin{array}{cc } 0 & e^{i \omega t } \\ e^{-i \omega t } & 0 \end{array } \right ) , \label{zeemantermexplicitform}\end{gathered}\ ] ] where we defined @xmath69 , @xmath70 , and the states of the matrices correspond to @xmath71 . when @xmath72 , the final term of the chosen representation of @xmath73 , eq .
( [ zeemantermexplicitform ] ) , can be omitted because it only superimposes a fast and negligibly small oscillation to the dynamics . within this rotating wave approximation ,
the resulting set of differential equations is exactly solvable . when the spin is originally in the @xmath12 state and the oscillating field
is applied for @xmath74 , the probability @xmath75 of a spin flip oscillates according to @xmath76 where @xmath77 is the detuning from the resonance condition @xmath78 , and @xmath79 is the cycle duration .
we note that the resulting spin - flip probability is completely equivalent to eq .
( [ esrresult ] ) when the spin is initially down .
remarkably , in the presence of soi one finds that an oscillating electric field @xmath80 leads to an effective magnetic field @xmath81 with , in general , non - zero components perpendicular to the static magnetic field .
hence , spin rotations can efficiently be driven by purely electrical means ( edsr ) .
this may be achieved by applying ac voltages to nearby gates , at frequencies which are in resonance with the zeeman splitting , as recently exploited in experiments on nanowire - based inas and insb qds @xcite .
explicit expressions for @xmath82 are lengthy , and in the following we therefore comment on important properties found in an analysis for lateral qds with growth axis @xmath50 $ ] and harmonic confinement in the @xmath64-@xmath83 plane @xcite .
first , in contrast to esr , the edsr arises from coupling to other orbital states and therefore depends on the level spacing .
this can easily be seen , since @xmath84 , so that neither the dipolar term @xmath85 nor @xmath86 or @xmath87 couple the spin states in lowest order .
a unitary schrieffer - wolff transformation shows that the leading term for edsr is a combination of zeeman coupling and soi .
more precisely , the effective magnetic field in the ground state is @xmath88 , where @xmath89 is the static magnetic field and @xmath90 depends linearly on the electric field components in the @xmath64-@xmath83 plane and the parameters @xmath46 and @xmath91 @xcite .
we note that the resulting magnetic field is fully transverse and therefore most efficient .
it can be quenched if @xmath92 .
for typical gaas qds , edsr allows spin manipulation on a time scale of 10 ns with the current experimental setups @xcite .
an analysis for heavy - hole qds can be found in ref .
@xcite .
soi , in combination with the phonon field , also leads to spin relaxation , characterized by the relaxation time @xmath4 . for electrons in @xmath41$]-grown 2d
qds it has been calculated that this phonon - mediated mechanism results in @xmath93 at low temperatures , where @xmath94 is the zeeman splitting induced by a magnetic field @xmath95 , and @xmath96 is the orbital level spacing @xcite . for moderate magnetic fields
, this dependence agrees very well with experimental results @xcite .
as the magnetic fields become very large , the wavelengths of the phonons with energy @xmath94 eventually become much smaller than the size of the qd , i.e. , the phonon - induced effects average out when integrating over the electron wave function and @xmath4 increases rather than decreases @xcite .
maximal relaxation rates are usually observed when the phonon wavelength matches the dot size @xcite . in the other limit , for very small magnetic fields , the derived expression for @xmath4 diverges .
this is because the above theory focuses on single - phonon processes , so that only phonons in resonance with the zeeman transition can contribute .
theorem forbids soi - induced spin relaxation in the absence of a magnetic field @xcite , which is also the reason why edsr requires the presence of a finite magnetic field .
when two - phonon processes are included , @xmath4 converges to a finite value @xcite .
holes are an attractive alternative to electrons because of the suppressed contact hyperfine interaction with nuclear spins ( see subsection below ) .
phonon - mediated spin relaxation has also been analyzed in detail for flat @xmath41$]-grown heavy - hole qds with magnetic fields along the confinement axis @xcite . for low magnetic fields , one finds @xmath97 due to dresselhaus soi , which is the same dependence as in the electron case , while the contribution due to rashba soi is @xmath98 .
the analysis shows that the spin relaxation time for heavy - holes can be comparable to or even longer than that for electrons when the qd is strongly 2d , illustrating that holes are very sensitive to confinement @xcite .
for instance , @xmath99 has been measured for heavy - holes in self - assembled ingaas qds @xcite . in the limit @xmath100 ,
the relaxation times are determined by two - phonon processes .
these have been included theoretically @xcite , suggesting times @xmath4 on the order of milliseconds , in good agreement with values observed in experiments @xcite .
notably , it has been shown that the upper limit @xmath101 is fulfilled in both the electron @xcite and the hole @xcite cases discussed above , in contrast to the naively expected relation @xmath29 .
furthermore , theory predicts that electron spin relaxation is drastically suppressed for a certain magnetic field direction when @xmath102 in @xmath41$]-grown qds @xcite .
[ in passing we note that in this special limit , a new symmetry in spin space emerges giving rise to interesting spintronics effects in quasi-2d systems @xcite .
] tuning the rashba coefficient via electric fields , this effect should be observable in an experiment with a vector magnet .
the analysis of soi - mediated relaxation has been extended to qds with two electrons , forming spin singlet and triplet states , where magnetic field orientations with strongly suppressed spin relaxation were found to exist for arbitrary rashba and dresselhaus coefficients @xcite .
the ( relative ) strengths of @xmath46 and @xmath91 may be found via the singlet - triplet anticrossings for magnetic fields applied in growth direction , or by measuring the `` magic angles '' at which the singlet - triplet anticrossings and thus the corresponding singlet - triplet relaxations vanish in leading order of the soi @xcite .
recently , a formula has been derived for lateral double quantum dots ( dqds ) , which quantifies the level splitting at the singlet - triplet anticrossing in terms of various parameters @xcite .
this formula should allow to extract both the spin - orbit parameters and also the hyperfine coupling from transport or charge sensing experiments in such dqds @xcite .
effects of hyperfine interaction will be discussed in following .
a qd typically consists of @xmath103-@xmath104 atoms , so that an electron or hole confined to the qd overlaps with a large number of nuclear spins .
the nuclear spin bath itself reveals large lifetimes , indicated by the long dipole - dipole correlation time among nuclear spins , @xmath105 0.1 ms in gaas @xcite , but presents the main source of electron and hole spin decoherence .
this is due to the hyperfine interaction among electron and nuclear spins , for which three different mechanisms can be derived from the dirac equation @xcite .
the first one is the ( isotropic ) contact hyperfine interaction , which is the most relevant mechanism for conduction band electrons . for holes , where the bloch functions are @xmath106- as opposed to @xmath107-type ,
the anisotropic hyperfine interaction and coupling to the orbital angular momentum become dominant instead . below
we summarize the effects and opportunities of a nuclear spin bath in more detail .
when we assume that the external magnetic field , if present , is oriented along the @xmath44 axis , the hamiltonian of an electron spin @xmath108 coupled to a bath of nuclear spins @xmath109 reads @xmath110 where @xmath111 is the vector of spin-1/2 pauli matrices , @xmath1 is the effective electron @xmath1 factor , @xmath61 is the external magnetic field , @xmath112 ( positive for in , ga , as ) are the contact hyperfine coupling constants , and @xmath113 , @xmath114 , are the raising and lowering operators for nuclear and electron spin , respectively @xcite .
the effects of the nuclear spins on an electron spin in a qd can thus be described in terms of an effective magnetic field @xmath115 , referred to as the overhauser field .
its component @xmath116 along the @xmath44 axis changes the total zeeman splitting by the overhauser shift , while transverse components couple the spin states @xmath12 and @xmath13 through electron - nuclear - spin flip - flop processes @xcite . to avoid confusion
, we mention that the term overhauser field is typically used both for the 3d effective nuclear magnetic field and its component @xmath116 .
the largest possible value for @xmath117 , obtained for a fully polarized bath , is @xmath118 , where @xmath119 is the averaged effective hyperfine coupling constant and @xmath120 is the ( average ) quantum number for the nuclear spin . for gaas , @xmath121 and @xmath122 , thus @xmath123 for the bulk @xmath1 factor -0.44 ,
and we note that @xmath124 is independent of the dot size @xcite . without further preparation ,
the @xmath125 nuclear spins inside a qd are in a superposition of states with different fields @xmath116 , statistically distributed around a mean value @xmath126 , where @xmath127 is the nuclear spin polarization along the @xmath44 axis .
unless @xmath128 , the width of this distribution is on the order of @xmath129 , i.e. , a few ( tens of ) mt for typical gaas qds of @xmath130 nuclear spins @xcite .
these internal fluctuations lead to dephasing and reduce the electron spin coherence time in gaas dots to a few ns only @xcite .
the associated decay of the transverse spin is gaussian and the decay time scales @xmath131 , when @xmath132 and homogeneous coupling are assumed for simplicity @xcite .
one possibility to prolong the lifetimes , apart from increasing the dot size , is hence to polarize the nuclear spins , which will be discussed in more detail in paragraph [ secsubsub : dnp ] .
however , referring to the factor @xmath133 obtained for @xmath134 , @xmath135 is required to reduce the decoherence by a factor of ten . ,
@xmath13 , decoherence refers to transitions of type `` @xmath12 and @xmath13 '' @xmath136 `` either @xmath12 or @xmath13 '' , and can be quantified by the time decay of the transverse spin @xmath25 as described in sec .
[ sec : systemsanddefinitions ] .
main : schematic decay of @xmath25 for a quantum dot electron with large zeeman splitting , assuming that the nuclear spin bath has been prepared in a narrowed state and that echo pulses are absent .
the sketch illustrates the variety of decay laws which the system proceeds through .
we note that the initial quadratic decay occurs on an ultrashort time scale , while an additional quadratic shoulder appears at the transition from the power law to the exponential loss of coherence .
details can be found in refs .
the picture was taken from @xcite and is property of science magazine ( www.sciencemag.org ) . _ ] a second , attractive approach for lifetime prolongation is to narrow the intrinsic distribution for @xmath137 @xcite .
when the nuclear spin bath is initially in a less noisy , `` narrowed '' state , the electron spin decoherence induced by the finite width of possible @xmath116 is suppressed . in particular , this dephasing mechanism is overcome when the system is initially in an eigenstate to a field @xmath116 . in this case , the decoherence time is no longer @xmath138 , but @xmath139 , so that the coherence times can be increased by several orders of magnitude @xcite .
the decay dynamics clearly differ from the gaussian behavior which results from internal dephasing .
in fact , an entire zoo of decoherence laws has been found , with a time decay that can proceed through several different stages , fig .
[ figuredecoherencescience ] @xcite .
the reason for this feature is a rather long bath correlation time of order @xmath140 @xcite .
the dynamics of the isolated electron spin interacting with the nuclear bath are therefore history - dependent ( non - markovian ) , and a markov approximation , for which the longitudinal and transverse spin components decay exponentially , is typically invalid . on time scales @xmath141 , where the dipolar coupling among nuclear spins can be ignored , this non - markovian decay has been analyzed in great detail , and we list a few key results in the following . assuming that the externally induced zeeman splitting is large , such that @xmath142 for @xmath120 of order unity , i.e. , @xmath143 for gaas , direct electron spin flips are energetically forbidden , which gives rise to pure dephasing of the electron spin @xcite . under these conditions ( perturbative approach possible ) , the various stages passed through by the transverse electron spin dynamics have been calculated with one unified and systematic method based on expansion of a generalized master equation @xcite .
these stages include an ultrashort quadratic decay and an initial ( partial ) power law decay , followed by a quadratic shoulder , a dominant exponential decay , and a long - time power law decay
. however , the exact behavior depends on various parameters , such as the qd dimensionality @xcite . the markovian regime , which gives rise to the exponential decay , is reached for sufficiently large @xmath61 , and analytic expressions for the decoherence time @xmath144 have been found @xcite .
this analysis was also of interest from a technical point of view , since it verified that calculations based on high - order expansions of a leading - order effective hamiltonian can have limited validity .
in particular , a notable modulation of the decay envelope found at long times can not be obtained with an effective hamiltonian ( see ref .
@xcite and references therein ) . observing these additional oscillations
experimentally would be a desirable confirmation of the theory . at low magnetic fields ,
an expansion of the generalized master equation is not possible , and the spin dynamics still are not understood in detail . a list of available approaches allowing for some insight into the low - field regime can be found in refs
. @xcite .
for instance , the system is exactly solvable in the special case when the nuclear spin bath is initially fully aligned @xcite . independently of @xmath61 , it turns out that both @xmath17 and @xmath25 first perform small oscillations due to coherent exchange with the nuclear spin bath . after the bath correlation time of order @xmath140 ,
electron spin coherence is irreversibly lost and the spin components converge to constant values , slightly below the initial values only , where the system remains until dipole - dipole interactions among the nuclear spins become relevant @xcite . for @xmath145 , assuming a gaussian envelope wave function , this asymptotic time decay evolves according to @xmath146 and @xmath147 for 3d and 2d qds , respectively , where @xmath148 . at large @xmath61 , this decay is @xmath149 and @xmath150 in 3d and 2d dots , in agreement with the systematic solutions of the generalized master equation @xcite .
the spin dynamics have also been investigated for heavy - holes in quasi-2d qds @xcite . assuming the strong confinement along the @xmath44 axis , we recall that the states can be classified according to their angular momenta ( effective spins ) @xmath151 , where @xmath152 quantifies the size and @xmath153 is the @xmath44-projection in units of @xmath154 .
this results from the strong soi in the valence band , coupling the electron spin to the @xmath106-type bloch functions .
heavy - holes ( hhs ) have @xmath155 , while light - hole ( lh ) states have @xmath156 , and the two bands are energetically well separated in 2d - like qds .
even though the contact hyperfine term is absent , the remaining mechanisms , i.e. , anisotropic hyperfine interaction and coupling to the orbital angular momentum , turn out to be rather strong , in typical iii - v compounds only one order of magnitude weaker than the contact hyperfine interaction for electrons @xcite .
remarkably , the coupling of the hh to the nuclear spin bath takes on a simple ising form in leading order , @xmath157 , where @xmath158 is the coupling to the @xmath159th nucleus , @xmath160 denotes the @xmath44 component of the @xmath159th nuclear spin in units of @xmath154 , and @xmath161 is the hh pseudospin operator with eigenvalues @xmath162 for @xmath163 @xcite .
this clearly differs from the heisenberg - type contact hyperfine interaction @xmath164 , eq .
( [ equationflipflop ] ) , because transverse components are basically absent in the case of holes . as for electrons
, one can distinguish between different initial configurations . in the case of an unprepared , inhomogeneously broadened nuclear spin bath ,
the transverse hole spin decay depends on the orientation of the external magnetic field @xmath60 . for zero field or @xmath165 , dephasing results in a gaussian time decay , as for electrons , with time scales of typically a few tens of ns @xcite
. however , due to the ising - type rather than heisenberg - type hh - nuclear - spin interaction , the situation changes drastically for an in - plane magnetic field @xmath166 .
since the hyperfine fluctuations are now purely perpendicular to the applied field , one finds that the transverse hole spin decays @xmath167 at long times in the limit @xmath168 , where @xmath169 is the externally induced zeeman splitting , @xmath170 is the in - plane hh @xmath1 factor , and @xmath171 is the variance of the nuclear field @xmath172 @xcite . for typical gaas qds and magnetic fields of a few tesla ,
the associated decay times are long , around tens of @xmath173s @xcite . only a few months after these calculations were published , an experiment on self - assembled ingaas qds with @xmath166 confirmed that hh spins in 2d - like qds are highly coherent , with @xmath174 ( @xmath175 with @xmath176 probability ) reported for the setup under study @xcite . for applications which require large zeeman splittings ,
an in - plane magnetic field may be inconvenient because @xmath170 is usually much smaller than the hh @xmath1 factor @xmath177 along the axis of strong confinement .
long coherence times for @xmath165 can be achieved as well , namely by preparing the nuclear spin bath in a narrowed state @xcite .
when the nuclear spins are initially in an eigenstate of @xmath172 , with @xmath165 , decoherence can only result from additional transverse terms in the hh - nuclear - spin coupling which then allow for flip - flop processes .
these additional terms mainly arise from coupling to neighboring bands , i.e. , the conduction band , lh band , and split - off band , and are about one to two orders of magnitude weaker than the dominant ising term @xcite .
it turns out that the time decay of the transverse spin due to band hybridization is purely exponential , and that the decoherence time @xmath5 can be tuned over several orders of magnitude via the applied magnetic field .
in fact , decoherence due to nuclear spins can be so strongly suppressed that other mechanisms , like the dipole - dipole interaction or the phonon bath , should take over as the dominant sources of transverse spin decay @xcite .
calculations on self - assembled qds showed that the ising - like form of the hyperfine coupling is preserved for realistic strain distributions @xcite .
the strain considerably affects the hyperfine - induced hole spin decoherence , largely through coupling to the conduction band , allowing to tune @xmath5 by an order of magnitude for fixed zeeman splittings @xcite .
as summarized above , the electron and hole spin decoherence induced by nuclear spins can be strongly suppressed when the nuclear spin bath is initialized in a narrowed state .
moreover , the effective nuclear magnetic field , up to @xmath178 for electrons in gaas qds , allows to realize large magnetic field gradients among neighboring qds @xcite and to tune the resonance energies in optically active qds over several tens of @xmath173ev @xcite .
therefore , nuclear spins are more and more considered a source of opportunity rather than trouble , leading to enormous experimental efforts in this field .
dynamic nuclear polarization ( dnp ) schemes are usually based on the isotropic contact hyperfine interaction among electron spin and nuclear spins .
as illustrated in paragraph [ secsubsub : electronspindecoherence ] , eq .
( [ equationflipflop ] ) , the transverse components of the overhauser field allow to polarize the nuclei via electron - nuclear - spin flip - flop processes , building up a large nuclear field @xmath179 with maximum overhauser shift @xmath180 .
an example for dnp via the hole spin can be found in ref .
@xcite .
dnp has been achieved experimentally by optical @xcite , electrical @xcite , and also magnetical @xcite driving , where it is impossible to completely list the large variety of approaches .
polarizations @xmath181 have been reported so far @xcite , however , achieving @xmath182 remains a very challenging task .
interestingly , many dnp schemes feature an intrinsic feedback mechanism which drives the system towards fixed , stable nuclear field values , so that the width of the nuclear field distribution is narrowed at the same time @xcite .
similar effects have been demonstrated using pulsed optical excitation on an ensemble of qds @xcite , and it has been shown that efficient feedback loops may also be included intentionally @xcite .
further promising approaches for the preparation of narrowed states are based on indirect measurement @xcite .
we mention that all decay properties described earlier in this section correspond to the free - induction decay , i.e. , the case where the system evolves in the absence of externally applied control sequences .
the dephasing due to inhomogeneous broadening can be undone to a large extent by applying spin - echo pulses , notably increasing the spin coherence times @xcite . in parallel
, an alternative trend is to bypass the interaction with nuclear spins completely by switching to host materials such as ge and si , which can be grown nuclear - spin - free .
examples for both approaches will be discussed in the next section , where we review recent progress toward quantum computation with spins in qds .
finally , we note that an alternative route to reduce the nuclear spin noise ( besides narrowing and dnp ) would be to polarize the nuclear spins by freezing them out , either by applying a sufficiently strong magnetic field on the order of @xmath183 at a few mk @xcite or by inducing an ordering transition of the nuclear spin system due to rkky interactions .
the latter phenomenon has attracted a lot of interest in recent years , and we refer the interested reader to the literature @xcite .
since the first proposals in the 1990s , researchers worldwide have been working hard toward the ambitious goal of implementing a quantum computer . in quantum dots ( qds ) , which we focus on in this report , seminal progress has been made within the past few years . in the following ,
we summarize and discuss some of the key results , where we distinguish between self - assembled , lateral , and nanowire - based qds .
a table at the end of this section , table [ tablecomparison ] , summarizes relevant information such as the measured lifetimes , and compares @xmath5 to reported operation times as commented below .
quantum systems are sensitive , and errors inevitably occur in any realistic device .
therefore , schemes for fault - tolerant quantum computing have been developed , where errors can automatically be corrected as long as they occur with low enough probability .
the latter condition can usually be quantified in terms of a threshold rate @xcite .
for instance , standard error correction schemes require that at least @xmath184 gate operations can be carried out within the decoherence time of a qubit @xcite .
a few years ago , a novel scheme has been presented , referred to as the surface code @xcite .
the logical qubits are encoded within several physical qubits each , all of which are arranged in a 2d lattice with nearest neighbor interactions .
logical single- and two - qubit gates are then performed via a series of projective measurements .
this code is probably the most powerful quantum computing scheme presently known , with a remarkably large error threshold around 1% @xcite , so that @xmath185 operations per decoherence time may already be sufficient . when errors for readout become negligible , error correction in the surface code is even possible up to a threshold rate of currently 18.5% @xcite .
the viability of topological error correction has recently been demonstrated in a first proof - of - principle experiment @xcite .
for further information on fault - tolerant quantum computation with the surface code we refer to refs .
a general overview on fault - tolerance is provided in @xcite . as opposed to gate - defined qds
, the confinement in self - assembled dots purely arises from the conduction and valence band offsets of the involved materials , leading to strong confinement on a very small scale in all three dimensions .
therefore , self - assembled qds are typically operational at @xmath186 , the boiling temperature of @xmath187he .
moreover , they are optically active and feature strong interband transitions with `` almost hard '' selection rules @xcite . exploiting this property , self - assembled qds
are primarily studied using optical means , and heterostructures have been designed which allow precise control over the charge states @xcite . for a qubit with basis @xmath188 and @xmath189
, the general qubit state can be written as @xmath190 , neglecting global phases , where @xmath191 and @xmath192 correspond to the polar and azimuthal angle of a point on the bloch sphere .
prerequisites for the implementation of a quantum computer are the abilities to initialize , to control , and to read out such a qubit state .
all of this has been achieved by now .
first , the spins of both electrons @xcite and holes @xcite can be initialized with @xmath193 fidelity .
second , ultrashort optical pulses , combined with an externally induced zeeman splitting , have allowed for complete quantum control , i.e. , arbitrary rotations on the bloch sphere , with operation times on the order of only a few ps .
again , this has successfully been demonstrated both on single electrons @xcite and single holes @xcite .
recently , initialization and coherent control have also been reported for two - particle qubits defined by the spin singlet and triplet states of electrons @xcite and holes @xcite in vertically stacked qds .
the latter , also referred to as qd molecules , form naturally during growth when a second layer of qds is grown on top of another .
even though the position of the dots in the first layer is arbitrary , the qds in the second layer will form right on top of the first ones due to strain in the tunnel barrier grown inbetween .
finally , several methods have been developed to read out the spin states .
these include time - averaged readout via faraday rotations @xcite , kerr rotations @xcite , and resonance fluorescence @xcite . in addition
, time - resolved kerr rotation spectroscopy has been reported @xcite , and it has been shown that qd molecules allow to measure the spin state of a single electron in real time via the resonance fluorescence @xcite .
in the latter case , the presence of a second dot enables to use different optical transitions ( laser energies ) for the initialization and readout steps @xcite .
dephasing and decoherence times have been measured .
for single - electron spin qubits , dephasing results in @xmath194 @xcite depending on the exact system , while the observed decoherence times are @xmath195 @xcite .
however , it has been verified already that @xmath26 can be significantly prolonged by narrowing the distribution of the nuclear field @xcite .
for single holes , the reported @xmath196 @xcite and @xmath197 @xcite are similar to those for electrons , hence shorter than initially expected , which is attributed to electrical noise @xcite .
measurements based on coherent population trapping , i.e. , in the frequency domain as opposed to the time domain , have revealed a much longer hole - spin dephasing time @xmath198 @xcite .
electrical noise is also the reason for the short @xmath199 @xcite and @xmath200 @xcite observed for singlet - triplet qubits from electrons and holes , respectively .
when the qd molecule is operated in a regime where the singlet - triplet splitting is less sensitive to fluctuating electric fields , @xmath26 can be increased by several orders of magnitude , and dephasing times up to have recently been measured for coupled electrons @xcite .
importantly , comparing the listed @xmath5 to the notably short operation times of a few ( tens of ) ps illustrates that the threshold factor of @xmath184 for standard quantum error correction schemes has already been exceeded both in electron and hole systems . for
reported spin relaxation times @xmath4 , see table [ tablecomparison ] .
the optical properties make self - assembled qds highly promising candidates for applications as single - photon sources , and designs for enhanced extraction efficiencies are being developed @xcite .
moreover , they may present an interface between `` stationary '' and `` flying '' qubits .
recently , interference of single photons from two separate qds has been demonstrated @xcite , which is a promising approach for generating entanglement and implementing two - qubit gates between distant spins . since different qds , unlike atoms , have different resonance energies , this requires the ability to tune the level structure of a dot for the photon energies to be matchable .
such fine tuning may be achieved via strain @xcite , via the stark effect @xcite , or by polarizing the nuclear spins up to a desired ( overhauser ) nuclear magnetic field .
dynamic nuclear polarization in single self - assembled qds has been studied in detail @xcite .
for instance , schemes exist both for high @xcite and low @xcite external magnetic fields which allow for continuous , bidirectional tuning of the overhauser field , fully controlled by the laser wavelength .
both schemes are also capable of narrowing the width of the nuclear field distribution @xcite . as an alternative to the widely studied iii - v compounds ,
self - assembled qds can also be grown from group iv materials .
a prominent example are self - assembled ge qds on si substrates . due to the indirect band gap of ge and si ,
these dots are not as optically active as typical iii - v qds @xcite .
self - assembled ge / si qds are subject to large experimental efforts , and detailed knowledge about growth and the electrical and optical properties has been gained @xcite .
substantial progress has been made on implementing gate - controlled qubits within the two - dimensional electron gas ( 2deg ) of a heterostructure .
experiments have predominantly been carried out on gaas qds within algaas / gaas heterostructures , which therefore should be considered to be the host material in the following , unless stated otherwise .
mainly , two different approaches for encoding the qubit have emerged .
the first one follows the original proposal @xcite , using the spin eigenstates @xmath12 and @xmath13 of single electrons .
the second scheme uses the singlet @xmath201 and triplet @xmath202 states of two electron spins , forming an @xmath203-@xmath204 qubit @xcite .
we summarize the results for both approaches in the following .
alternatively , @xmath203-@xmath205 qubits @xcite and qubits from three - spin states @xcite are currently under investigation . in the @xmath203-@xmath204 approach ,
every qubit is formed by two electrons in two adjacent qds , where the charge configuration and the overlap of the electron wave functions depend on the shape of the confining potential , which in turn is determined by the applied gate voltages . within the frame of this review
it is sufficient to distinguish the two charge configurations ( 1,1 ) and ( 0,2 ) , which denotes that the electrons are found either in different dots or both in the `` right '' qd , respectively .
coherent rotations @xmath206 are induced by a magnetic field gradient between the two qds , typically achieved by dynamic polarization of the nuclear spins @xcite or by a nearby positioned micromagnet @xcite . when the barrier is reduced such that the wave functions strongly overlap
, the finite exchange energy gives rise to coherent rotations @xmath207 , i.e. , @xmath208 , thus allowing for arbitrary rotations on the bloch sphere @xcite .
these single - qubit operations can be carried out within a few ns only @xcite , and two - qubit gates may be implemented by capacitive coupling , where the charge configuration in one double quantum dot ( dqd ) affects the exchange energy and hence the precession frequency in the other dqd @xcite .
charge - conditional phase flips of an @xmath203-@xmath204 qubit have recently been demonstrated in four - dot systems @xcite , and it was verified experimentally that the resulting cphase gate between two @xmath203-@xmath204 qubits is entangling @xcite . for initialization , the potential energy in one of the dots is reduced such that both electrons tend to occupy the ground state of the same qd .
this sets the qubit in the singlet state , because the symmetric orbital part of the two - electron wave function requires an antisymmetric contribution of the spin .
after operation in the ( 1,1 ) regime , the same idea also allows for spin - to - charge conversion and therefore presents a popular basis for readout schemes .
having reduced the potential energy in one qd , the system changes to ( 0,2 ) for @xmath209 , but remains in ( 1,1 ) for @xmath210 due to pauli exclusion .
the charge state can be detected via nearby quantum point contacts ( qpcs ) @xcite , which may furthermore be embedded in radio - frequency ( rf ) impedance matching circuits to allow for faster readout @xcite .
however , the quality of the outcome ( readout visibility , see also sec .
[ secsub : overviewprogress ] ) strongly depends on the triplet relaxation time at the measurement point , which may be clearly reduced in the presence of large magnetic field gradients @xcite , and implementing schemes for fast and reliable single - shot readout therefore remains an important task .
recently , single - shot measurements with measurement times down to 100 ns have been reported , using an rf sensor qd which is much more sensitive than standard qpcs @xcite . as an alternative approach , dispersive readout of spin singlet and
triplet states has been demonstrated with an rf resonant circuit coupled to a dqd @xcite .
also , readout via spin - dependent tunnel rates has been achieved for the singlet and triplet states in a single qd @xcite .
even though the random distribution of the nuclear spins leads to rapid dephasing within @xmath211 @xcite , two - electron spin states feature very long coherence times .
a single echo pulse increases the dephasing time to a few @xmath173s already @xcite , and more sophisticated pulse sequences have demonstrated coherence times on the order of a hundred @xmath173s @xcite , where @xmath212 currently corresponds to the largest value reported so far @xcite . assuming that the two - qubit gates can be operated on a similar time scale ( ns ) as the single - qubit gates , the threshold value of @xmath184 operations per decoherence time is hence clearly exceeded .
as we summarize in the following , this similarly holds for the case where the qubits are encoded in the spin states of single electrons .
the two eigenstates of a single electron in a qd , @xmath12 and @xmath13 , are split by an effective zeeman energy via coupling to external and internal ( overhauser ) magnetic fields .
this energy difference gives rise to coherent single - qubit rotations @xmath213 .
rotations about the second axis of the bloch sphere , @xmath214 , can be driven by means of e(d)sr , which we introduced in sec .
[ sec : soiandnuclearspins ] of this review .
finally , two - qubit gates such as @xmath32 and swap can be implemented by controlling the overlap of the wave functions and hence the exchange energy for neighboring electrons .
thirteen years after publication of the original proposal @xcite , the elementary unit of an all - electrical `` spin - qubit processor '' has now been implemented for the first time , demonstrating independently controllable single - spin rotations combined with inter - dot spin exchange in a dqd @xcite .
an important part of the setup in ref .
@xcite is a micromagnet at the sample surface , whose stray field provides a time - independent magnetic field gradient at the qds .
this gradient is useful for two reasons .
first , it allows for efficient and electrically driven esr ; an oscillating electric field slightly shifts the position of the electron in the qd , so that the electron `` feels '' an oscillating magnetic field of the same frequency without the need for soi @xcite .
second , it leads to shifted resonance frequencies in adjacent qds via a difference in the zeeman energies , so that neighboring qubits can be addressed individually @xcite .
performance of the implemented spin - qubit processor was tested via a time - averaged readout scheme , using a nearby qpc to measure the average charge configuration when the system is tuned to the pauli spin blockade regime ( see also @xmath203-@xmath204 qubits above ) @xcite .
a fully operational unit for quantum computation requires precise initialization and single - shot readout of the individual qubit states , which therefore still needs to be included .
latest developments are promising @xcite , and independent single - shot readout of two electron spins in a dqd , with fidelities close to 90% , has recently been reported @xcite . in the experiments of ref .
@xcite , a time around @xmath215 was required for spin readout and initialization to @xmath12 , slightly shorter than the measured spin relaxation time of @xmath216 .
the latter strongly depends on the regime of operation , i.e. , the applied gate voltages , and can take values @xmath217 @xcite .
the decoherence times @xmath5 can be assumed to be much longer than a @xmath173s . for a single echo pulse and a magnetic field of @xmath218 only , a decay time near @xmath219 has been measured , and clearly longer coherence times can be expected at higher magnetic fields and more sophisticated pulse sequences @xcite .
comparing @xmath5 to the typical operation times , we find that the threshold rates of proposed ( standard ) error correction protocols have already come close within reach , as summarized in the following . two - qubit gates are fast .
for instance , swap operations @xmath220 require a time of order @xmath221 , where @xmath222 is planck s constant and @xmath152 is the exchange energy .
they are routinely carried out within @xmath223 @xcite , and swap times @xmath224 have been demonstrated @xcite .
furthermore , single - qubit rotations induced by the zeeman splitting typically occur on a sub - ns time scale . using the bulk @xmath1 factor @xmath225 for gaas and @xmath226 as the magnetic field along the quantization axis ,
the @xmath227-rotation @xmath228 requires a time of order @xmath229 .
the efficiency of the universal set of qubit gates is therefore limited by the operation time @xmath230 for coherently driven spin flips @xmath16 , where @xmath2 is the amplitude of the oscillating magnetic field perpendicular to the quantization axis ( see also paragraph [ secsubsub : esrandedsr ] ) . recently reported spin - flip times are on the order of several tens of ns , both for magnetically @xcite and electrically @xcite driven rotations , corresponding to @xmath231 .
the efficiency of electrically controlled rotations may , among others , be improved by increasing the magnetic field gradient @xcite , or even , in a brute - force approach , by switching to host materials with stronger soi @xcite and larger @xmath1 factors .
alternatively , coherent single - spin rotations about an arbitrary axis may be implemented using an auxiliary spin in a nearby qd with different zeeman field , which enables a purely exchange - based control without the need for soi or oscillating fields @xcite .
such exchange - controlled single - qubit gates have not yet been realized experimentally ; theoretical results , however , are promising and point toward high fidelities , with gating times @xmath232 @xcite . for both qubit encoding schemes , control over the nuclear spin bath is of great benefit .
various methods for dynamic nuclear polarization ( dnp ) have been developed and demonstrated @xcite , all of which may be used to generate large magnetic field gradients between neighboring qds .
such gradients can be measured quantitatively via the cycle duration of oscillations @xmath206 @xcite , and differences @xmath233 in the overhauser field have been induced and subsequently exploited for coherent rotations on the bloch sphere @xcite .
in addition , schemes exist to reduce the width of the nuclear field distribution @xcite , allowing to strongly prolong the dephasing times @xcite . besides gaas qds , other promising host materials are under investigation .
in particular , rapid progress has been made on si qds within si / sige heterostructures @xcite .
for instance , slow electron spin relaxation @xmath234 has recently been observed in this material via single - shot readout @xcite .
also , single - shot measurements have allowed to extract relaxation times @xmath235 for @xmath203-@xmath204 qubits operated near the ( 1,1)-(0,2 ) charge state transition @xcite . moreover , a hyperfine - induced dephasing time @xmath236 has been deduced from ensemble - averaged measurements on an @xmath203-@xmath204 qubit @xcite , which is nearly two orders of magnitude longer than the dephasing times measured in gaas .
as opposed to self - assembled qds , gate - defined systems can not confine electrons and holes at the same time and are therefore optically inactive @xcite .
currently , experiments are almost exclusively carried out on gated 2degs , and new experimental challenges may be encountered when 2d hole gases are used instead @xcite . however , considering the long lifetimes predicted for heavy - holes in qds @xcite and the fact that strong magnetic field gradients may be induced via micromagnets @xcite , we think that hole spins can present a valuable alternative to electron spins .
semiconducting nanowires attracted a lot of interest as promising platforms for majorana fermions @xcite , field effect transistors @xcite , programmable circuits @xcite , single - photon sources @xcite , lasers @xcite , and others .
qds therein form when the confinement in the transverse directions , provided by the wire geometry , is supplemented with an additional confinement in the longitudinal direction , which can be achieved both via electric gates and via structured growth of materials with suitable band offsets . in the first case , coupling between neighboring qds is easily controllable via the gate voltages , while the second case is highly attractive for optical processes because electrons and holes can be stored at the same time ( see also self - assembled qds ) .
nanowires are thus versatile and may present a valuable link between `` stationary '' and `` flying '' qubits .
for instance , similarly to self - assembled qds @xcite , voltage - controlled charging and spin - dependent selection rules have been demonstrated on optically active inasp qds embedded in inp wires @xcite .
nanowires have been grown from a variety of materials , all of which feature different properties and advantages .
a prominent host material is inas , which is known for its strong soi and large @xmath1 factors .
experiments on electrons in inas nanowire qds revealed a spin - orbit length @xmath237 , i.e. , a spin - orbit energy @xmath238 @xcite , along with @xmath1 factors @xmath239 @xcite .
these features allow for fast spin rotations via edsr @xcite , and spin - flip times @xmath240 have already been demonstrated on single - electron qubits @xcite . on the other hand , the coherence time in ref .
@xcite was found to be rather short , only @xmath241 for a single echo pulse and @xmath242 for several pulses , which is considered to be attributable to the large nuclear spin 9/2 of in .
the relaxation times clearly exceeded the measurement range , i.e. , the @xmath173s time scale @xcite , consistent with the theoretically predicted @xmath4 of several @xmath173s to ms @xcite .
experiments with singlet - triplet states in an inas nanowire dqd in the multi - electron regime showed that spin relaxation may be suppressed via tuning of the interdot coupling @xcite . besides inas
, also insb has recently attracted a lot of attention , where edsr spectroscopy on the two - electron states of a gate - defined nanowire dqd revealed @xmath243 and very large @xmath1 factors @xmath244 @xcite . , surrounded by a si shell of thickness @xmath245 leading to static strain . when the nanowire is supplemented with confinement along the wire axis @xmath44 , a quantum dot of radius @xmath246 and ( effective ) length @xmath247 forms .
( b ) energy gap between the two lowest kramers doublets in a longitudinal ge quantum dot of @xmath248 and different lengths @xmath247 , as a function of relative shell thickness @xmath249 .
the si shell allows to change the level splitting by several mev , which is particularly useful for implementing spin qubits .
( c ) top : hole spectrum as a function of longitudinal wave number @xmath250 for a typical ge / si wire of @xmath248 and @xmath251 in the presence of an electric field @xmath252 and a magnetic field @xmath253 along @xmath64 ( see panel a ) , illustrating strong spin - orbit interaction and sensitivity to magnetic fields , which are prerequisites for efficient qubit manipulation .
the spin - orbit energy @xmath254 , resulting from direct , dipolar coupling to @xmath255 , is more than 15 times larger than the reported value for inas @xcite . at @xmath256
, @xmath257 opens a zeeman gap of @xmath258 , corresponding to @xmath259 .
bottom : expectation value of the effective hole spin components @xmath260 , @xmath261 , and @xmath262 for the ground state plotted above . when the fermi level is set within the gap at @xmath263 , the wire transports opposite spins in opposite directions , the characterisitic feature of a helical mode
. therefore , ge / si nanowires also provide a promising basis for spin filters and majorana fermions @xcite . ] a promising alternative to the iii - v compounds are ge / si core / shell nanowires , which can be grown nuclear - spin - free . the valence band offset at the ge / si interface is large , @xmath264 , so that holes accumulate naturally in the ge core without the need for dopants @xcite .
high mobilities @xcite and very long mean free paths @xcite have been observed , along with a highly coherent charge transport seen through proximity - induced superconductivity @xcite .
progress has also been made on gate - controlled ge / si nanowire qds , even though the single - hole regime has not quite been reached yet @xcite .
experiments on such qds have recently revealed spin relaxation times near 1 ms @xcite .
furthermore , as illustrated in fig .
[ figurenanowires]b , the relative thickness of the si shell allows to vary the qd level splitting by several mev via the static strain @xcite .
remarkably , ge / si nanowires were also found to feature strong soi : theoretical studies of the low - energy hole states showed that direct , dipolar coupling to an electric field gives rise to a soi of rashba type ( `` direct rashba spin - orbit interaction '' ) which exceeds the standard rashba soi by one to two orders of magnitude @xcite .
figure [ figurenanowires]c ( top ) plots the ground state spectrum for a typical nanowire of @xmath265 core radius and @xmath266 shell thickness in a moderate electric field of @xmath267 perpendicular to the wire . the corresponding spin - orbit energy is @xmath268 , more than 15 times larger than the reported value for inas @xcite . the additional magnetic field in fig .
[ figurenanowires]c ( top ) along the electric field axis opens a gap in the spectrum ( @xmath269 at @xmath263 ) , illustrating sensitivity to magnetic fields , a prerequisite for efficient qubit manipulation .
the hole @xmath1 factors in ge / si nanowire qds are tunable both via the confinement and the magnetic field orientation @xcite .
all these properties should allow for electrically controlled qubits with long lifetimes and short operation times .
furthermore , ge / si nanowires present an outstanding platform for helical hole states and majorana fermions @xcite .
finally , we mention that helical states @xcite and majorana fermions @xcite can alternatively be realized in armchair carbon nanotubes without the need for a magnetic field , allowing for purely electrical setups . the required field , on the order of v / nm @xcite , is stronger than in the case of ge / si wires , but still below experimentally achievable limits .
> m0.109 | > m0.138 | > m0.138 | > m0.138 | > m0.138 | > m0.138 | > m0.138 | & & & & electrons & holes & single spins & @xmath203-@xmath204 qubits & electrons & holes lifetimes & + @xmath270 & + @xmath271 & + @xmath272 & + @xmath273 & + @xmath274 & + @xmath275 operation times & + @xmath276 & + @xmath277 & + @xmath278 & + @xmath279 & + @xmath280 & n.a .
+ @xmath281 & + @xmath282 & + @xmath283 & + 22 & + @xmath284 & + n.a . & + n.a .
readout & fidelities @xmath285 ( visibilities @xmath286 ) & + @xmath287a + resonance fluorescence in a qd molecule + + other : + faradayb and kerrc rotation spectroscopy , resonance fluorescenced & absorptione and emissionf spectroscopy + ( selection rules ) & + @xmath288 g + @xmath289i + spin - selective tunneling + + @xmath290h + spin - selective tunneling + ( two spins ) + + other : + photon - assisted tunnelingi & + @xmath291j + @xmath292k + spin - dependent charge distribution ( rf - qpcj / sqdk ) + + @xmath293l + spin - dependent tunneling rates + + other : + dispersive readoutm & @xmath294n pauli spin blockade & spin - dependent charge distribution ( sensor dot coupled via floating gate)o initialization & fidelities @xmath295 & @xmath296p + optical pumping & @xmath297q + optical pumping + @xmath296r + exciton ionization & + spin - selective tunneling , s adiabatic ramping to ground state of nuclear fieldt & pauli exclusionw & pauli spin blockaden & n.a .
+ ( single - hole regime not yet reached ) scalability & & & > m0.98 + a@xcite ; b@xcite ; cmeasured for single electrons in a narrowed nuclear spin bath @xcite and for two - electron states in qd molecules with reduced sensitivity to electrical noise @xcite , both via coherent population trapping . without preparation @xmath298 ,
see sec .
[ secsub : progressselfassembled ] . ; d@xcite ; e@xcite ; f@xcite .
measured through coherent population trapping .
other experiments revealed @xmath299 attributed to electrical noise @xcite . ; g@xcite ; h@xcite ; i@xcite ; j@xcite ; k@xcite ; l@xcite . achieved by narrowing the nuclear spin bath , @xmath300 otherwise @xcite . ; m@xcite ; n@xcite .
hyperfine - induced .
; o@xcite ; p@xcite ; q@xcite ; r@xcite ; s@xcite ; t@xcite ; uwhile ref .
@xcite gets close , we are currently not aware of a ramsey - type experiment where coherent rotations about the bloch sphere @xmath44 axis have explicitly been demonstrated as a function of time .
thus no value is listed .
however , @xmath301 should be short , on the order of 0.1 ns assuming a magnetic field of 1 t and @xmath302 as in bulk gaas . ; v@xcite ; w@xcite ; x@xcite ; y@xcite .
swap gates for @xmath203-@xmath204 qubits have not yet been implemented so far .
we therefore list the duration @xmath303 of a charge - state conditional phase flip . ;
zin ref .
@xcite , rotations about an arbitrary axis in the @xmath64-@xmath83 plane of the bloch sphere are reported instead , controlled via the phase of the applied microwave pulse . ; a@xcite ; b@xcite ; c@xcite ; d@xcite ; e@xcite ; f@xcite ; g@xcite ; h@xcite ; i@xcite ; j@xcite .
rf - qpc : radio - frequency quantum point contact @xcite . ; k@xcite .
rf - sqd : radio - frequency sensor quantum dot @xcite . ;
the paper demonstrates readout of the singlet and triplet states in a single quantum dot . ;
a radio - frequency resonant circuit is coupled to a double quantum dot . ;
n@xcite . the readout and initialization schemes in this experiment only determine whether two neighboring spin qubits are equally or oppositely oriented . ;
the scheme , operated in the multi - hole regime , distinguishes the spin triplet states from the spin singlet . ; p@xcite ; q@xcite ; r@xcite ; s@xcite ; t@xcite .
information about the nuclear field is required for the ground state to be known . ; u@xcite ; v@xcite .
[ tablecomparison ] table [ tablecomparison ] summarizes important information about the systems covered in this section , such as the longest measured lifetimes and shortest reported operation times ( as of april 2012 ) .
we would like to point out that , even though we studied the literature carefully , the provided summary is not intended to be complete , as further improvements might already have been achieved of which we had not been aware when completing this review .
the table also lists established initialization and readout schemes for each system , along with a ratio of the observed decoherence and gating times .
the latter illustrates that decoherence no longer presents the massive stumbling block it had been considered years ago .
typical values for characterizing readout schemes are the measurement fidelities @xmath285 and the visibilities @xmath286 .
we note that these are not exactly equivalent , as briefly explained in the following .
when @xmath304 denotes the error probability that the qubit state @xmath189 is incorrectly read as @xmath188 , the measurement fidelity for state @xmath189 is @xmath305 @xcite .
analogously , the measurement fidelity for the qubit state @xmath188 is @xmath306 .
one may define the readout fidelity for a particular experiment as @xmath307 , where @xmath308 ( @xmath309 ) is the probability that the system is initially in @xmath189 ( @xmath188 ) @xcite .
weighting @xmath189 and @xmath188 equally , this results in @xmath310 @xcite .
a more general quantity is the visibility @xmath311 , which is independent of @xmath308 and @xmath309 and presents a lower bound for the readout fidelity in a system @xcite .
we note that readout may also be characterized by the measurement efficiency defined in ref .
@xcite , where various readout schemes have been analyzed theoretically .
recently , pairwise control of the exchange interaction via electric gates has been demonstrated in a triple quantum dot @xcite .
such control is an essential requirement for most quantum computer architectures ( see fig .
[ schemeldivqc]b ) , and so the experiments of ref .
@xcite present an important proof of scalability .
large - scale quantum computers , however , must be capable of reaching a system size of several thousands of qubits .
this poses serious architectural challenges to the exchange - based qd scheme from ref .
@xcite , fig .
[ schemeldivqc ] , since the large amount of wires and metallic gates needs to be installed and operated on a very small scale .
a promising strategy to meet this challenge has recently been proposed @xcite ; long - distance spin - spin coupling can be achieved capacitively via floating gates , allowing to move the ( d)qds far apart .
the effective qubit - qubit coupling @xmath312 via floating gates can take remarkably large values @xmath313 @xcite .
these are comparable to the achievable exchange energies @xmath314 in typical gaas dqds @xcite , where we note that @xmath152 close to has already been realized @xcite .
the floating gates may be positioned on top of the sample , as sketched in fig .
[ figurefloatinggates]a , or may even be defined within the 2deg .
qubit - qubit coupling can be switched on and off by changing the relative positions of the qds ( charges ) with respect to the gates , allowing for all - electrical control @xcite .
proposed , scalable architectures for a quantum computer with floating gates are shown in figs .
[ figurefloatinggates]b - c . a key feature of the architectures suggested in ref .
@xcite is that they all consist of a two - dimensional lattice of spin qubits with nearest neighbor qubit - qubit interactions .
therefore , they all allow for the implementation of the surface code with its strikingly large error threshold around 1% ( see also sec .
[ sec : recentprogress ] and refs .
@xcite for further information ) .
can be reached @xcite .
quantum computer architecture using metallic floating gates on top of a two - dimensional electron gas .
( c ) an alternative architecture with qubits ( black dots ) implemented in nanowires ( vertical black lines ) . in both ( b ) and ( c ) , qubit - qubit interactions can be switched on ( off ) via gates by moving the qubits close to ( away from ) the corresponding metal discs .
the architectures provide a platform for the powerful surface code @xcite . _
all pictures were taken from @xcite and are property of the american physical society ( aps.org ) . _ ] is separated from the neighboring ground planes by a .
electron spin states ( thick green arrows ) in inas nanowire quantum dots ( qds ) serve as qubits . the qd confinement , illustrated by brown discs within the nanowires ( pink cylinders ) ,
can be realized with a suitable barrier material , such as inp , or with electric gates .
( a ) a large - diameter nanowire is oriented parallel to the transmission line and hosts qds with strong longitudinal confinement .
( b ) small - diameter nanowires , each hosting a qd with strong transverse confinement , are placed perpendicular to the transmission line .
an insulating material separates the wires from the superconducting resonator to prevent a current flow . in both setups
, the spin - orbit interaction enables all - electrical operation along with strong long - range interactions mediated by the photon field of the cavity @xcite . for the system in ( b ) , the time to coherently swap two spins was estimated to be around , and further improvement seems clearly possible . _
the pictures were taken from @xcite and are property of the american physical society ( aps.org ) . _ ] another scheme for long - distance coupling between spins uses the photon field of a cavity .
the original idea goes back to a proposal from 1999 , which suggests that laser - induced raman transitions can be used to couple the electron spin states of distant qds when embedded in an optical cavity @xcite .
a few years later , as an alternative to the standard cavity quantum electrodynamics ( qed ) , 1d superconducting transmission line resonators , which operate as on - chip microwave cavities , have been introduced and have launched the field of circuit qed @xcite . since then , several proposals for long - range spin - spin coupling via circuit qed have been made @xcite .
the direct coupling of a single spin to the magnetic component of the cavity electromagnetic field , however , is weak , and achieving strong interactions between a spin qubit and a cavity thus requires other mechanisms , resulting in a variety of suggested approaches .
these , for instance , are based on raman - type transitions among single - electron states in dqds @xcite , on @xmath203-@xmath204 qubits in qd molecules with ( nuclear ) magnetic field gradients @xcite , or on the single - electron states in dqds contacted to ferromagnetic insulators @xcite .
relatively recently , it has been proposed to use electron spins in inas nanowire qds , which feature a strong soi and thus enable efficient coupling between the qubits and the electric component of the cavity field @xcite .
investigated architectures are shown in fig .
[ figurestriplinesinas ] , where nanowires with strong longitudinal ( transverse ) qd confinement are placed parallel ( perpendicular ) to the transmission line in panel a ( b ) .
rotations of individual qubits can be driven through edsr @xcite , and two - qubit interactions can be turned on and off by changing the qd confinements with nearby gates , allowing for all - electrical control @xcite . for the setup of fig .
[ figurestriplinesinas]b , an operation time around was estimated for swapping two spins coherently , and further optimization seems clearly possible .
notably , it has been shown that circuit qed also allows for long - range coupling between certain types of molecular magnets @xcite .
when electron spins in qds were proposed for quantum computation , the experimental situation was not encouraging .
gate - controlled qds within 2degs were limited to around 30 or more confined electrons each , and techniques for single - qubit manipulation and readout were not available @xcite .
furthermore , decoherence from interactions with the environment was considered an almost insurmountable obstacle . within the past decade ,
the situation has changed dramatically , owing to continuous experimental and theoretical progress .
qds are now routinely controlled down to the last electron ( hole ) , and various schemes have been applied both for qubit initialization and readout . reducing the occupation number of qds to
the minimum is desirable for high - fidelity quantum computation @xcite ; however , larger fillings with a well - defined spin-1/2 ground state are also useful @xcite .
in addition , efficient single- and two - qubit gates have been demonstrated , allowing for universal quantum computing when combined .
the achieved gating times are much shorter than measured lifetimes , and it seems that one will soon be able to overcome decoherence to the required extent .
this is a major step toward the realization of a quantum computer .
while the field is very advanced for the `` workhorse '' systems such as lateral gaas qds or self - assembled ( in)gaas qds , rapid progress is also being made in the quest for alternative systems with further optimized performance .
first , this includes switching to different host materials .
for instance , ge and si can be grown nuclear - spin - free , and required gradients in the zeeman field may be induced via micromagnets .
second , both electron- and hole - spin qubits are under investigation , exploiting the different properties of conduction and valence band , respectively . finally , promising results are obtained from new system geometries , particularly nanowire qds .
future tasks can probably be divided into three categories .
the first one consists of studying new quantum computing protocols ( such as the surface code ) which put very low requirements on the physical qubits .
the second category refers to further optimization of the individual components listed in table [ tablecomparison ] .
for instance , longer lifetimes are certainly desired , as well as high - quality qubit gates with even shorter operation times . however , as decoherence no longer seems to present the limiting issue , particular focus should also be put on implementing schemes for highly reliable , fast , and scalable qubit readout ( initialization ) in each of the systems . finally , since the results in table [ tablecomparison ] are usually based on different experimental conditions , the third category consists of merging all required elements into one scalable device , without the need for excellent performance .
such a `` complete spin - qubit processor '' should combine individual single - qubit rotations about arbitrary axes , a controlled ( entangling ) two - qubit operation , initialization into a precisely known state , and single - shot readout of each qubit . while ref .
@xcite presents an important step toward this unit , prototypes of a complete spin - qubit processor could present the basis for continuous optimization . in summary ,
considering the impressive progress achieved within the past decade , one may be cautiously optimistic that a large - scale quantum computer can indeed be realized .
+ + + * summary points : * * the experimental situation has dramatically changed since 1998 , when the original proposal for quantum computation with quantum dots @xcite was published .
quantum dots are now precisely controlled down to the last spin .
single - qubit rotations around different axes , two - qubit operations , and various initialization and readout schemes have successfully been demonstrated ( table [ tablecomparison ] ) . *
it seems that one can soon overcome decoherence to the required extent , which is a big step toward the implementation of a quantum computer . * nuclear spins and spin - orbit interaction , on the one hand ,
present a source of decoherence and relaxation . on the other hand
, they can generate large overhauser fields and are useful for realizing quantum gates .
schemes exist to narrow the width of the nuclear field distribution . *
a variety of quantum dot systems is currently under investigation , including host materials with or without nuclear spins , operation in the conduction or valence band , and different geometries .
holes in ge / si - nanowire - based quantum dots are promising examples @xcite . *
electrically pulsed , pairwise control of the exchange interaction has been demonstrated in a triple quantum dot @xcite , i.e. , the scalability of exchange - based schemes for quantum computing @xcite has now been proven experimentally .
* long - distance spin - spin coupling via floating gates may be used to overcome architectural challenges of a large - scale quantum computer , and several two - dimensional architectures have been proposed @xcite .
alternatively , distant spins may be coupled via the photon field of a cavity @xcite . *
recent development toward quantum computation with quantum dots has been very positive , and one can be curious about the progress of the next few years .
+ * future issues : * * the surface code is a powerful protocol for fault - tolerant quantum computing @xcite .
can the requirements on the physical qubits be reduced further ?
* can longer lifetimes and more efficient qubit gates be reported ?
in particular , can schemes for fast and highly reliable single - shot readout and initialization be implemented in each of the discussed systems ? * schemes for exchange - controlled single - spin rotations ( about arbitrary axes ) have been proposed as an efficient alternative to rotations driven by oscillating fields @xcite .
can this be realized experimentally ? *
electrically controlled single - qubit and two - qubit operations were demonstrated in the setup of ref .
can a `` complete spin - qubit processor '' be reported soon ? +
the authors are not aware of any affiliations , memberships , funding , or financial holdings that might be perceived as affecting the objectivity of this review .
we thank s. gangadharaiah , f. maier , p. stano , d. stepanenko , m. trif , r. j. warburton , and j. r. wootton for helpful discussions and acknowledge support from the swiss nf , nccrs nanoscience and qsit , darpa , and iarpa . * * * * * + the error threshold exceeds 1% for quantum computation based on the surface code * * + electrical and pairwise control of the exchange interaction in a triple quantum dot proves scalability of exchange - based quantum computing schemes ( ref .
@xcite ) * * + experimental realization of a spin - qubit processor following ref .
@xcite , demonstrating two - qubit interactions and individual single - qubit rotations * * + `` direct rashba spin - orbit interaction '' and other attractive features are predicted for holes in ge / si nanowires * * * + + + + + + * further reading *
[ cols="<,^ , < " , ] | experimental and theoretical progress toward quantum computation with spins in quantum dots ( qds ) is reviewed , with particular focus on qds formed in gaas heterostructures , on nanowire - based qds , and on self - assembled qds .
we report on a remarkable evolution of the field where decoherence one of the main challenges for realizing quantum computers no longer seems to be the stumbling block it had originally been considered .
general concepts , relevant quantities , and basic requirements for spin - based quantum computing are explained ; opportunities and challenges of spin - orbit interaction and nuclear spins are reviewed .
we discuss recent achievements , present current theoretical proposals , and make several suggestions for further experiments . |
a common phenomenon of active galactic nuclei , which presumably harbor supermassive black holes with masses of @xmath0
@xmath1 ( rees 1984 ) , is the strong variability which can be observed in x - ray lightcurves .
these agn lightcurves seem to show featureless red noise , i.e. scale - free , divergent variability at low frequencies , often also described as flickering or @xmath2 fluctuation ( lawrence et al . 1987 ) .
the @xmath2 term describes the power law distribution of the spectral power with the function @xmath3 in the power spectrum , often denoted as @xmath2 behavior .
we present an alternative model to analyse the variability seen in the x - ray lightcurves of agn .
the standard method of analyzing time series in the frequency domain is discussed briefly in section 2 .
the alternative is known as a linear state space model ( lssm ) based on the theory of autoregressive processes ( scargle 1981 , honerkamp 1993 ) which usually can not be observed directly since the observational noise ( i.e. detectors , particle background ) overlays the process powering the agn .
a lssm fit applied to the time series data yields the dynamical parameters of the underlying stochastic process .
these parameters should be strongly correlated to the physical properties of the emission process .
the corresponding lssm power spectrum exhibits both the decrease of power at medium frequencies and a limitation of spectral power at low frequencies .
the detailed mathematical background of lssm and the fit procedure are described in section 3 and 4 .
finally we present first results using this technique with exosat data from the seyfert galaxy ngc 5506 in section 5 .
although measured astronomical data are time domain data , a commonly applied method works in the frequency domain by analyzing the power spectrum of the time series .
as the observational window function is convoluted to the true spectrum of the source , artefacts might be produced in the power spectrum , which make a proper interpretation more difficult ( papadakis and lawrence 1995 , priestley 1992 ) . in most cases ,
the power spectra are fit by a power law function with an offset described as @xmath5 , with values of @xmath6 ranging from 0 to 2 and a mean of about 1.5 ( lawrence and papadakis 1993 ) .
the value @xmath7 is often denoted as the ` observational noise floor ' which describes the random process comprising the observational errors whereas the ` red noise ' component is the signal of interest . in the case of long agn observations , however ,
a flattening at low frequencies occurs which can not be modelled by the @xmath4-model ( mchardy 1988 ) .
the @xmath4-model is an ad hoc description of the measured periodogram , without any direct physical motivation .
however , it is possible to generate time series with a @xmath4-spectrum using self - organized criticality models simulating the mass flow within an accretion disc of the agn ( mineshige et al .
such models produce a stationary time series that exhibits a @xmath4-power spectrum by limiting the timescales occurring in the simulated accretion process .
a @xmath4-model without limited timescales would be stationary only if the power law slope is smaller than unity ( samorodnitsky and taqqu 1994 ) .
the observed time series is composed by the superposition of single luminosity bursts .
the slope of the @xmath4-spectrum of data simulated in that way is about 1.8 , significantly higher than those measured from real data ( lawrence and papadakis 1993 ) .
if the inclination of the accretion disk is brought in as an additional model parameter the slope can be diminished , but not in a way that leads to convincing results ( abramowicz et al .
another point that contradicts this assumption is that there is no correlation between the spectral slope and the type of the seyfert galaxy ( green et al .
this correlation should be present since the seyfert type is believed to be caused by the inclination of the line of sight ( netzer 1990 ) .
the periodogram which is used to estimate the true source spectrum is difficult to interpret in the presence of non - equispaced sampling time series arising from real astronomical data ( deeter and boynton 1982 and references therein ) .
the estimation of the @xmath4-spectrum is hampered even in the absence of data gaps .
this is due to the finite extent of the observed time series .
therefore , the transfer function ( fourier transform of the sampling function ) is a sinc - function which will only recover the true spectrum if this is sufficiently flat ( deeter and boynton 1982 ; deeter 1984 ) . in the case of ` red noise '
spectra the sidebands of the transfer function will cause a spectal leakage to higher frequencies which will cause the spectra to appear less steep ( the spectral slope will be underestimated ) .
even periodograms of white noise time series deviate from a perfectly flat distribution of frequencies as the periodogram is a @xmath8-distibuted random variable with a standard deviation equal to the mean ( leahy et al . 1983 ) . thus the periodograms fluctuate and their variances are independent of the number of data points in the time series . due to the logarithmic frequency binning
, agn periodograms will always show this strong fluctuation due to the low number of periodogram points averaged in the lowest frequency bins ( see fig.1 ) .
fig.1 : a ) exosat me x - ray lightcurve of the quasar 3c273 ( jan .
1986 ) , b ) corresponding periodogram .
each dot represents the spectral power at its frequency , stepped with @xmath9 .
the periodogram is binned logarithmically ( squares indicates a single point within the frequency bin ) .
furthermore , additional modulations can be created in white noise periodograms if the time series consists of parts which slightly differ in their means and variances , respectively ( krolik 1992 ) . in the case of the exosat me x - ray lightcurves
this effect is due to the swapping of detectors as each detector has its own statistical characteristics which can not be totally suppressed ( grandi et al .
1992 ; tagliaferri et al .
fig.1a shows a typical x - ray lightcurve which mainly consists of uninterrupted 11 ksec observation blocks before detectors are swapped .
if the periodogram frequency corresponds to the observation block length , the calculated sum of fourier coefficients equals its expected white noise value of @xmath10 due to the constant mean and variance within the entire oscillation cycle . at other , mainly lower , frequencies the fourier sum yields non - white values due to temporal correlations caused by different means and variances of observation blocks located in the test frequency cycle .
these deviations from a flat spectrum will be very strong at frequencies which correspond to twice the observation block length .
the arrows in fig.1b clearly show this minimum feature at @xmath11hz and another shortage of power at @xmath12hz which corresponds to the long uninterrupted 72 ksec observation block starting at the second half of the exosat observation ( fig.1a ) .
consequently a model is required which operates in the time domain and avoids any misleading systematical effects occuring in power spectra .
in this section we briefly introduce the linear state space model ( lssm ) . for a detailed discussion , see honerkamp ( 1993 ) and hamilton ( 1995 ) .
the lssm is a generalization of the autoregressive ( ar ) model invented by yule ( 1927 ) to model the variability of wolf s sunspot numbers .
we follow wold s decomposition theorem ( wold 1938 ; priestley 1992 ; fuller 1996 ) which states that any discrete stationary process can be expressed as the sum of two processes uncorrelated with one another , one purely deterministic ( i.e. a process that can be forecasted exactly such as a strictly period oscillation ) and one purely indeterministic .
further , the indeterministic component , which is essentially the stochastic part , can be written as a linear combination of an innovation process , which is a sequence of uncorrelated random variables . a given discrete time series @xmath13 is considered as a sequence of correlated random variables .
the ar model expresses the temporal correlations of the time series in terms of a linear function of its past values plus a noise term and is closely related to the differential equation describing the dynamics of the system .
the fact that @xmath13 has a regression on its own past terms gives rise to the terminology ` autoregressive process ' ( for detailed discussions see scargle 1981 ; priestley 1992 ) .
a time series is thus a realization of the stochastic process or , more precisely , the observation of a realization of the process during a finite time interval .
the ar model expresses the temporal correlations in the process in terms of memory , in the sense that a filter ( @xmath14 ) remembers , for a while at least , the previous values @xmath15 .
thus the influence of a predecessor value decreases as time increases .
this fading memory is expressed in the exponential decay of the ar autocorrelation function ( see eq .
[ eq : rar ] ) .
the ar processes variable @xmath13 remembers its own behavior at previous times , expressed in a linear relationship in terms of @xmath16 plus @xmath17 which stands for an uncorrelated ( gaussian ) white noise process . @xmath18
the number of terms @xmath19 used for the regression of @xmath13 determine the order of the ar process , which is abbreviated to an ar[p ] process .
the parameter values @xmath14 have to be restricted for the process to be stationary ( honerkamp 1993 ) .
for a first order process this means @xmath20 .
depending on the order @xmath19 , the parameters @xmath14 of the process represents damped oscillators , pure relaxators or their superpositions .
for the first order process ar[1 ] the relaxation time @xmath21 of the system is determined from @xmath22 by : @xmath23 in the case of a damped oscillator for an ar[2 ] process the parameters , the period @xmath24 and the relaxation time @xmath25 respectively , are related by : @xmath26 for a given time series the parameters @xmath14 can be estimated e.g. by the durbin - levinson- or burg - algorithm ( honerkamp 1993 ) . by statistical testing
it is possible to infer whether a model is compatible with the data .
fig.2 : a ) exosat me x - ray lightcurve of ngc 5506 ( jan .
1986 ) , b ) hidden ar[1]-process , estimated with the lssm fit .
a first generalization of ar models are the autoregressive - moving - average ( arma ) models that include also past noise terms in the dynamics : @xmath27 both models , ar and arma processes , assume that the time series is observed without any oberservational noise . in presence of such noise the parameters @xmath14 will be underestimated and statistical tests will reject the model even if its order is specified correctly .
lssms generalize the ar and arma processes by explicitly modelling observational noise .
furthermore , lssms use the so called markov property , which means that the entire information relevant to the future or for the prediction is contained in the present state .
the variable @xmath13 that has to be estimated can not be observed directly since it is covered by observational noise @xmath28 . following the markov property it is possible to regressively
predict the values @xmath13 , though .
the measured observation variables @xmath29 may not necessarily agree with the system variables @xmath13 that provide the best description of the system dynamics .
thus a lssm is defined with two equations , the system or dynamical equation ( [ eq : sys ] ) and the observation equation ( [ eq : obs ] ) .
@xmath30 this definition is a multivariate description , which means that the ar[p ] process is given as a @xmath19-dimensional ar process of order one , with a matrix @xmath31 that determines the dynamics . by combining the different dimensional terms of the multivariate description the typical ar[p ] ( see eq .
[ eq : ar ] ) form can be derived easily . the observation @xmath29 is formulated as a linear combination of the random vectors @xmath32 and @xmath28 . the matrix @xmath33 maps the unobservable dynamics to the observation . the terms @xmath34 and @xmath28 represent the dynamical noise with covariance matrix @xmath35 and the observational noise with variance @xmath36 , respectively .
the estimation of the parameters in lssms is more complicated than for ar or arma processes .
there are two conceptually different procedures available to obtain the maximum likelihood parameters estimates .
both are iterative and start from some initial values that have to be specified .
the first procedure uses explicit numerical optimization to maximize the likelihood .
the other applies the so called expectation - maximization algorithm .
the latter procedure is slower but numerically more stable than the former and is described in detail by honerkamp ( 1993 )
. statistical evaluation of a fitted model is generally based on the prediction errors .
the prediction errors are obtained by a kalman filter which estimates the unobservable process @xmath37 ( hamilton 1995 ) .
such a linear filter allows us to arrive at the variables @xmath38 ( and its prediction errors ) , used to describe the system dynamics , starting from a given lssm and the given observations @xmath29 ( brockwell and davis 1991 , koen and lombard 1993 ) . multiplying the estimated process @xmath38 with the estimated @xmath33 yields an estimate @xmath39 of the observed time series @xmath29 .
a necessary condition that the model fits to the data is that the difference @xmath40 represents white noise , i.e. the time series of prediction errors should be uncorrelated .
this can for example be judged by a kolmogorov - smirnov test that tests for a flat spectrum of the prediction errors or by the portmanteau test using their autocorrelation function .
we have used the first method to quantify the goodness of fit of the tested lssms ( see table 1 ) . another criterion to judge fitted models is the decrease in the variance of prediction errors with increasing order of the fitted models . a knee in this function
gives evidence for the correct model order .
any further increase of the model order will not reduce the variance significantly . the so called akaike information criterion ( aic )
formulizes this procedure including the different number of parameters of the models ( hamilton 1995 ) .
any oscillators and relaxators which might occur in unnecessarily more complex lssms should be highly damped and can be neglected therefore . the last method to judge
a fitted model is to compare the spectrum that results from the fitted parameters with the periodogram of the sample time series .
the spectrum of a lssm is given by : @xmath41 the superscript @xmath24 denotes transposition .
spectra of ar or arma processes are special cases of equation ( [ eq : arsp ] ) . in the simplest case of an ar[1 ] process
modelled with a lssm , the corresponding spectrum is given by : @xmath42 } = \frac{q}{1 + a_1 ^ 2 - 2a_1\cos(\omega)}+r\end{aligned}\ ] ] this function provides both the flattening at low and the decrease of power at medium frequencies seen in periodograms ( e.g. see fig . 4 ) . in a first approach gaps in the observed lightcurve
were filled with white noise with the same mean and rms as the original time series in order to create a continuous time series . in a second run
these gaps were refilled with the predictions of the kalman filter plus a white noise realization with the original lightcurves variance .
generally , gaps in an observed time series can be handled by the lssm in a natural way avoiding the filling of gaps with poisson noise .
the key is again the kalman filter .
the kalman filter considers the fact that there are still decaying processes taking place even if the object is not observed . in each cycle of the iterative parameter estimation procedure @xmath32 is estimated based on an internal prediction , corrected by information obtained from the actual data @xmath29 . in case of gaps no information from @xmath29 is available and the internal prediction decays in its intrinsic manner until new information is given . in the case of the lightcurve of ngc 5506
the resulting parameters are consistent with those of the first approach due to the high duty cycle of the original time series .
as the x - ray lightcurves from exosat are the longest agn observations available , we have used the longest individual observation of about 230ks of the seyfert galaxy ngc 5506 for applying the lssm ( fig .
the data which have been extracted from the heasarc exosat me archive , are background subtracted and dead time corrected , with a 30 sec time resolution obtained over 18kev energy range .
the seyfert galaxy ngc 5506 holds a special place in agn variability studies , as it is both bright and one of the most variable agn .
the chosen lightcurve contains only few gaps providing a duty cyle of 92.4% .
the mean and rms of the lightcurve are 6.87 and 1.55 counts in 30s bins .
ccccc model & @xmath43 & periods & @xmath44 & ks test@xmath45 + lssm ar[p ] & & ( s ) & ( s ) & + 0 & 1 & - & - & 0.0% + 1 & 0.722 & - & 4799 & 93.5% + 2 & 0.701 & - & 26.1 & 66.8% + & & - & 5011 & + 3 & 0.510 & - & 10.6 & 88.2% + & & - & 18.9 & + & & - & 4798 & + 4 & 0.395 & 236.3 & 71.1 & 92.1% + & & - & 6.7 & + & & - & 4780 & + + + + + we applied lssms with different order ar processes .
an lssm using an ar[0 ] process corresponds to a pure white noise process without any temporal correlation and a flat spectrum .
the used kolmogorov - smirnov test rejects this model at any level of significance ( see table 1 ) . without loss of generality ,
@xmath46 is set to unity , the mean and variance are set to 0 and 1 , respectively .
we see that the x - ray lightcurve of ngc 5506 can be well modelled with a lssm ar[1 ] model , as the residuals between the estimated ar[1 ] process and the measured data are consistent with gaussian white noise .
fig.3 shows the distribution and the corresponding normal quantile plot of the fit residuals which both display the gaussian character of the observational noise .
the standard deviation of the distribution is 0.738 which is in good agreement to the estimated observational variance of 0.722 for the lssm ar[1 ] fit ( see table 1 ) .
furthermore , the lightcurve of the estimated ar[1 ] looks very similar to the temporal behavior of the hidden process ( fig.2 ) .
the corresponding dynamical parameter @xmath22 of the lssm ar[1 ] fit is 0.9938 which corresponds to a relaxation time of about 4799 s. fig.3 : a ) distribution and b ) normal quantile plot of the residuals of the lssm ar[1 ] fit to the exosat me ngc 5506 lightcurve ( the dotted lines in a ) indicate the mean and rms of the observational noise ) .
a normal quantile plot arranges the data in increasing order and plot each data value at a position that corresponds to its ideal position in a normal distribution .
if the data are normally distributed , all points should lie on a straight line .
the lssm ar[1 ] gives a good fit to the exosat ngc 5506 data as the variance of the prediction errors nearly remains constant from model order 1 to 2 and the residuals conforms to white noise .
the decrease in the variance for higher model orders might be due to correlations in the modelled noise , generated by the switching of the exosat detectors . since each detector has its own noise charateristics a regular swapping between background and source detectors would lead to an alternating observational noise level ( see section 2 ) .
the higher order lssm ar[p ] fits try to model the resulting correlations with additional but negligible relaxators and damped oscillators ( @xmath47 , @xmath48 ) .
we have used the durbin - levinson algorithm ( see section 3 ) to estimate the parameters of a competing simple ar[p ] model ( see table 2 ) . as expected for time series containing observational noise , the characteristic timescales
are underestimated by fitting a simple ar process and the statistical test rejects the ar[p ] model .
a test for white noise residuals fails , which means that there are still correlations present which can not be modelled with an ar[p ] procces .
we have performed ar[p ] fits for model orders up to 10 and we never found residuals consitent with white noise , indicating that there is no preferred model order .
all occuring relaxators and damped oscillators are insignificant due to their short relaxation timescales compared with the bintime of 30s . as the observational noise is not modelled explicitly in ar models , it is included accidentally in the inherent ar noise term .
thus , any correlation in the observed time series which can be detected in the lssm fits , is wiped out and the higher order ar fits only reveal fast decaying relaxators and oscillators .
ccccc model & @xmath49 & periods & @xmath44 & ks test@xmath45 + ar[p ] & & ( s ) & ( s ) & + 0 & 1 & - & - & 0.0% + 1 & 0.9235 & - & 23.3 & 0.5% + 2 & 0.8814 & - & 55.6 & 0.3% + & & - & 29.8 & + 3 & 0.8566 & - & 97.0 & 0.4% + & & 197.4 & 40.6 & + 4 & 0.8362 & - & 153.2 & 0.4% + & & - & 51.2 & + & & 127.7 & 55.1 & + + + + + one might expect that the resulting best fit lssm light curve ( fig .
2b ) might also be produced by just smoothing the original lightcurve .
this assumption is wrong as a smoothing filter would pass long timescales and suppress all short time variability patterns .
thus all information about the variations on short timescales would be lost ( brockwell and davis 1989 ) .
the kalman filter concedes not only the time series values @xmath13 but also its prediction errors .
these errors are much smaller than the errors of the observed lightcurve @xmath29 . in the case of the ngc 5506 observation ( fig.2 )
the estimation errors are about 0.18 counts / sec and the errors of @xmath29 are about 1.3 counts / sec , respectively .
both lightcurves in fig.2 are shown without error bars due to reasons of clarity .
we have used monte carlo simulations to determine the error of the dynamical parameter @xmath22 . using the distribution of the estimated parameters of 1000 simulated ar[1 ] time series with the best fit results
, we found @xmath50 .
as the dynamical parameter is close to unity the corresponding relaxation time error is high , with @xmath51s . to prove the quality of the lssm results we have fitted a lssm ar[1 ] spectrum to the periodogram data
this fit yields the dynamical parameter @xmath52 which is consistent with the lssm ar[1 ] fit in the time domain , but the corresponding error is much higher due to the lower statistical significance of frequency domain fits ( see section 2 ) .
the autocovariance function of the ar[1 ] process is given by : @xmath53}({\delta } ) = \frac { q } { 1 - a_1 ^ 2 } \
, e^{\log(a_1 ) \ , \delta } \label{eq : rar}\end{aligned}\ ] ] which is an exponentially decaying function for stationary ( @xmath54 ) time series , very similar to the temporal behavior of the autocorrelation function of a shot noise model ( papoulis 1991 ) : @xmath55 the variable @xmath56 denotes the density and @xmath21 is the lifetime of the shots .
this similarity means that an ar[1 ] process can also be modelled by a superposition of poisson distributed decaying shots ( papoulis 1991 ) .
the shot noise model , which has been used as an alternative to the @xmath4 model , appears to give a good fit to the power spectrum of ngc 5506 ( papadakis and lawrence 1995 , belloni and hasinger 1990 and references therein ) .
but instead of all the shots having the same lifetime , papadakis and lawrence ( 1995 ) used a distribution varying as @xmath57 between @xmath58 and @xmath59 .
they fixed @xmath59 arbitrarily at 12000s and found that @xmath58 is around 300s for ngc 5506 , much lower than the relaxation time of about 4800s found with the lssm fit . a possible explanation for this difference could be the distribution of lifetimes .
since the power law slope of the shot noise model is constantly @xmath60 at medium and high frequencies , this distribution is necessary to modify the slope and to maintain a good fit to the spectrum .
the advantage of a lssm is a variable slope at medium frequencies which depends on the dynamical parameter ( see fig . 4 ) .
fig.4 : periodogram of the exosat me x - ray lightcurve of ngc 5506 ( dots ) and the spectrum of the best fit lssm ar[1 ] model in the time domain ( line ) ( see fig.2a ) .
the spectra of the higher order lssm ar fits differ less than @xmath61 from the lssm ar[1 ] spectrum .
the dashed lines display the @xmath62 - spectra of the corresponding frequency domain fit .
the time domain fit yields @xmath63 errors which are more than 3 times smaller ( see text for details ) .
the shot noise model can be regarded as an approximation of an ar[1 ] model for values @xmath22 near unity .
the mean density of the poisson events @xmath56 then corresponds to the variance @xmath46 of the dynamical noise in the lssm system equation ( [ eq : sys ] ) .
thus @xmath46 could be used to quantify and compare the rate of the accretion shots occuring in agns .
we obtain a convincing fit to the observed x - ray lightcurve of an agn using a lssm ar[1 ] process as well in the time and in the frequency domain .
the explicit modelling of observational noise allows to estimate the covered ar[1 ] process , indicating that the stochastic process is dominated by a single relaxation timescale .
we show that the general ar[p ] model ( see eq.1 ) can be restricted to a simple ar[1 ] process which succeeds in describing the entire dynamics of the observed agn x - ray lightcurve .
it has been suggested by mchardy ( 1988 ) that the single shots , which are supposed to be superimposed to build the lightcurve , may arise from subregions of an overall larger chaotic region which are temporarily lit up , perhaps by shocks . since one would expect a non uniform electron density throughout this region ( probably decreasing with distance from the central engine ) , the resulting difference in cooling timescales yields the different decay timescales ( green et al .
as the lssm predicts that the stochastic process is dominated by a single relaxator , we presume the existence of a single cooling timescale or a uniform electron density in the emission region following the shot noise model ( see sutherland et al . 1978 ) .
the assumption of an exponentially decaying shot seems to be reasonable as time - dependent comptonisation models lead to such a pulse profile .
the scenario for a thermal comptonisation model ( payne 1980 , liang and nolan 1983 ) starts with uv photons which arise as the accretion inflows inhomogeneities , each producing a single flare when gravitational energy is set free as radiation .
the impulsive emission of the poisson distributed delta peaks in a cloud of hot electrons triggers x - ray flares with a specific pulse profile depending on the seed photon energy , the density , and the temperature of the electrons .
this impulsive emission is delayed and broadened in time and spectrally hardened due to repeated compton scattering .
some approximate analytic solutions of this process show that the temporal evolution of the generated x - ray pulse can be described by a nearly exponentially decaying function ( miyamoto and kitamoto 1989 ) .
the only difference to the ` shots ' used above is the ( more realistic ) non - zero rise time . using this model
it should be possible to associate the estimated relaxator timescale @xmath64}$ ] with the physical properties of the comptonisation process .
the presented lssm can also be used to analyse x - ray variability of galactic x - ray sources . as both ,
relaxators and ( damped ) oscillators can be estimated , it is possible to use the algorithm to search for periodocities and qpo phenomena in the lightcurves of x - ray binaries ( see robinson and nather 1979 , lewin et al .
1988 , van der klis 1989 ) .
_ acknowledgements : _ we would like to thank j.d .
scargle , r. staubert , m. maisack , j. wilms and k. pottschmidt for helpful discussions and c. gantert for writing the code of the lssm program .
furthermore , we thank the anonymous referee for constructive comments .
abramowicz a.r .
, chen x. , kato , s. et al . , 1995 ,
apjl 438 , l37 begelman m.c . , de kool m. , 1991 , in variability in active galactic nuclei , ed .
h.r.miller & p.j.wiita ( cambridge : cambridge univ.press ) , 198 belloni t. , hasinger g. , 1990 , a&a 227 , l33 brockwell p.j . ,
davis r.a . , 1991 ,
time series : theory and methods , springer verlag , 2nd.ed .
deeter j.e . , 1984 ,
apj 281 , 482 - 491 deeter j.e . ,
boyton p.e . , 1982 ,
apj 261 , 337 - 350 fuller w.a . , 1996 ,
introduction to statistical time series , new york , john wiley , 2nd.ed .
grandi p. , tagliaferri g. , giommi p. et al . , 1992 , apj suppl .
ser . , 82 , 93 green a.r .
, mchardy i.m . ,
lehto h.j .
, 1993 , mnras 265 , 664 - 680 hamilton j.d . , 1995 , time series analysis , princeton university press honerkamp j. , 1993 , stochastic dynamical systems , vch publ .
new york , weinheim koen c. , lombard f. , 1993 , mnras 263 , 287 krolik j.h . , 1992 , statistical challenges in modern astronomy , e.feigelson and g.j.babu , eds .
, springer verlag new york , 349 lawrence a. , watson m.g .
, pounds k.a .
et al . , 1987 , nature 325 , 694 lawrence a. , papadakis p. , 1993
, apj suppl .
414 , 85 leahy d.a .
, darbro w. , elsner r.f . , weisskopf m.c , sutherland p.g . , 1983 ,
apj 266 , 160 - 170 lehto h.j . , 1989 , in proc .
23rd eslab symp . on two topics in x - ray astronomy , vol.1 , ed .
j.hunt & b.battrick ( esa sp-296 ) , noordwijk , 499 lewin w.h.g . ,
van paradijs j. , van der klis m. , 1988 , space sci .
review 46 , 273 liang e.p . ,
nolan p.l . , 1983 , space sci
. review 38 , 353 mchardy i. , czerny b. , 1987 , nature 325 , 696 mineshige s. , ouchi n.b . , nishimori h. et al . , 1994 , pasj 46 , 97 miyamoto s. , kitamoto s. , 1989 , nature 342 , 773 netzer h. , 1990 , saas - fee advanced course 20 on active galactic nuclei , eds .
blandford , h.netzer , l.woltjer , 57 - 160 papadakis i.e. , lawrence a. , 1995 , mnras 272 , 161 papoulis a.p . ,
1991 , probability , random variables and stochastic processes , new york , mcgraw - hill , 3rd.ed .
payne d.g .
, 1980 , apj 232 , 951 priestley m.b . , 1992 , spectral analysis and time series , san diego , academic press rees m.j . , 1984 , ann.rev.astron.astrophys .
22 , 471 robinson e.l .
, nather r.e . , 1979 , apj suppl .
39 , 461 samorodnitsky g. , taqqu m.s .
, 1994 , stable non - gaussian random processes , new york , chapman and hall scargle j.d . , 1981 , apj suppl .
45 , 1 sutherland p.g .
, weisskopf m.c . ,
kahn s.m . , 1978 , apj 219 , 1029 tagliaferri g. , bao g. , israel g.l .
1996 , apj , accepted for publication van der klis m. , 1989 , ann.rev.astron.astrophys . 27 , 517 wold h.o.a . , 1938 , a study in the analysis of stationary time series , uppsala , almqvist and wiksell , 2nd.ed
yule g. , 1927 , phil .
a 226 , 267 + | in recent years , autoregressive models have had a profound impact on the description of astronomical time series as the observation of a stochastic process .
these methods have advantages compared with common fourier techniques concerning their inherent stationarity and physical background .
however , if autoregressive models are used , it has to be taken into account that real data always contain observational noise often obscuring the intrinsic time series of the object .
we apply the technique of a linear state space model which explicitly models the noise of astronomical data and allows to estimate the hidden autoregressive process . as an example , we have analysed the x - ray flux variability of the active galaxy ngc 5506 observed with exosat . |
in cellular dynamical mean - field theory ( c - dmft ) a cluster is embedded in a bath of non - interacting electrons that simulates the effect of the rest of the infinite lattice by injecting and removing electrons on the cluster with the appropriate single - particle propagator .
the bath is determined self - consistently by requiring that the self - energy of the infinite system and that of the cluster be the same . to break the symmetry , frequency independent source fields
are allowed on bath sites only . more details can be found in @xcite .
all the calculations are done with exact diagonalization at zero temperature @xcite . for the case of a @xmath45 plaquette , which we shall consider throughout this work ,
the nambu spinor is defined by @xmath70 , and the greek letters @xmath71 label the degrees of freedom within the cluster .
we compute the cluster propagator @xmath72 by solving the cluster impurity hamiltonian that will be described shorty .
given the @xmath73 on the cluster that results from the presence of the bath , we extract the cluster self energy from @xmath74 . here , @xmath75{cc}\hat{g}_{\uparrow}\left
( \tau,\tau^{\prime}\right ) & \hat{f}\left ( \tau,\tau^{\prime}\right ) \\ \hat{f}^{\dagger}(\tau,\tau^{\prime } ) & -{\hat{g}}_{\downarrow}\left ( \tau^{\prime},\tau\right ) \end{array } \right ) \label{nambugreen}\ ] ] is an @xmath76 matrix , @xmath77 and @xmath78 are the imaginary - time ordered normal and anomalous green functions respectively . using the self - consistency condition , @xmath79 ^{-1}+\hat{\sigma}_{c}(i\omega_{n } ) \label{selfcon}\ ] ] with @xmath80 ^{-1 } , \label{lattice_g}\ ] ] we recompute the weiss field @xmath81 obtain the corresponding bath parameters by minimizing a distance function described below , and iterate till convergence . here
@xmath82 is the fourier transform of the superlattice hopping matrix with appropriate sign flip between propagators for up and down spin and the integral over @xmath83 is performed over the reduced brillouin zone of the superlattice .
a @xmath45 plaquette is embedded in a bath of non - interaction electrons . to solve the cluster impurity problem
, we express it in the form of a hamiltonian @xmath84 with a discrete number of bath orbitals coupled to the cluster and use the exact diagonalization technique ( lanczos method ) @xcite : @xmath85 here @xmath71 label the sites in the cluster and @xmath86 represents the hopping and the chemical potential within the cluster .
the energy levels in the bath are grouped into replicas of the cluster ( @xmath87 ) ( two replicas in the present case ) with the labels @xmath88 and @xmath89 such that we have @xmath90 bath energy levels @xmath91 coupled to the cluster via the bath - cluster hybridization matrix @xmath92 . using lattice symmetries we take @xmath93 and @xmath94 .
the quantity @xmath95 represents the amplitude of superconducting correlations in the bath .
no static mean - field order parameter acts directly on the cluster sites .
the parameters @xmath96 , @xmath97 and @xmath95 are determined by imposing the self - consistency condition in eq .
[ selfcon ] using a conjugate gradient minimization algorithm with a distance function @xmath98 that emphasizes the lowest frequencies of the weiss field by imposing a sharp cutoff at @xmath99 .
( energies are given in units of hopping @xmath100 and we take @xmath101 . ) the distance function in eq.([distance ] ) is computed on the imaginary frequency axis ( effective inverse temperature , @xmath102 ) since the weiss field @xmath103 is a smooth function on that axis . with the bond superconducting order parameter defined as @xmath104
we consider @xmath105-wave singlet pairing ( @xmath106 ) .
the average is taken in the ground state of the cluster .
all quantities depending on wave vector , including self - energy , are obtained from the green function periodization scheme .
the finite size of the bath in the exact - diagonalization technique is an additional approximation to the cdmft scheme .
the accuracy of this approximation can be verified by comparing the cdmft solution for the one - band hubbard model with the solution from the bethe ansatz @xcite .
we have also used this comparison in one dimension as a guideline to fix the choice of parameters in the distance function ( @xmath99 and @xmath107 ) .
these results in one dimension also compare well with those obtained using the hirsch - fye quantum monte carlo algorithm as an impurity solver where the bath is not truncated @xcite .
further , using finite - size scaling for these low ( but finite ) temperature calculations @xcite , it was shown that , at intermediate to strong coupling , a @xmath45 cluster in a bath accounts for more than @xmath108 of the correlation effect of the infinite size cluster in the single - particle spectrum . because of the finite size of the bath , one also needs to use a finite linewidht broadening @xmath109 when plotting the figures .
we can also perform an internal consistency check on the effect of the finite bath on the accuracy of the calculation . with an infinite bath
, convergence insures that the density inside the cluster is identical to the density computed from the lattice green function . in practice , we find that there can be a difference of @xmath110 between the density estimated from the lattice and that estimated from the cluster .
we display results as a function of cluster density since benchmarks with the one - dimensional hubbard model show that , with a finite bath and the procedure described above , one can reproduce quite accurately bethe ansatz results for @xmath111 when the cluster density is used .
nevertheless , we should adopt a conservative attitude and keep in mind the error estimate mentioned above .
the off - diagonal self - energy @xmath10 is extracted from the periodized nambu green function and plotted with @xmath112 the maxima of the plot are marked with black dots .
the imaginary part of the local spin susceptibility @xmath60 is calculated on the cluster and plotted with the same value of @xmath113 the dots indicate the maxima in the limit @xmath114 the various densities evaluated on the cluster are plotted for the following values of @xmath115 starting from the normal state @xmath116 @xmath117 @xmath118 @xmath119 @xmath120 @xmath121 @xmath122 @xmath123 @xmath124 @xmath125 @xmath126 @xmath127 @xmath128 @xmath129 @xmath130 @xmath131 @xmath132 @xmath133 @xmath134 @xmath135 fig .
[ fig_delta_vs_delta ] illustrates how the single - particle gap in the superconducting state @xmath136 ( not necessarily the superconducting gap ) and the shift @xmath105 between the position of the peaks in @xmath10 and @xmath11 change with doping .
( open circles and blue line ) between the first peak in the @xmath11 and the first peak in @xmath10 at the antinodal point . also shown as a function of doping , the single particle gap @xmath136 ( triangles and solid red line ) measured from the single - particle density of states .
on the left panel , @xmath13 @xmath137 . on the right panel @xmath27 and the band parameters are those appropriate for yba@xmath28cu@xmath29o@xmath138 , width=302 ] in conventional migdal - eliashberg theory ,
the real part of the self energy @xmath139 times the quasiparticle renormalization factor is essentially the gap function .
we find that this function , illustrated in fig [ fig_re_sigma_doping ] , increases as one approaches half - filling , consistent with the increase in the single particle gap found earlier @xcite and illustrated in fig .
[ fig_delta_vs_delta ] .
@xmath139 has weak frequency dependence near zero frequency only over a range of order @xmath140 for @xmath141 as can be seen in fig .
[ fig_resigma_u ] .
if there were a static piece to the gap , @xmath139 would have a frequency independent component at frequencies larger than @xmath55 , at least until frequencies of order @xmath142 we find that this is not the case . for the @xmath0 model
@xcite one finds a small instantaneous contribution to @xmath139 , thus making connection with mean - field theories .
we show in the main text how mean - field theories can also be seen as approximations to the present approach , even though we do not find an instantaneous contribution to pairing . as a function of frequency @xmath61 at the antinodal point for fixed doping @xmath145 and different values of @xmath146 and @xmath147 @xmath148 represented respectively by solid bleue line , short - dashed red line and long - dashed green line .
the nearly flat part near @xmath149 decreases with @xmath56 the range of frequencies where @xmath31 is positive also decreases as @xmath3 increases or @xmath55 decreases .
, width=302 ] for @xmath13 @xmath144 expressed in cluster momenta for underdoping on the top panel ( @xmath150 ) and for overdoping on the bottom panel ( @xmath151)@xmath53 even in the latter case , a sizable @xmath65 component is left .
recall that the momenta refer to averages over a quarter of the brillouin zone.,width=302 ] the lehman representation for the nambu green function allows us to find the following result for the @xmath8 value of @xmath36@xmath152 with @xmath153 the heaviside step function .
excited states @xmath154 that contribute have an energy less than @xmath61 above the ground state @xmath155 for bcs s - wave theory , @xmath36 can be computed analytically .
one obtains , using @xmath156 with @xmath157 @xmath158 .
\label{integrated_off_diagonal_sw}\ ] ] the results for the d - wave case in the main text were obtained by numerical integration and a sharp cutoff . in eliashbergh
theory , the function @xmath159 is@xmath160 d\omega ^{\prime}\ ] ] where @xmath161 is the single - particle density of states at the fermi level , the square root is in the upper half - plane and @xmath162 with @xmath163 the diagonal components of the self - energy in nambu space , @xmath164 and @xmath165 .
the phase is chosen such that there is no contribution from the second pauli matrix in nambu space @xcite .
the anomalous green function @xmath166 entering the calculation of @xmath36 in cdmft was obtained by fourier transforming the anomalous lattice green function calculated with the band lanczos approach .
we used three different values @xmath167 and @xmath168 for the small imaginary part that must be added to the real frequency to obtain the retarded @xmath169 the final result for @xmath36 is the extrapolation to @xmath114 this is done to smooth the function while preserving as much as possible the asymptotic large frequency value .
it differs by only a few percent from the value of the order parameter calculated on the cluster .
[ [ effect - of - the - self - energy - in - decreasing - the - pairing - tendency - as - one - approaches - half - filling ] ] effect of the self - energy in decreasing the pairing tendency as one approaches half - filling ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ fig .
[ pairing_suscep ] shows that in the normal state , the pairing susceptibility calculated without vertex corrections decreases as one approaches haff - filling .
this is an illustration of the detrimental effect of the pseudogap .
the self - energy in the dressed green functions entering the calculation leads to a decrease in the number of states that can pair near the fermi level .
the cutoff frequency enters very clearly in the integrated off - diagonal spectral weight eq .
( [ integrated_off_diagonal_sw ] ) .
since one expects that the attractive ( instead of repulsive ) hubbard model should behave more like the bcs model , we checked that @xmath36 calculated with c - dmft for that model does have the structure of the bcs result for @xmath170 wave . in other words , it vanishes below the gap , and increases monotonically until a sharp cutoff frequency that depends somewhat on @xmath3 but is of the order of the bandwidth , as expected from the mean - field solution .
there is some structure in the frequency dependence that is probably caused in part by the finiteness of the bath used in the calculation , but does not change the overall trend . | confirmation of the phononic origin of cooper pair formation in superconductors came with the demonstration that the interaction was retarded and that the corresponding energy scales were associated with phonons . using cellular dynamical mean - field theory for the two - dimensional hubbard model , we identify such retardation effects in d - wave pairing and associate the corresponding energy scales with short - range spin fluctuations .
we find which frequencies are relevant for pairing as a function of interaction strength and doping and show that the disappearance of superconductivity on the overdoped side coincides with the disappearance of the low energy feature in the antiferromagnetic fluctuations , as observed in neutron scattering experiments . in ordinary superconductors ,
the origin of attraction between electrons , the pairing glue , manifests itself in observable quantities .
indeed , the characteristic frequencies of phonons appear directly in the frequency dependence of the gap function , which in turn enters observables such as the single - particle density of states or the infrared conductivity .
migdal - eliashberg theory @xcite has been extremely successful to extract from these experiments the spectral function of the phonons that provide the glue .
high - temperature superconductors , heavy - fermion and layered organic superconductors all have phase diagrams where non s - wave superconducting order parameters lie in close proximity to antiferromagnetic phases . in the case of high - temperature superconductors ,
much effort has been devoted to find out whether antiferromagnetic fluctuations could be the pairing glue @xcite .
even though its assumptions are not generally valid in that case , eliashberg theory has been used to extract the amplitude and frequency dependence of a spectral function that is found to be similar to that for antiferromagnetic fluctuations directly measured by neutron scattering @xcite .
but understanding the origin of pairing in high - temperature superconductors requires an approach that does _ not _ rely on the assumptions entering eliashberg theory and that takes into account mott insulating behavior .
this seems to rule out theories that are based purely on early weak - coupling ideas of boson exchange @xcite .
in fact , anderson @xcite has argued that the appropriate starting point consistent with mott physics is the strong - coupling version of the hubbard model , or the @xmath0 model , and that in this case interactions are instantaneous , as suggested by mean - field factorization @xcite . in this paper , we show , for the hubbard model , that spectral features of the imaginary part of the anomalous ( off - diagonal ) self - energy do correspond to those of the spectral function for short - range spin fluctuations and that the energy scales relevant for pairing are also those of spin fluctuations .
the hubbard model hamiltonian is given by @xmath1 where @xmath2 and @xmath3 correspond to the hopping matrix and the onsite screened coulomb repulsion respectively with @xmath4 the destruction ( creation ) operators for an electron at site @xmath5 with spin @xmath6 and @xmath7 the number operator .
the theoretical method that has been most successful to date to treat the mott transition starting from the one - band hubbard model is dynamical mean - field theory ( dmft ) @xcite .
cluster generalizations of dmft @xcite are necessary to study problems in two dimensions where correlations beyond single site must be taken into account to study , for example , d - wave superconductivity .
they lead to phase diagrams that have the same features as those observed experimentally for both electron- and hole- doped high - temperature superconductors @xcite and for organic conductors .
in addition , observable quantities such as the density of states , @xcite the arpes spectrum @xcite and the optical conductivity @xcite have the experimentally observed behavior .
the method that we use , c - dmft with exact diagonalization at @xmath8 , is described in refs .
@xcite and in ref .
@xcite .
we stress that it does not involve _ any _ eliashberg - like approximation . in c
- dmft , antiferromagnetism and d - wave superconductivity coexist over part of the phase diagram .
this is seen in stoichiometric cuprates with intrinsically doped planes @xcite and in a few other cases , @xcite but does not appear to be a completely generic property of the phase diagram . our work is thus restricted to showing that spin fluctuations are relevant for high - temperature superconductivity all the way to the overdoped regime , leaving open the possibility that additional types of fluctuations may either contribute to or hinder superconductivity in the underdoped phase . at the fermi wave vector
nearest to the antinodal point , for various dopings . in ( b ) imaginary part of the local spin susceptibility @xmath9 .
black dots in ( a ) and ( b ) identify peaks .
the position of the peaks of @xmath10 in ( a ) are reported as magenta dots in ( b ) at the same height as the corresponding @xmath11 to illustrate the correspondence between the main peaks of the two functions .
the frequency splitting between the peaks decreases with doping , like the single particle gap .
the red curves are for the normal state .
the lower frequency peak present in the superconducting state disappears and the next peak moves to higher frequency with doping . in all the figures
lorentzian broadening is @xmath12 @xmath13 @xmath14 @xmath15 , @xmath16 , for la@xmath17sr@xmath18cuo@xmath19,width=302 ] the correspondence between the imaginary part of the anomalous self - energy , @xmath20 and the imaginary part of the local spin susceptibility , @xmath11 , is seen in fig .
[ fig_imsigma_im_chi ] .
we take band parameters appropriate for la@xmath17sr@xmath18cuo@xmath21 namely @xmath22 for nearest - neighbor and @xmath23 for next - nearest - neighbor hopping .
cdmft with @xmath24 then leads to superconductivity in the doping range observed experimentally @xcite .
the anomalous self - energy @xmath25 is defined as minus the off - diagonal part of the inverse green function in nambu space .
numerical results are presented in energy units where @xmath26 for all different dopings , the positions of the first two peaks in the spin fluctuations ( black dots on middle panel ) are just shifted down with respect to the corresponding peaks in @xmath10 ( black dots on top panel ) . in eliashberg theory for the electron - phonon interaction , the first two peaks in the phonon density of states are shifted down with respect to those in @xmath10 by the bcs gap @xcite .
similarly , the down shift of peaks in @xmath11 seen in fig .
[ fig_imsigma_im_chi]b increases as we underdope , like the single - particle gap . for @xmath27 and realistic band structure for yba@xmath28cu@xmath29o@xmath30
the shift is very weakly doping dependent @xcite . in migdal - eliashberg theory ,
the real part of the self energy @xmath31 times the quasiparticle renormalization factor is the gap function .
we find that this function , has no static ( frequency independent ) contribution , @xcite contrary to what was found in the @xmath0 model .
@xcite .
to identify the energy scales relevant for the pairs , we introduce the function @xmath32 @xmath33 is the retarded gorkov function defined in imaginary time by @xmath34 with @xmath5 and @xmath35 nearest - neighbors .
the infinite frequency limit of @xmath36 is equal to @xmath37 which in turn is proportional to the @xmath8 d - wave order parameter ( it changes sign under @xmath38 rotation ) .
it was shown in ref .
@xcite that @xmath39 scales like @xmath40 . for all these reasons ,
@xmath41 is useful to estimate the frequencies relevant for binding .
its meaning is illustrated by the d - wave bcs result in fig .
[ fig_ig_wide]a ) .
the function @xmath36 rises monotonically until it reaches the sharp bcs cutoff frequency @xmath42 above which no binding occurs .
@xmath36 extracted from the eliashberg calculation @xcite for lead is also displayed in fig .
[ fig_ig_wide]a ) .
the maximum is reached at a frequency just above the largest phonon frequency . for a d - wave bcs superconductor with a cutoff at @xmath43
the magenta line is obtained from eliashberg theory for pb in ref . .
frequencies in that case are measured in units of the transverse phonon frequency .
the two glitches before the maximum correspond to the transverse and longitudinal peaks in the phonon density of states .
the scale of the vertical axis is arbitrary .
( b ) @xmath36 calculated for various dopings .
the horizontal lines for the asymptotes mark the value of the order parameter .
( c ) the three independent fourier components of @xmath44on a @xmath45 plaquette for an underdoped case.the @xmath46 component dominates at low frequencies.,width=302 ] @xmath36 is plotted in fig .
[ fig_ig_wide]b for underdoping @xmath47 optimal doping @xmath48 and overdoping @xmath49 .
the asymptotic large frequency value of @xmath41 indicated by horizontal lines gives the order parameter that , as a function of doping , has the dome shape dependence @xcite .
the functions @xmath36 cross their respective asymptotic values at progressively lower frequencies as doping increases .
the spin fluctuations that dominate at the lower frequencies come from wave vectors around @xmath50 as illustrated in fig .
[ fig_ig_wide]c for an underdoped case .
the maximum of @xmath36 is more pronounced in the underdoped regime .
the form of @xmath51 in the overdoped regime is closer to the bcs limit with just a sharp cutoff .
our calculations are less precise at high frequencies , but nevertheless they suggest that , in all cases , @xmath36 undershoots very slightly its asymptotic value and then recovers at frequencies that are of order @xmath52 where the upper hubbard band opens new scattering channels @xcite@xmath53 this has no analog in ordinary superconductors . in fig .
[ fig_ig_low ] we focus on the low - frequency behavior . on the top panel
, @xmath36 crosses its asymptotic value for the first time near its maximum .
this crossing point shown by vertical lines follows the first peak in the corresponding @xmath11 in the bottom panel . by studying the cases
@xmath54 we have verified that these features scale with @xmath55 . clearly , if we wished to design an approximate mean - field theory @xcite for this problem that would play a role analogous bsc theory as an approximation of the migdal - elisahberg theory , we would use a frequency cutoff of order @xmath56 let us now discuss how the properties of the spin fluctuations @xmath57 correlate with those of the d - wave superconducting state for la@xmath17sr@xmath18cuo@xmath19 in fig .
[ fig_imsigma_im_chi]b , one sees that in the underdoped regime the low frequency peak is the most prominent feature .
optimal doping @xmath58 is reached when the intensity of the low frequency peak becomes comparable to the next one at higher frequency .
more importantly one sees that , around doping @xmath59 , superconductivity disappears with the low frequency peak in @xmath60 below @xmath61 about @xmath62 .
that low frequency peak is the one involved in the dynamics of the pairs as shown by the intersection of the @xmath36 function with its asymptotic value in fig .
[ fig_ig_wide]b .
the leftover peak in the normal state , indicated in red in fig .
[ fig_imsigma_im_chi]b , moves to higher frequencies as we dope .
these features are those found in neutron scattering experiments .
@xcite the large tails and monotonic decrease of the weight of the low frequency peak in @xmath11 as we dope are consistent with the glue function extracted from recent optical conductivity experiments .
@xcite the position of the low frequency peak near @xmath63 at optimal doping is consistent with the experimental value @xcite if we take @xmath64 one should recall that fig .
[ fig_imsigma_im_chi]b for the local spin spectral weight @xmath11 gives information integrated in wave vector so the properties of the neutron resonance located at @xmath46 have to be found by a different approach . in a recent calculation with a related cluster method ,
@xcite it has been found that the peak located at @xmath65 in the infinite lattice decreases with frequency in the underdoped regime . given the small weight of this neutron resonance ,
this does not contradict the fact that @xmath66 whether local or averaged over one - quarter of the brillouin zone near @xmath46 , has the opposite doping dependence @xcite the magnetic resonance itself has small weight @xcite .
we stress that , despite the similarities , the results obtained in this paper are not identical to those that are obtained in ordinary superconductors .
in particular , in the underdoped regime the spin fluctuations are strongly pair - breaking .
the pair - breaking effect of the pseudogap can be seen from the fact that in the normal state , the pairing susceptibility decreases as one approaches half - filling when vertex corrections are neglected @xcite .
following the suggestion of ref .
@xcite , we also checked whether the pseudogap is pair breaking by computing the ratio @xmath67 at the antinodal fermi surface crossing as a function of matsubara frequency .
the ratio is far from constant , in agreement with the existence of strong pair - breaking effects in the pseudogap . from a diagrammatic point of view , spin fluctuations both scatter electrons ( self - energy ) , decreasing the density of states at the fermi surface , and provide the glue ( vertex ) .
this can lead to the dome shape of the transition temperature as a function of doping @xcite . in the underdoped regime , the pair - breaking effect wins over the glue provided by the vertex , whereas in the overdoped regime the vertex dominates . in conclusion
, we have found that the imaginary part of the anomalous self - energy has a structure that is correlated with the spectrum of spin fluctuations .
this correlation is similar to the one found with the phonon spectrum in the migdal - eliashberg theory of ordinary superconductors .
this suggests the importance of spin - one excitations in pair formation .
our approach also allows for mutual feedback between spin fluctuations and pairing @xcite .
the frequencies most relevant for the pair dynamics scale with the heisenberg exchange @xmath55 .
superconductivity disappears for sufficient overdoping when the first peak in @xmath68 below frequencies about @xmath69 becomes negligible .
there are , however , major differences with ordinary superconductors . even though the anomalous self - energy increases as we approach half - filling
, the order parameter decreases because of large self - energy effects in the normal part of the propagator .
these come from mott physics at half - filling .
the magnetic fluctuations that we find have the same doping and energy dependence as that found in optical , @xcite tunneling @xcite and neutron experiments @xcite .
this work leaves open the possibility that in the underdoped regime there exists other instabilities that compete with antiferromagnetism and d - wave superconductivity .
we are grateful to d.j .
scalapino for insights at the origin of this work , to c. bourbonnais , j. carbotte , m. civelli , t. maier for discussions , and to m. greven and g. kotliar and l. taillefer for careful reading and comments on the manuscript .
this work was supported by nserc ( canada ) , cfi ( canada ) , cifar , and the tier i canada research chair program ( a .- m.s.t . ) .
computations were carried out on clusters of the rseau qubcois de calcul de haute performance ( rqchp ) and on the elix cluster at universit de sherbrooke .
a .- m.s.t .
thanks the aspen center for physics .
99 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 and , * * , ( ) .
, * * , ( ) . and , * * , ( ) . , , and , * * , ( ) , issn . , , and , _ _ ( , ) . , , and , * * , ( ) . , , and , * * , ( ) . , et al . , *
* , ( ) . , et al . , * * , ( ) . , et al . ,
* * , ( ) . , et al . , ( ) .
beal - monod:1986 , et al . , * * , ( ) . , et al . , *
* , ( ) . , et al . , *
* , ( ) . ,
* * , ( ) . and , * * , ( ) .
georges:1996 , et al . , * * , ( ) . , et al . , *
* , ( ) . , et al . , * * , ( ) . , * * , ( ) .
maier:2005 , et al . , * * , ( ) . , et al . , *
* , ( ) . , et al . ,
* * , ( ) . and , * * , ( ) .
macridin:2005 , et al . , * * , ( ) .
supplementretardation:2009 , et al . ,
( ) . , et al . , ( ) . , et al . , *
* , ( ) . , et al .
, et al . , * * , ( ) .
scalapino:1966 , et al . , * * , ( ) . , et al . ,
* * , ( ) . , et al . , *
* , ( ) . , et al . , *
* , ( ) .
brehm:2008 , et al . ,
( ) . , et al . , * * , ( ) . , et al . , *
* , ( ) . , et al . , *
* , ( ) . , et al . , * * , ( ) .
schiro:2008 , et al . , * * , ( ) . , et al . , *
* , ( ) .
c. bourbonnais and a. sedeki ( unpublished ) . , et al . , *
* , ( ) . ,
* * , ( ) .
, , , , and , * * , ( ) . , , and , * * , ( ) . |
long - duration gamma - ray bursts ( lgrbs ) , associated with the core - collapse deaths of massive stars , are among the most energetic events observed in our universe . as a result , they are widely cited as powerful and potentially unbiased tracers of the star formation and metallicity history of the universe out to @xmath0 ( e.g. bloom et al .
2002 , fynbo et al . 2007 , chary et al .
2007 , savaglio et al . 2009 ) .
however , in recent years potential biases in the star - forming galaxy population sampled by lgrbs have become a matter of debate .
recent work on a small number of nearby lgrbs suggested a connection between lgrbs and low - metallicity environments ( e.g. fruchter et al .
2006 , wainwright et al .
nearby host galaxies appeared to fall below the luminosity - metallicity and mass - metallicity relations for star - forming galaxies out to @xmath1 ( e.g. modjaz et al .
2008 , kocevski et al .
2009 , levesque et al .
2010a , b ) .
these results could potentially introduce key biases that would impact the use of lgrbs as cosmic probes .
a metallicity bias , or some correlation between metallicity and lgrb host or explosive properties , is indeed expected under the most commonly - cited progenitor model for lgrbs , the collapsar model ( woosley 1993 ) . under the classical assumptions of stellar evolutionary theory ,
the progenitor is a single rapidly - rotating massive star which maintains a high enough angular momentum over its lifetime to generate an lgrb from core - collapse to an accreting black hole .
in addition , lgrbs have been observationally associated with broad - lined type ic supernovae ( e.g. galama et al .
1998 , stanek et al . 2003 ,
malesani et al .
2004 , modjaz et al .
2006 , starling et al .
2011 ) , requiring the progenitors to have shed mass , and therefore angular momentum , as a means of stripping away their outer h and he shells .
mass loss rates for these evolved massive stars are dependent on stellar winds ( vink & de koter 2005 ) , which in turn are dependent on the stars metallicity ( kudritzki 2002 , vink et al .
2001 ) . for young massive stars ,
the metallicities of their natal environments can be adopted as the metallicities of the stars themselves .
it therefore stands to reason that the wind - driven mass loss rates in high - metallicity environments would rob the stars of too much angular momentum , preventing them from rotating rapidly enough to produce a lgrb and suggesting that lgrbs should either be restricted to low - metallicity environments ( e.g. hirschi et al . 2005 , yoon et al .
2006 , woosley & heger 2006 ) , or produce weaker explosions at higher metallicities ( macfadyen & woosley 1999 ) .
the work presented here originally aimed to observationally confirm and quantify the predicted role of metallicity in lgrb production and progenitor evolution .
however , the results illustrate that the effects of metallicity on lgrbs are complex and do not agree with these expectations , suggesting that the predictions of stellar evolutionary theory and progenitor models may require further development .
, top ) and intermediate - redshift ( @xmath2 , bottom ) lgrb host galaxies ( filled circles ) .
the nearby lgrb hosts are compared to binned mass - metallicity data for a sample of @xmath353,000 sdss star - forming galaxies , where the open diamonds represent the median of each bin and the dashed / dotted lines show the contours that include 68%/95% of the data ( tremonti et al .
2004 ) . for the intermediate - redshift hosts we plot binned mass - metallicity data for a sample of 1330 emission line galaxies from the deep2 survey ( open squares ; zahid et al .
for the @xmath4 host galaxy of grb 980703 , where we can not distinguish between the upper and lower metallicities given by the @xmath5 diagnostic , we plot both metallicities and connect the data point with a dotted line to indicate their common origin from the same host spectrum.,width=480 ] in levesque et al .
( 2010a , b ) we conducted a uniform rest - frame optical spectroscopic survey of @xmath6 lgrb host galaxies , using the keck telescopes at mauna kea observatory and the magellan telescopes at las campanas observatory .
the sample was restricted to confirmed long - duration bursts with well - associated and observable host galaxies . from these spectra
we were able to determine a number of key parameters for the star - forming lgrb host galaxies , including metallicity , ionization parameter , young stellar population age , sfr , and stellar mass .
the primary metallicity diagnostic used in this work was the ( [ oiii ] @xmath75007 + [ oiii ] @xmath74959 + [ oii ] @xmath73727)/h@xmath8 ( @xmath5 ) diagnostic ( kewley & dopita 2002 , kobulnicky & kewley 2004 ) ; for our full sample we found an average @xmath5 metallicity of log(o / h ) + 12 = 8.4 @xmath9 0.3 .
our stellar mass estimates were determined using the _ le phare _ code ( ilbert et al .
2009 ) , fitting multi - band photometry for the host galaxies ( savaglio et al .
2009 ) to stellar population synthesis models adopting a chabrier imf , the bruzual & charlot synthetic stellar templates , and the calzetti extinction law ( bruzual & charlot 2003 , chabrier 2003 , calzetti et al .
the fitting yielded a stellar mass probability distribution for each host galaxy , with the median of the distribution serving as our estimate of the final stellar mass . for our sample
, we found a mean stellar mass of log(@xmath10 .
these metallicities and stellar masses were used to construct a mass - metallicity relation for lgrb host galaxies , which we plot in figure 1 . for comparison
, we also compare our results to two samples of star - forming galaxies with comparable redshifts .
the nearby ( @xmath11 ) lgrb hosts are compared to @xmath12 star - forming sdss galaxies , while the intermediate - redshift ( @xmath2 ) hosts are compared to 1,330 galaxies from the deep extragalactic evolutionary probe 2 ( deep2 ) survey ( tremonti et al .
2004 , zahid et al .
surprisingly , we found a strong and statistically significant correlation between stellar mass and metallicity for lgrb hosts out to @xmath6 ( pearson s @xmath13 , @xmath14 ) , with the relation showing no evidence for a clear metallicity cut - off above which lgrbs can not be formed - instead , the overall lgrb mass - metallicity relation is offset from the mass - metallicity relation for star - forming galaxies by an average of @xmath15 dex in metallicity .
the phenomenological explanation for this offset is unclear .
( center ) , and @xmath16 ( bottom ) .
the hosts are separated into redshift bins in order to better illustrate redshift effects .
two hosts with both lower- and upper - branch@xmath5 metallicities ( the hosts of grb 020405 at @xmath17 and grb 980703 at @xmath4 ) are shown as lower and upper data points connected by dotted lines .
upper and lower limits are indicated by arrows .
hosts with both upper _ and _ lower limits on their @xmath16 values are shown as data points connected by solid lines.,width=480 ] lacking observational evidence for a pure cut - off metallicity for lgrb formation , we instead consider the possibility that lgrbs at high metallicity may simply produce less energetic explosions ; that is , explosions with a lower isotropic ( @xmath18 ) or beaming - corrected ( @xmath19 energy release in the gamma - ray regime , where @xmath20 is the grb jet opening angle ) energy release ( frail et al .
previous studies of several local lgrbs suggested a strong correlation between these parameters , with high - metallicity lgrbs producing markedly - lower @xmath18 ( stanek et al .
2006 ) . by combining energetic parameters available in the literature with our previously - determined host galaxy metallicities
, we were able to reproduce this comparison ( levesque et al .
2010c ; figure 2 ) .
a comparison with redshift was considered as well , to highlight any potential correlation that may appear as an artifact of metallicity evolution with redshift .
however , we found that there is no statistically significant correlation between metallicity and redshift , _ or _ between metallicity and @xmath18 or @xmath16 .
this result is at odds with the previously - predicted and tentatively - observed inverse correlation , and appears to demonstrate that metallicity has no clear impact on the final explosive properties and gamma - ray energy release of an lgrb progenitor .
0.1 dex below the theoretical relation where explosion site metallicity and host metallicity are identical , plotted here as a gray dashed line , though this is within the uncertainty of the metallicity diagnostics.,width=480 ] it is important to note one strong limitation of current lgrb studies : their reliance on global metallicities . for the majority of lgrb hosts at @xmath21 , pinpointing the lgrb explosion site and acquiring site - specific spectra within the small , faint host galaxies is a difficult proposition .
however , these site - specific studies _ are _ possible for a key sample of seven nearby spatially - resolved lgrb host galaxies . for this subset of hosts
we can determine metallicities and star - formation rates directly at the lgrb host site as well as in the surrounding star - forming regions of the galaxy .
this allows us to pinpoint the precise environments that produce lgrbs and place these sites in context with their global host galaxy properties .
despite the enormous value of such observations , only a small handful of spatially - resolved lgrb hosts have been previously studied .
christensen et al . (
2008 ) obtained integral field unit spectroscopy of the @xmath22 host galaxy of grb 980425 , determining metallicities at 23 different sites across the host .
thne et al .
( 2008 ) examined spatially - resolved ism properties in the @xmath23 host galaxy of grb 060505 .
levesque et al .
( 2010d ) measured high metallicities at both the nucleus and grb explosion site within the massive @xmath24 host galaxy of grb 020819b .
most recently , levesque et al . (
2011 ) presented a detailed analysis of the grb 100316d host environment at @xmath25 . by obtaining longslit spectra of the host complex at two different position angles using ldss3 on magellan
, we were able to extract spatially - resolved profiles for a number of key diagnostic emission features , thus constructing metallicity and star formation rate profiles across the host that focused on both the specific lgrb explosion site and the diffuse emission of the host complex .
based on this analysis , we determined that grb 100316d happened near the lowest - metallicity and most strong star - forming region of the host complex .
however , this work also revealed only a very weak metallicity gradient within the host complex .
combined with the previous studies of nearby lgrb hosts , we found that , within this small sample , host " or
global " metallicities were comparable to metallicities at the grb explosion sites ( figure 3 ) , suggesting that global metallicities may indeed be valid proxies for explosion site metallicities in higher - redshift lgrb host galaxy studies . expanding this work to the remaining resolved lgrb hosts ( the hosts of grbs 020903 , 030329 , and 060218 )
would allow us to further explore this interesting result .
based on the work described above , the role of metallicity in lgrb production and progenitor evolution remains a mystery .
lgrbs occur preferentially in low - metallicity environments but do not show any evidence of a cut - off metallicity above which lgrb production is suppressed .
there is also no statistically significant correlation between the gamma - ray energy release of lgrbs and the metallicity of their host environments .
finally , it appears that these results can not be attributed to effects of local metallicities within a globally - sampled host , given that several nearby lgrb hosts show evidence of minimal metallicity gradients and explosion site metallicities that are representative of the global environment . in light of these results , it is worth considering alternative models of lgrbs progenitors and stellar evolution , as well as new analytical means of examining metallicity effects in lgrbs .
for example , it is possible that additional comparisons with other explosive properties of lgrbs , such as x - ray fluence or blastwave velocity , could still reveal a correlation with host metallicity .
alternative progenitor scenarios , such as magnetars or binary channels , could also potentially agree with these observed results .
binary progenitor scenarios in particular are an intriguing possibility .
one of the most common binary progenitor scenarios for lgrbs invokes a terminal common envelope phase where the outer envelope is ejected and the stellar cores coalesce .
this manner of binary is predicted to occur at a higher rate - but _ not _ exclusively - in low - metallicity environments due to stellar wind effects , with weaker stellar winds permitting the evolution of binaries at closer proximities ( podsiadlowski et al .
a second common progenitor model considers an interim common envelope phase , where the outer envelope is ejected , followed by a contact binary phase .
this is also predicted to occur at a higher rate in low - z environments , due to a widening range of roche lobe radii that can permit a binary to enter and survive an interim common envelope phase while still maintaining roche lobe overflow ( linden et al .
2010 ) .
finally , in addition to new progenitor scenarios , new treatments of stellar evolution with rotation are also compelling .
detailed treatments of differential rotation in massive stars have profound effects on the properties and populations of massive stars ( ekstrom et al .
2012 , levesque et al .
2012 ) , and at low metallicities these effects are expected to be further enhanced ( leitherer 2008 ) .
georgy et al . ( 2012 ) recently examined the effects of rotation on the production of evolved massive stars , supernovae , and lgrbs , using the new stellar rotation models of ekstrom et al .
( 2012 ) at solar metallicity , and were able to produce favorable conditions for lgrb formation in 40 - 60m@xmath26 stars at solar metallicity . indeed
, the latest stellar rotation models actually overproduce the predicted rate of lgrbs , although the introduction of additional parameters in the stellar interiors , such as strong coupling of the core to the stellar surface due to interior magnetic fields , could decrease this rate and bring predictions of the models into very good agreement with observations .
collaborators on this work included megan bagley , edo berger , ryan chornock , andrew fruchter , john graham , lisa kewley , and h. jabran zahid .
the author is supported by nasa through einstein postdoctoral fellowship grant number pf0 - 110075 awarded by the chandra x - ray center , which is operated by the smithsonian astrophysical observatory for nasa under contract nas8 - 03060 .
2002 , _ apj _ , 121 , 1111 2003 , _ mnras _ , 344 , 1000 2000 , _ apj _ , 533 , 682 2003 , _ pasp _ , 115 , 763 2007 , _ apj _ , 671 , 272 2008 , _ a&a _ , 490 , 45 2012 , _
a&a _ , 537 , 146 2001 , _ apj _ , 562 , 55 2006 , _ nature _ , 441 , 463 2007 , _ arxiv : astro - ph/0703458v2 _ 1998 , _ nature _ , 395 , 670 2012 , _
a&a _ , in press ( arxiv:1203.5243 ) 2005 , _ a&a _ , 443 , 581 2009 , _ apj
_ , 690 , 1236 2002 , _ apjs _ , 142 , 35 2004 , _ apj _ , 617 , 24 2009 , _ apj _ , 702 , 377 2002 , _ apj _ , 557 , 389 2008 , _ iau symp .
255 , low - metallicity star formation : from the first stars to dwarf galaxies _
l. k. hunt , s. madden , & r. schneider ( cambridge : cup ) , 305 2010a , _ aj _ , 139 , 694 2010b , _ aj _ , 140 , 1557 2010e , _ apj _ , 725 , 1337 2010c , _ apj _ , 712 , l26 2011 , _ apj _ , 739 , 23 2012 , _ apj _ , in press ( arxiv:1203.5109 ) 2010 , _ apj _ , 725 , 1984 1999 , _ apj _ , 524 , 262 2004 , _ apj _ , 609 , l5 2006 , _ apj _ , 645 , l21 2008 , _ aj _ , 135 , 1136 2010 , _ mnras _ , 406 , 840 2009 , _ apj _ , 1091 , 182 2003 , _ apj _ , 591 , l17 2006 , _ acta astron .
_ , 56 , 333 2011 , _ mnras _ , 411 , 2792 2008 , _ apj _ , 676 , 1151 2004 , _ apj _ , 613 , 898 2001 , _
a&a _ , 369 , 574 2005 , _ a&a _ , 442 , 587 2007 , _ apj _ , 657 , 367 1993 , _ apj _ , 405 , 273 2006 , _ ara&a _ , 44 , 507 2006 , _ a&a _ , 460 , 199 2011 , _ apj _ , 730 , 137 | host galaxies are an excellent means of probing the natal environments that generate gamma - ray bursts ( grbs ) .
recent work on the host galaxies of short - duration grbs has offered new insights into the parent stellar populations and ages of their enigmatic progenitors .
similarly , surveys of long - duration grb ( lgrb ) host environments and their ism properties have produced intriguing new results with important implications for long grb progenitor models .
these host studies are also critical in evaluating the utility of lgrbs as potential tracers of star formation and metallicity at high redshifts .
i will summarize the latest research on lgrb host galaxies , and discuss the resulting impact on our understanding of these events progenitors , energetics , and cosmological applications . |
residual symmetries in the mass matrices are a means to determine rows and columns in the mixing matrix @xmath1 as pure numbers , independent of the values of the masses , by assuming finiteness of the underlying flavour symmetry group @xmath2for recent reviews on group theory see for instance @xcite .
this model - independent approach contrasts , for instance , with the idea of texture zeros , where mass ratios are related to mixing angles and cp phases . in recent years , assuming neutrinos are majorana particles , much effort has gone into the discussion of the lepton mixing matrices @xmath1 from residual symmetries . though a lot of results have been achieved @xcite , the present status has not been reached in a systematic way .
guts @xcite . ]
thus it is by no means clear if the cases existing in the literature encompass all relevant mixing matrices . in the present paper we close this gap by accomplishing a complete classification of all possible @xmath1 completely determined by residual symmetries .
it turns out that indeed all relevant mixing matrices can already be found in the literature .
before we delve into the advertised classification , we shortly review the idea of residual symmetries , which also allows us to introduce our notation . the mass lagrangian obtained through flavour symmetry breaking has the form @xmath3 where @xmath4 is the charge - conjugation matrix , @xmath5 and @xmath6 contain the three left and right - handed charged - lepton fields , respectively , and @xmath7 consists of the three left - handed neutrino fields .
the neutrino mass matrix @xmath8 is symmetric , due to the assumed majorana nature of the neutrinos , but complex in general .
residual symmetries in the mass matrices are defined via @xmath9 where @xmath10 and @xmath11 are unitary matrices . in the standard model , the fields @xmath7 and @xmath5 are together in a weak doublet , and we assume that this is also true for any extension of it .
therefore , @xmath10 and @xmath11 are given in the same weak basis .
the idea of residual symmetries is that any matrix @xmath10 and any matrix @xmath11 of equation ( [ residual ] ) contributes to the complete flavour symmetry group @xmath2 of some unknown extension of the standard model .
the group @xmath2 is broken in the charged - lepton sector to the group @xmath12 with @xmath13 and in the neutrino sector to @xmath14 with @xmath15 . because of the specific form of the mass term in the case of majorana neutrinos , @xmath14 must be a subgroup of @xmath16 @xcite , or @xmath17 the klein four - group , if we confine ourselves to matrices @xmath11 with @xmath18 .
in this paper we want to discuss all possibilities such that the residual symmetry groups @xmath12 and @xmath14 fully determine the lepton mixing matrix @xmath1 .
thus we identify @xmath14 with the klein four - group : @xmath19 with three commuting matrices @xmath20 with the properties @xmath21 and @xmath22 for @xmath23
. as we will see in the following , in most cases @xmath12 is generated by a single matrix @xmath10 . in some cases , however , one needs two matrices @xmath24 , @xmath25 in order to fully determine @xmath1 ; in such instances , one of these matrices alone fixes only a row in @xmath1 . how can the residual symmetries determine the mixing matrix @xmath1 ?
let us denote the diagonalising matrices of @xmath26 and @xmath8 by @xmath27 and @xmath28 , respectively .
then @xmath29 and the lepton mixing matrix is given by @xmath30 since @xmath31 one can show that @xmath28 not only diagonalises @xmath8 , but also the matrices @xmath20 .
therefore , the requirement that @xmath32 is diagonal for @xmath33 already determines @xmath28 and no knowledge of @xmath8 is necessary in this framework .
the same argument applies to @xmath26 and @xmath10 .
the diagonalisation of @xmath10 determines @xmath27 , if @xmath10 has non - degenerate eigenvalues .
the only other non - trivial case , differing from the previous one , is that of two commuting @xmath24 , @xmath25 , which after diagonalisation have the form @xmath34 or permutations thereof . in a nutshell , in the framework of residual symmetries , @xmath28 and @xmath27 are determined by @xmath14 and @xmath12 , respectively , whence one gets a handle on the lepton mixing matrix @xmath1 .
is only partially determined by the residual symmetries , then either a column or a row is fixed @xcite . ] however , the framework set forth above is still too general to be treatable .
an important extra ingredient is that the group @xmath2 generated by @xmath14 and @xmath12 is _
finite_. this is an _
ad hoc _ assumption , but it has the important consequence that the eigenvalues of all @xmath13 , the eigenvalues of all products of the generators of @xmath2 , _ e.g. _
@xmath35 , and the eigenvalues of all multiple products of the generators are roots of unity .
are @xmath36 , thus they are trivially roots of unity . ] by taking traces of these matrices we obtain sums of roots of unity , and therefore known mathematical results concerning these type of sums are applicable for a previous application see @xcite , which form the basis for our classification of all cases of @xmath1 . note that in order for the group @xmath2 to be finite it is necessary but in general not sufficient that the eigenvalues of its generators are roots of unity .
nevertheless , in the cases discussed in this paper , it turns out that , if the group generators and some of their products have finite order , then @xmath2 is finite . in summary , the classification of possible mixing matrices @xmath1
is based on the following premises : 1 .
the standard model with three families of leptons is the low - energy model of some theory of lepton flavour .
neutrinos have majorana nature .
the residual group @xmath14 is the klein four - group , _
i.e. _ @xmath37 .
the group @xmath2 , generated by the elements of @xmath12 and @xmath14 , is finite .
the mixing matrix @xmath1 is completely determined by the residual symmetries . in the following
, we will always work in a basis where the @xmath20 are diagonal : @xmath38 in this convenient basis , @xmath28 is an unknown diagonal matrix of phase factors .
therefore , in the framework of residual symmetries , the majorana phases in @xmath1 are indeterminate and the lepton mixing matrix @xmath1 is simply given by @xmath39 up to rephasings . moreover , since in our framework the order of the charged - lepton and neutrino masses is undefined , it is only possible to determine @xmath1 up to independent row and column permutations . the paper is organized as follows . in section [ tools ]
we discuss some mathematical results .
these are used later , in section [ basic forms ] , to determine the possible forms of @xmath40 which is defined as the matrix of absolute values of @xmath10 : @xmath41 it turns out that there are only five such basic forms ( later on reduced to three by additional considerations ) , modulo independent permutations of rows and columns .
then , in section [ equivalent ] , we perform a general discussion of equivalent forms of @xmath10 , which are those forms which lead to trivial variations of the mixing matrix @xmath1 .
in particular , we investigate the freedom of permutations and argue that , without loss of generality , we can confine ourselves to matrices @xmath10 which can be written as @xmath42 in this formula , @xmath43 contains the `` internal '' or ckm - type phase of a unitary matrix and @xmath44 is a diagonal matrix of phase factors .
we identify in section [ permutations ] the genuinely different forms of @xmath40 which emerge from the five basic forms by permutations .
it remains to investigate the phases of @xmath10 .
we first determine in section [ internal ] the `` internal '' phase of @xmath10 for each form of @xmath40 . in section
[ not finite ] we demonstrate that forms 1 and 4 do not lead to finite groups .
the extensive section [ external ] is devoted to the computation of @xmath44 or `` external '' phases in @xmath10 for the basics forms 2 , 3 and 5 . at this point
we have completely determined @xmath10 up to equivalent forms , so in this section we also provide our main result , the possible cases of @xmath1 . in section [ combining ] we discuss the remaining solutions where @xmath12 has two generators , each with a twofold degenerate eigenvalue
see equation ( [ t1t2 ] ) .
we conclude with section [ concl ] .
lengthy calculational details are deferred to appendices .
[ [ vanishing - sums - of - roots - of - unity ] ] vanishing sums of roots of unity : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we will make use of three theorems related to roots of unity . the first theorem concerns vanishing sums of roots of unity .
some remarks are appropriate before we reproduce the theorem of conway and jones ( theorem 6 in @xcite ) .
formal sums of roots of unity with rational coefficients form a ring @xcite .
a sum of roots of unity @xmath45 is _ similar _ to a sum of roots of unity @xmath46 if there is a rational number @xmath47 and a root of unity @xmath48 such that @xmath49 .
the length of a sum of roots of unity @xmath46 is the number of distinct roots involved .
note , however , that @xmath50 for any root of unity @xmath51 , _
_ i.e.__@xmath52 does _ not _ count separately for the length of @xmath46 ; in other words , @xmath53 has length zero .
the roots occurring in the following theorem are @xmath54 note that @xmath55 while @xmath56 expressed in radicals is too complicated to be shown here .
[ tcj ] let @xmath46 be a non - empty vanishing sum of length at most 9 .
then either @xmath46 involves @xmath57 , @xmath58 and @xmath59 for some root @xmath57 , or @xmath46 is similar to one of 1 .
@xmath60 , 2 .
@xmath61 , 3 .
@xmath62 , 4 .
@xmath63 , 5 .
@xmath64 , 6 .
@xmath65 , 7 .
@xmath66 , 8 .
@xmath67 .
[ [ non - vanishing - sums - of - roots - of - unity ] ] non - vanishing sums of roots of unity : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the second theorem concerns sums of roots of unity with values on the unit circle in the complex plane .
[ sum=1 ] let @xmath68 be an @xmath69-th root of unity , i.e. @xmath70 , and let @xmath71 be a sum with integer coefficients @xmath72 . if @xmath73 , then @xmath46 is itself a root of unity .
a proof of this theorem can for instance be deduced from lemma 1.6 in @xcite and a discussion of this issue can be found on @xcite .
the link between theorem [ sum=1 ] and the problems in the present paper is provided by the following theorem .
[ troots ] let @xmath2 be a finite group with @xmath13 and let @xmath74 be one of the numbers @xmath75 , @xmath76 or @xmath77 .
moreover , @xmath78 is a diagonal element and @xmath79 , @xmath80 are off - diagonal elements of @xmath10 . then the following holds : @xmath81 , @xmath82 .
* proof : * it is easy to show that @xmath83 since @xmath84 and @xmath2 is finite , the eigenvalues of @xmath10 and @xmath85 must be roots of unity .
therefore , the traces of these matrices are sums over three roots of unity and @xmath86 is a sum over six roots of unity .
thus , if @xmath87 , theorem [ sum=1 ] tells us that @xmath86 is a root of unity .
now suppose that @xmath88 or @xmath89 .
since @xmath90 we find that both @xmath91 are normalized sums over roots of unity .
again , theorem [ sum=1 ] applies .
this finishes the proof of the first part of the theorem . for the second part we note that , since @xmath10 is a unitary matrix , @xmath92 for @xmath23 .
we know that @xmath93 is a product of three roots of unity and that @xmath94 , @xmath95 are sums over roots of unity . therefore , @xmath96 is a sum over roots of unity as well .
now we apply once more theorem [ sum=1 ] and the proof is finished .
@xmath97 [ [ root - of - unity - or - not - that - is - the - question ] ] root of unity or
not , that is the question : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the next theorem addresses the problem of finding out whether a complex number @xmath68 with @xmath98 is a root of unity or not .
we can answer this question if we know a polynomial @xmath99 with rational coefficients such that @xmath100 . because then from @xmath99 we can determine the minimal polynomial @xmath101 of @xmath68 , which is defined as being irreducible over the rational numbers and normalized .
since the minimal polynomial is unique the following statement holds .
[ roots ] let @xmath68 be a complex number with @xmath98 and @xmath101 its minimal polynomial . then @xmath68 is a root of unity if and only if @xmath101 is a cyclotomic polynomial .
note that cyclotomic polynomials are the minimal polynomials of roots of unity .
there is a straightforward corollary to theorem [ roots ] which makes use of the fact that cyclotomic polynomials have integer coefficients .
[ roots1 ] if there are non - integer coefficients in @xmath101 , then @xmath68 is not a root of unity .
in this section we determine the basic forms of @xmath40 which follow from our assumptions listed in the introduction .
we begin by considering the matrices @xmath102 we argue that these matrices have eigenvalues 1 , @xmath103 , @xmath104 .
+ * proof : * the matrices @xmath105 fulfill @xmath106 therefore , the complex conjugate of every eigenvalue of @xmath105 is also an eigenvalue .
moreover , the product of the three eigenvalues must be one .
so the spectrum of @xmath105 contains one complex eigenvalue @xmath103 , its complex conjugate and 1 .
@xmath97 + since @xmath107 it follows that @xmath108 written in terms of the eigenvalues , these relations are @xmath109 for @xmath110 .
these equations can be tackled with theorem [ tcj ] .
suppose we have obtained the possible @xmath103 .
defining @xmath111 we observe that the unitarity of @xmath10 gives @xmath112 furthermore , we obtain @xmath113 in this way we arrive at the relation @xmath114 thus in order to obtain @xmath115 we are left with the task of finding all possible solutions of the generic equation @xmath116 with roots of unity @xmath117 , @xmath118 , @xmath119 . these solutions are derived in appendix [ possible eigenvalues - app ] through application of theorem [ tcj ] .
it turns out that , up to permutations and complex conjugations , equation ( [ lambda2 ] ) has only three solutions : @xmath120 where @xmath121 is an arbitrary root of unity .
the details of the tedious manipulations for finding all possible forms of the matrix @xmath122 are given in appendix [ basic forms - app ] .
here we mention only that it is important to take into account that @xmath122 stems from a unitary matrix see inequality ( [ ineq ] ) , which rules out many cases .
the surprising result of appendix [ basic forms - app ] is that , up to permutations of rows and columns , there are only five basic forms of @xmath122 . in the following , however , we will rather use @xmath40 instead of @xmath122 .
therefore , we display here the basic forms of @xmath40 obtained by inserting the solutions ( [ abc ] ) into equation ( [ t - lambda ] ) : [ basic ] @xmath123 note that form 5 derives from equation ( [ basic5-var ] ) via @xmath124 and , therefore , @xmath125 with an arbitrary integer @xmath126 . since we know that @xmath127 is a rational angle , _
i.e. _ a rational multiple of @xmath128 , it follows that @xmath57 must be a rational angle too .
it is expedient to reflect on the ambiguities in the determination of @xmath1 with the method used here .
the first observation is that the choice of basis for the @xmath20 in equation ( [ sss ] ) leaves still some freedom for basis transformations .
the reason is that @xmath129 where @xmath130 is a @xmath131 permutation matrix and @xmath132 is a diagonal matrix of phase factors , may transform one @xmath20 into another , but the set of matrices of equation ( [ sss ] ) is invariant under this similarity transformation . on the other hand , the transformation ( [ v ] ) acts also on @xmath10 as @xmath133 and we can use this freedom to fix some conventions for @xmath10 .
this is an important issue because it prevents us from over - counting the number of cases and from studying equivalent cases twice .
two matrices @xmath10 , @xmath134 , which are connected via @xmath135 , are called equivalent in the following . the freedom in reordering and rephasing expressed by equation ( [ tv ] )
can be used for a strategy to determine all inequivalent forms of @xmath10 . 1 .
for each of the five basic forms of equation ( [ basic ] ) , we take the matrix displayed there as the starting point and indicate it by the subscript @xmath136 .
2 . due to equation ( [ tv ] )
we are allowed to confine ourselves to permutations from the right in order to find inequivalent matrices @xmath137 with @xmath138 3 .
in general there will be less than six inequivalent matrices @xmath40 for each of the basics forms , as some matrices which emerge from each other by a permutation from the right might still be equivalent due to the equality of some matrix elements or , in the case of form 5 , because it is possible to make the exchange @xmath139 .
4 . after having found , for each basic form , the inequivalent matrices @xmath140 , where @xmath141 is a permutation matrix , we determine the internal phase associated with each @xmath137 . denoting the resulting matrix by @xmath142 , we can choose phase conventions such that @xmath143 for all @xmath144 .
the matrix @xmath10 will also have external phases . applying again equation ( [ tv ] )
, we can assume that these phases are taken care of by multiplying @xmath43 by a diagonal matrix of phase factors @xmath44 from the right .
in essence , the @xmath44 will be determined by the requirement that @xmath10 has finite order .
note , however , that , given a basic form and one of its possible @xmath142 , there can be several solutions of @xmath44
. 6 . finally ,
since the matrices @xmath20 are real , for every solution @xmath10 there is the complex conjugate solution @xmath145 .
eventually we are not interested in the possible @xmath10 but in the possible mixing matrices @xmath1 . according to equation ( [ u1 ] )
the two matrices are linked via @xmath146 the order of the eigenvalues is indeterminate and @xmath147 is invariant under phase transformations .
therefore , @xmath1 can undergo rephasing and permutations from the left .
equation ( [ tv ] ) implies that the same holds from the right , with rephasings and permutations independent from those on the left .
to this indeterminacy one has to add the possibility of complex conjugation of @xmath1 . given these trivial phase variations of the mixing matrix and the need to factor them out , we will be focusing on @xmath148 .
the group @xmath2 , determined by the three @xmath20 and by one or two matrices @xmath10 which generate @xmath12 , is not changed by the above manipulations of @xmath10 . actually , what we have at hand is not directly the group @xmath2 but its representation on the three leptonic gauge doublets . in this sense , the change @xmath149 for the @xmath13 corresponds to switching from one representation to its complex conjugate . we can also remove an overall phase factor @xmath150 from a @xmath10 , if @xmath150 is root of unity
this will , in general , change the group , but the group will remain finite .
in order to investigate the inequivalent permutations of the columns of @xmath151 , it is appropriate to use the representation of the permutations @xmath152 as permutation matrices : @xmath153 it is easy to check that the representation property @xmath154 is fulfilled for any two @xmath155 .
the basic relation we need is given by latexmath:[\[m(q)^t
|t|_a m(q ) = |t|_a \quad \rightarrow \quad m(q)^t \big ( |t|_a m(p ) \big ) m(q ) = @xmath151 under some @xmath157 means that @xmath158 and @xmath159 are equivalent provided that permutations @xmath160
, @xmath161 are related through conjugation by @xmath47 , _
i.e. _ @xmath162 .
now we discuss all five basic forms under this aspect .
it is easy to see that @xmath151 of form 1 , equation ( [ basic1 ] ) , is invariant under _ all _ @xmath157 .
since @xmath163 has three conjugacy classes , there are three inequivalent forms .
apart from @xmath151 , we choose @xmath164 $ ] and @xmath165 $ ] . turning to form 2 , equation ( [ basic2 ] )
, @xmath151 is obviously invariant under @xmath166 . given that @xmath167 = \left|t\right|_{a } m \left[\left(132\right)\right]$ ] , @xmath168 , and @xmath169 , we can conclude that the only other inequivalent form is @xmath170 $ ] .
concerning form 3 , equation ( [ basic3 ] ) , we find that it is invariant under cyclic permutations .
since one transposition is transformed into the other two by conjugation with the two cyclic permutations , we can for instance choose @xmath164 $ ] , as representative of the transpositions .
finally , cyclic permutations give the inequivalent forms @xmath165 $ ] and @xmath171 $ ] .
the matrix @xmath151 of form 4 , equation ( [ basic4 ] ) , is invariant under @xmath166 .
in contrast to form 2 , all columns are different .
given that @xmath172 and @xmath173 , we are lead to a choice of inequivalent forms @xmath164 $ ] , @xmath174 $ ] and @xmath175 $ ] .
since in form 5 , equation ( [ basic5 ] ) , the angle @xmath57 is free apart from being rational , the exchange @xmath139 relates equivalent forms .
so in effect we are lead back to the argumentation applied to form 2 .
for definiteness , we choose @xmath176 $ ] . in summary ,
the inequivalent forms obtained by the above discussion are given by [ basic5-permutations ] @xmath177
in order to compute the internal phase of @xmath10 , we can use the formulas provided in @xcite : @xmath178 it is well known that the existence of an internal phase of a unitary matrix is independent of the phase convention @xcite .
this can also be seen from the above equations for @xmath179 and @xmath180 . however , where to place the internal phase in @xmath10 is , of course , convention - dependent .
it suffices to find the internal phase for subform a in equation ( [ basic5-permutations ] ) for each of the five basic forms , since the other subforms emerge from a by permutation of the columns . for each form
we choose a suitable set of elements @xmath181 of @xmath10 which are real and positive by convention .
this set contains three of the elements occurring in @xmath179 and @xmath180
. then we apply equations ( [ r ] ) and ( [ j ] ) and compute the phase of the remaining element . from this , taking into account that the columns of a unitary matrix form an orthonormal system , the matrix @xmath43 , defined in equation ( [ t ] ) , is obtained .
since these procedures are standard methods in linear algebra , we only display the results : @xmath182 of the phase factors occurring in these formulas , @xmath183 is defined in equation ( [ obg ] ) and @xmath184 as discussed in section [ equivalent ] , @xmath43 only needs to be computed modulo complex conjugation .
we begin with @xmath185 theorem [ troots ] requires that the phase factors of @xmath186 , @xmath187 and @xmath188 are roots of unity .
therefore , these phases are @xmath189 , @xmath190 and @xmath191 , respectively , with roots of unity @xmath192 ( @xmath193 ) . from these equations
we derive @xmath194 , etc .
hence it follows that @xmath195 with roots of unity @xmath196 .
furthermore , @xmath197 since @xmath10 has finite order , @xmath198 has to be a root of unity .
but one can easily check that @xmath198 fulfills @xmath199 therefore , according to theorem [ roots1 ] , @xmath198 is not a root of unity and @xmath10 of equation ( [ 1a ] ) does not belong to a finite group .
next we consider @xmath200 applying again theorem [ troots ] , we find that the phase factors of @xmath186 , @xmath201 and @xmath202 are roots of unity . from this
it is easy to show that @xmath203 , @xmath204 and @xmath205 with roots of unity @xmath196 .
now we proceed as with form 1a and obtain again equation ( [ det1 ] ) , which excludes form 1b as well .
finally we discuss @xmath206 here theorem [ troots ] tells us that the phase factors of @xmath207 , @xmath201 and @xmath202 are roots of unity and one obtains the same relations for @xmath208 as for form 1b .
computing the determinant of @xmath10 , we are again lead to equation ( [ det1 ] ) .
hence form 1c is excluded .
we proceed analogously to form 1 .
we first consider @xmath209 theorem [ troots ] requires that the phase factors of @xmath210 , @xmath201 and @xmath202 are roots of unity , therefore , the @xmath208 must all be roots of unity . computing the determinant of @xmath10 , we obtain @xmath211.\ ] ] since @xmath10 has finite order , its determinant is a root of unity and so is @xmath212
. then also @xmath213 must be root of unity .
but @xmath214 is a root of the equation @xmath215 therefore , according to theorem [ roots1 ] , it can not be a root of unity , which is a contradiction .
hence form 4a does not lead to a finite group .
now we consider @xmath216 in the same way as for form 1a , we find that all @xmath208 are roots of unity . in this case @xmath217 and the argument excluding this form goes through as before .
next we discuss @xmath218 the determinant of this form is equal to the one of form 4b , and the phases @xmath208 have to be roots of unity as before .
so one can exclude this form as well .
the final case is @xmath219 again , all @xmath208 must be roots of unity and the determinant is the same as for form 4a .
it is interesting to note that forms 1 and 4 , the only two of the basic five forms which have a non - trivial internal phase , do not lead to finite groups .
thus we do not discuss them further .
it remains to investigate the remaining forms 2 , 3 and 5 .
we still have to determine the external phases @xmath220 of @xmath221 for a total of eight subforms of @xmath43 : [ subforms ] @xmath222 , & & & & \\ & \mbox{form 3 : } & \quad & \widetilde t_a , & \quad \widetilde t_b & = \widetilde t_a\ , m[(23 ) ] , & \quad \widetilde t_c & = \widetilde t_a\ , m[(132 ) ] , & \quad \widetilde t_d & = \widetilde t_a\ , m[(123 ) ] , \\ & \mbox{form 5 : } & \quad & \widetilde t_a , & \quad \widetilde t_b & = \widetilde t_a\ , m[(123 ) ] , & & & & \end{aligned}\ ] ] which are read off from equations ( [ form2ab ] ) and ( [ 2-tilde ] ) , equations ( [ form3abcd ] ) and ( [ 3-tilde ] ) , and equations ( [ form5ab ] ) and ( [ 5-tilde ] ) , for forms 2 , 3 and 5 , respectively .
all these forms have trivial internal phases . the basic idea to determine the possible @xmath44 for each subform is the observation that , for any element @xmath223 , the matrix @xmath224 must be of one of the basics forms 2 , 3 or 5 , after having excluded forms 1 and 4 .
in particular , this holds for @xmath225 .
therefore , for any of the eight subforms under consideration the elements @xmath226 must belong to a matrix @xmath227 where @xmath228 and @xmath229 are arbitrary permutation matrices .
two remarks are in order .
let us assume that we discuss for instance @xmath10 of form 2 . 1
. then @xmath230 can belong to form 2 , 3 or 5 .
since we have used up already the freedom of permutations for the subform we begin with , we have to admit _ all _ possible permutations @xmath228 and @xmath229 . in the following we will present , for each subform in an own subsection , the results for the phases @xmath44 , the eigenvalues @xmath231 ( @xmath33 ) of @xmath10 and the matrix @xmath148 .
every triple consisting of @xmath10 , including its external phases , the eigenvalues of @xmath10 and @xmath148 is called solution .
the latter matrices , whose entries are the @xmath232 , represent the main result of this paper .
the details of the computations for each subform are deferred to appendix [ details ] .
every solution obtains a tag @xmath233 in the case of a _ complete _ determination of @xmath148 by a single @xmath10 . in the case that @xmath10 is degenerate and determines only one row of @xmath148 , then the tag is @xmath234 which stands for _
partial_. as we will see , in general , there are several solutions of external phases @xmath44 for each of the eight subforms of equation ( [ subforms ] ) .
sometimes it occurs that for one @xmath43 there are two matrices @xmath235 and @xmath236 of external phase such that the corresponding matrices @xmath237 and @xmath238 have the same @xmath148 and eigenvalues which are related by @xmath239 in this case , we assign only one solution tag . in appendix
[ form 2a ] it is shown that form 2a @xmath240 requires @xmath241 and thus leads to two subcases . in the first subcase with solution tag @xmath242 , we have @xmath243 and the mixing matrix @xmath244 the second subcase is given by @xmath245 @xmath246 for @xmath247 , the quantities @xmath248 and @xmath249 are roots of unity related to @xmath250 and @xmath251 by @xmath252 respectively .
note that the transformations @xmath253 and @xmath254 with @xmath255 lead to a permutation of the mixing pattern in @xmath148 , which accounts for 12 of the 36 possible row and column permutations .
once this permutation freedom is taken into account , it becomes clear that two roots of unity @xmath249 , @xmath256 will yield the same mixing angles if and only if @xmath257 .
while @xmath258 is known as bimaximal mixing , @xmath259 corresponds to trimaximal mixing ; for the specific choice of @xmath249 such that @xmath260 , tribimaximal mixing is obtained .
it will turn out that @xmath259 is the only series of mixing matrices genuinely involving the three flavours . from residual symmetries
, @xmath259 has for instance been derived in @xcite , while it was obtained in @xcite from gap @xcite and the small groups library @xcite .
recently , this series has been accommodated in a model @xcite .
we note that the @xmath148 of @xmath259 has a trivial ckm - type phase .
if @xmath261 in @xmath258 , then @xmath262 and the residual symmetries fix only one row in the mixing matrix @xmath1 : @xmath263 with the mixing matrix @xmath264 the symbol `` @xmath265 '' indicates the positions in @xmath148 which are not fully determined by the @xmath10 of equation ( [ t2a ] ) with @xmath266 . according to the analysis carried out in appendix [ form 2b ] , the external phases allowed for form 2b are @xmath267 where @xmath268 and @xmath269 are defined in equation ( [ phi - rho0 ] ) and @xmath270 is an arbitrary root of unity .
complex conjugation of @xmath44 leads to further allowed cases .
however , as discussed earlier , this is a trivial variation of the solutions and we ignore it .
thus there are eight subcases of form 2b .
[ [ first - subcase ] ] first subcase : + + + + + + + + + + + + + + @xmath271 with eigenvalues @xmath272 we denote this solution by @xmath273 . it gives the mixing matrix @xmath274 [ [ second - subcase ] ] second subcase : + + + + + + + + + + + + + + + @xmath275 with eigenvalues @xmath276 in this case with the solution tag @xmath277 , the seventh root of unity , @xmath278 , occurs in the eigenvalues and the mixing matrix reads @xmath279 where the @xmath280 are the roots of the equation @xmath281 their approximate numerical values are @xmath282 , @xmath283 , @xmath284 .
[ [ third - subcase ] ] third subcase : + + + + + + + + + + + + + + @xmath285 with eigenvalues @xmath286 leads to solution @xmath287 with the mixing matrix @xmath288 [ [ fourth - subcase ] ] fourth subcase : + + + + + + + + + + + + + + + @xmath289 with eigenvalues @xmath290 does not produce a new mixing matrix , but repeats the mixing matrix of solution @xmath277 of the second subcase .
moreover , the eigenvalues of @xmath10 of the second and fourth subcases are related by complex conjugation . therefore , according to the philosophy put forward in the beginning of section [ external ] , we do not assign a new solution tag here . [
[ fifth - subcase ] ] fifth subcase : + + + + + + + + + + + + + + @xmath291 with eigenvalues @xmath292 in this case , the fifth root of unity , @xmath293 , appears .
the mixing matrix of this solution is given by @xmath294 [ [ sixth - subcase ] ] sixth subcase : + + + + + + + + + + + + + + @xmath295 with eigenvalues @xmath296 here we obtain the mixing matrix @xmath297 [ [ seventh - subcase ] ] seventh subcase : + + + + + + + + + + + + + + + + @xmath298 with eigenvalues @xmath299 here again the fifth root of unity , @xmath300 , occurs .
the corresponding mixing matrix is @xmath301 [ [ eighth - subcase ] ] eighth subcase : + + + + + + + + + + + + + + + @xmath302 with eigenvalues @xmath303 these eigenvalues are proportional to the complex conjugate ones of solution @xmath304 , sixth subcase , and the mixing matrix is also the same .
thus we do not assign a new solution tag . according to the analysis in appendix [ form 3a ] ,
the possible external phases for form 3a are @xmath305 with arbitrary roots of unity @xmath250 , @xmath251 and @xmath270 . [ [ first - subcase-1 ] ] first subcase : + + + + + + + + + + + + + + @xmath306 with eigenvalues @xmath307 the corresponding mixing matrix is @xmath308 [ [ second - subcase-1 ] ] second subcase : + + + + + + + + + + + + + + + next we consider @xmath309 with eigenvalues @xmath310 this subcase provides the mixing matrix @xmath311 according to equation ( [ kappa-3a ] ) there are two more subcases to consider , however , it turns out that these again lead to the eigenvalues and the mixing matrix of @xmath312 .
according to the analysis in appendix [ form 3b ] , there are four different cases of external phases : @xmath313 with arbitrary roots of unity @xmath250 , @xmath270 . [
[ first - subcase-2 ] ] first subcase : + + + + + + + + + + + + + + @xmath314 has the eigenvalues @xmath315 and provides the mixing matrix @xmath316 [ [ second - subcase-2 ] ] second subcase : + + + + + + + + + + + + + + + the next subcase is @xmath317 the eigenvalues are @xmath318 and the mixing matrix is given by @xmath319 [ [ third - subcase-1 ] ] third subcase : + + + + + + + + + + + + + + @xmath320 with eigenvalues @xmath321 gives the mixing matrix @xmath322 solution @xmath323 has the same mixing matrix as @xmath324 , however , its eigenvalues are not related by way of equation ( [ lambda - related ] ) .
therefore , we count it as a separate solution . [ [ fourth - subcase-1 ] ] fourth subcase : + + + + + + + + + + + + + + + the last subcase @xmath325 has two degenerate eigenvalues : @xmath326 therefore , the mixing matrix is only partially determined : @xmath327 according to the analysis in appendix [ form 3c ] there are four solutions for the external phases : @xmath328 with arbitrary roots of unity @xmath251 , @xmath270 .
[ [ first - subcase-3 ] ] first subcase : + + + + + + + + + + + + + + we begin with @xmath329 which has the eigenvalues @xmath330 and leads to the mixing matrix @xmath331 [ [ second - subcase-3 ] ] second subcase : + + + + + + + + + + + + + + + next we tackle @xmath332 this matrix has the eigenvalues @xmath333 and provides the mixing matrix @xmath334 according to equation ( [ external-3c ] ) there are two more subcases with eigenvalues @xmath335 respectively .
it turns out that in both subcases the mixing matrix is that of @xmath336 .
since by choice of @xmath270 and @xmath251 , respectively , we can even match the eigenvalues of @xmath10 in equation ( [ eigen-3c-2 ] ) , we do not count these two subcases as separate solutions . according to appendix [ form 3d ]
there are the four solutions @xmath337 with arbitrary roots of unity @xmath251 , @xmath270 , for the external phases .
[ [ first - subcase-4 ] ] first subcase : + + + + + + + + + + + + + + we begin with @xmath338 which has the eigenvalues @xmath339 and provides the mixing matrix @xmath340 [ [ second - subcase-4 ] ] second subcase : + + + + + + + + + + + + + + + next we consider @xmath341 its eigenvalues are given by @xmath342 and the corresponding mixing matrix is @xmath343 the third and fourth subcase have the eigenvalues @xmath344 respectively .
thus they are those of the second subcase multiplied by a phase factor .
since they also reproduce the mixing matrix of the second subcase , we do not consider them further .
if @xmath10 is of form 5a , then obviously we have two - flavour mixing , which is completely off from realistic lepton mixing .
there is a rather large number of such two - flavour mixing solutions , namely complete solutions @xmath345@xmath346 , and partial solutions @xmath347@xmath348 where one row of @xmath148 is determined .
since these solutions are not relevant from the physics point of view , we defer them to appendix [ 2f ] , where we list them for completeness .
we are considering here the matrix @xmath349 where @xmath350 is a root of unity . denoting the eigenvalues of @xmath10 by @xmath351 and those of @xmath352 , with @xmath353 defined in equation ( [ sss ] ) , by @xmath354 ,
we find , by taking the determinants and traces , @xmath355 thus we obtain a vanishing sum of six roots of unity @xmath356 according to theorem [ tcj ] , the vanishing sum over six roots of unity could be similar to sum b ) , but this case can be excluded by the first relation of equation ( [ system ] ) .
then , theorem [ tcj ] allows for the following two solutions of this equality : @xmath357 with arbitrary roots of unity @xmath358 ( @xmath359 ) and @xmath360 ( @xmath361 ) .
[ [ first - subcase-5 ] ] first subcase : + + + + + + + + + + + + + + first we discuss solution ( [ 5b-1 ] ) .
the assignment @xmath362 leads to a contradiction with the first relation in equation ( [ system ] ) .
so without loss of generality we are left with @xmath363 since @xmath364 , this first subcase yields @xmath365 .
choosing without loss of generality @xmath366 , we end up with the two - flavour mixing case @xmath367 with a mixing angle of @xmath368 .
note that @xmath251 and @xmath369 are roots of unity .
since two - flavour mixing is dealt with in subsection [ two - flavour ] , we do not assign an extra solution tag to this .
[ [ second - subcase-5 ] ] second subcase : + + + + + + + + + + + + + + + now we discuss solution ( [ 5b-2 ] ) .
if we make the assignment @xmath370 it immediately follows that @xmath371 and , therefore , @xmath372 .
taking without loss of generality @xmath373 , we obtain @xmath374 and the mixing matrix @xmath375 note that here the product @xmath376 must be a root of unity .
it remains to discuss possibilities of assignment other than that of equation ( [ as ] ) . without loss of generality ,
we can confine ourselves to @xmath377 with @xmath378 , @xmath379 , @xmath380 and permutations thereof and @xmath381 , @xmath382 , @xmath383 and permutations thereof . with this assignment ,
the first relation in equation ( [ system ] ) requires @xmath384 because of @xmath385 , @xmath386 and @xmath387 , we arrive at @xmath388 therefore , also for the general assignment in equation ( [ as1 ] ) we obtain @xmath371 and we are lead back to solution @xmath389 .
here we discuss the cases where two generators @xmath391 are necessary in order to fully determine the mixing matrix @xmath1 .
this happens if each @xmath392 has a twofold degenerate eigenvalue
see equation ( [ t1t2 ] ) : @xmath393 and roots of unity @xmath117 , @xmath394 , @xmath118 , @xmath395 fulfilling @xmath396 actually , we can only hope to discover a new @xmath148 , not already covered by the previous section , if _
none _ of the elements of @xmath12 is non - degenerate .
let us examine this point further .
the product @xmath397 is degenerate if and only if @xmath398 whence if follows that @xmath399 next we consider @xmath400 the only possibility to avoid non - degeneracy in this matrix is @xmath401 and we end up with @xmath402 with the diagonal sign matrices @xmath20 defined in equation ( [ sss ] ) .
finally , without changing the finiteness of @xmath2 we can remove the phase factors @xmath394 and @xmath403 .
thus , the only case we have to investigate is when @xmath12 is a klein four - group . according to the discussion in section [ basic forms ] ,
we know that for every pair of indices @xmath404 the matrix @xmath405 has eigenvalues @xmath406 , @xmath407 , @xmath408 , where @xmath407 is a root of unity .
so we have the same situation as in section [ basic forms ] , with @xmath10 replaced by @xmath1 .
therefore , instead of @xmath40 , @xmath409 itself must be of one of the five basic forms and we have the solutions @xmath410 in the last form , @xmath350 must be a root of unity .
scanning the genuine three - flavour mixing solutions of section [ external ] we find that the @xmath148 of @xmath258 is of form 2 of equation ( [ basic ] ) . naturally , the @xmath148 of all solutions with @xmath10 of form 5a are of form 5 as well .
but a @xmath148 of form 1 , 3 or 4 of equation ( [ basic ] ) does not occur in the previous section .
therefore , @xmath411 , @xmath412 and @xmath413 provide new mixing matrices .
we note also that , even though the previous argument is not enough to prove that the @xmath414 solutions lead to finite groups , it can be easily checked by explicit construction of the groups generated by @xmath353 , @xmath415 , @xmath24 , and @xmath25 that they are finite ( as are the ones associated with the @xmath233 and @xmath234 solutions ) .
in this paper , assuming that neutrinos are majorana particles , we have presented a complete classification of all possible lepton mixing matrices @xmath1 such that @xmath148 , defined as the matrix with the elements @xmath416 , is completely determined by residual symmetries . in this model - independent framework , the entries of @xmath148 are obtained as pure numbers , determined by group - theoretical considerations . evidently , the resulting matrices @xmath148 are independent of any parameter of a possible underlying theory and of the lepton masses . in our analysis
we used the _
ad hoc _ assumption that the flavour group @xmath2 is finite , which allowed us to use suitable theorems related to sums of roots of unity .
we have found 22 solutions associated with a genuine three - flavour mixing matrix @xmath1 , _
i.e. _ where all three flavours are mixed : @xmath417 , @xmath389 and @xmath418 .
since four pairs of solutions produce the same mixing matrix@xmath419 , @xmath420 , @xmath421 and @xmath422there are only 17 sporadic @xmath423 patterns and one infinite series denoted by @xmath259 .
the precise groups formed by the generators of the residual symmetries @xmath424 and @xmath425 will depend on one or more phases which are free parameters ; for example , the @xmath10 generator can always be multiplied by an arbitrary root of unity . nevertheless , with the help of gap @xcite
, one can reach the conclusion that for the 17 sporadic @xmath423 patterns , the minimal groups , @xmath426 , @xmath427 and @xmath428 it can be shown that the full flavour group must always contain these as subgroups . ]
are * @xmath429 for @xmath430 , * @xmath431 for @xmath432 , @xmath433 , @xmath434 , @xmath426 , * @xmath435 for @xmath436 , @xmath437 , @xmath438 , @xmath439 , @xmath312 , @xmath440 , @xmath441 , @xmath428 , * @xmath442 for @xmath443 , @xmath444 , @xmath427 , * @xmath445 for @xmath389 . at this point a comparison of our results with those in the literature is in order .
the list of mixing matrices in @xcite induced by @xmath446 agrees exactly with our list above and the same holds true for those induced by @xmath447 ; moreover , this reference contains , among some mixing matrices of the series @xmath247 , also the cases @xmath448 and @xmath389 with @xmath449 and @xmath450 , respectively .
finally , the list of mixing matrices above associated with @xmath435 agrees with those given in @xcite in table 5 ( with labels xia / xib ) and table 6 ( with labels beginning with k ) .
so we conclude that all our sporadic cases have already been discussed in the literature . for the phenomenologically interesting infinite series denoted by @xmath259 where @xmath451 for some root of unity @xmath452 with @xmath453 coprime to @xmath69
, we have discussed minimal groups and their generators in appendix [ minimal ] .
our result is * @xmath454 with @xmath455 when @xmath456 , * @xmath457 with @xmath458 when @xmath459 .
once all row and column permutations are considered , it turns out that in the four - dimensional space of the quadruples @xmath460 the sporadic @xmath423 patterns yield 228 distinct values
see figure [ fig : numerics ] for a graphical representation of the mixing angles . can take a total of 34 different values associated with the sporadic patterns .
however , note that in some cases the absolute values of the entries of the lepton mixing matrix do not depend on the dirac phase @xmath48 .
this happens when @xmath461 , in which case there is at least one null entry in @xmath1 , and consequently there is no cp - violation associated to @xmath48 . ] all of them are excluded at 3 sigma by current neutrino oscillation data .
if one leaves out @xmath462 , there are still 212 distinct points in the space of the triples @xmath463 .
this leaves us with the infinite series of mixing patterns given in equation ( [ c2mixpattern ] ) as the only phenomenologically viable case ; taking the 3 sigma range of @xmath464 of forero _ et al .
_ , arxiv:1405.7540 in @xcite , this translates into @xmath465 for roots of unity @xmath249 .
suitable values of @xmath69 include , sorted by group size , * @xmath466 with @xmath467 , @xmath468 and @xmath469 , * @xmath470 with @xmath471 , @xmath472 and @xmath473 , * @xmath474 with @xmath475 , @xmath476 and @xmath477 , * @xmath478 with @xmath479 , @xmath480 and @xmath481 , among others .
however , we note that in the past , with less precise neutrino oscillation data , values of @xmath69 equal to 5 and 16 were valid as well see for instance @xcite , see also @xcite for earlier references .
one can easily check that equation ( [ c2mixpattern ] ) gives a trivial dirac - type phase @xmath48 . in figure
[ fig : numerics ] , the case @xmath247 is represented by three lines which are obtained by varying the root of unity @xmath249 .
the colours correspond to different column permutations : * red : @xmath482 , * blue : @xmath483 , * green : @xmath484 . from the two lower plots in figure
1 we can read off that , if @xmath247 is realized in nature , then @xmath485 must be quite far from @xmath486 . plots with the values of the three lepton mixing angles in the standard parameterisation , for all possible mixing patterns involving genuine three - flavour mixing ( including permutations ) . for each case pertaining to the sporadic mixing matrices , a dot has been placed in these plots ( some of them are superimposed ) .
the three lines ( red , blue , green ) correspond to different column permutations of the series of patterns obtained from equation by varying the root of unity @xmath249 .
the lower plots are magnifications of the ones above in the physically interesting region , with the numbers on the red curves indicating values of @xmath487 .
the gray regions mark the 1 , 2 and 3 sigma ranges of the mixing angles calculated by forero _ et al .
_ , arxiv:1405.7540 in @xcite for the normal mass hierarchy .
] for completeness , we have also presented in appendix [ 2f ] the solutions where @xmath1 is block - diagonal , _
i.e. _ one flavour does not mix with the other two . as a byproduct
, our analysis has yielded all instances where one row of @xmath148 is fixed .
is determined .
] actually , if one takes only the cases without a zero in the row , then there are only two cases , @xmath488 and @xmath489 , which have for instance been found earlier in @xcite with the help of gap .
the result of our analysis is a bit sobering .
taking up the position that residual symmetries should be capable of reproducing the results of fitting @xmath1 to the data @xcite , there is only the series @xmath247 which can do the job ; some mixing matrices of this series were discovered earlier in @xcite by using gap . for the sporadic mixing matrices ,
the main obstacle is to reproduce the small quantity @xmath490 obtained from the oscillation data . on the one hand ,
some sporadic mixing matrices @xmath148 have a zero entry , which is ruled out as it would imply @xmath491 . on the other hand , the smallest non - zero entry of all sporadic @xmath148
is @xmath492 occurring in @xmath287 , which is significantly larger than the physical @xmath464 .
maybe the result of this paper gives more credibility to models where mixing angles and phases are related to mass ratios .
such models can be based on abelian and non - abelian flavour groups .
texture zeros are one possibility
these are , in effect , practically synonymous with abelian symmetries @xcite .
one might also try to relax the assumptions of our analysis , for example allowing one neutrino mass to be zero @xcite , adopting the idea that neutrinos are dirac particles , will in general not be the klein four - group .
however , in this is nontheless the case
see @xcite , then the analysis of this paper still applies . ] or giving up the _ ad hoc _ assumption of a finite flavour group @xmath2 . in particular , in the latter case a new mathematical approach would be necessary .
it is also worth mentioning that since neither of the residual symmetries ( @xmath12 and @xmath14 ) are symmetries of the full lagrangian , in concrete models there will be radiative corrections to the mixing patterns presented here .
* acknowledgments : * w.g . is very grateful to christoph baxa for an illuminating discussion about roots of unity and to patrick o. ludl for constant support and valuable help with group theory and gap .
we also thank mariam trtola for providing the up - to - date ranges of the mixing angles in figure 1 , which include data presented at the neutrino 2014 conference .
the work of r.f . was supported by the spanish _ ministerio de economa y competitividad _ through the grants fpa2011 - 22975 and multidark consolider csd2009 - 00064 , by the _ generalitat valenciana _ through the grant prometeo/2009/091 , and by the portuguese _ fundao para a cincia e a tecnologia _ through the grants cern / fp/123580/2011 and expl / fis - nuc/0460/2013 .
the relations in equation ( [ lambda ] ) have the generic form @xmath493 with roots of unity @xmath117 , @xmath118 , @xmath119 .
this equation can be conceived as a vanishing sum over eight roots of unity such that the trivial root , 1 , occurs in the sum and also the complex conjugate of every root .
a suitable theorem to deal with such a vanishing sum is theorem 6 of @xcite , which we have reproduced in this work as theorem [ tcj ] in section [ tools ] .
this mathematical result deals with all vanishing sums of at most 9 roots of unity , which we have labeled from a ) to h ) . in each sum , all roots are different .
let us now go through each of them .
sums g ) and h ) have each 9 roots , so these sums are irrelevant for a solution of equation ( [ lambda1 ] ) . on the other hand , the sums e ) and f ) have 8 distinct roots of unity each , but in equation ( [ lambda1 ] ) the trivial root 1 occurs twice , so sums e ) and f ) are also irrelevant for our purpose .
sums c ) and d ) have 7 roots each , therefore , adding a further root of unity does not lead to a vanishing sum .
sum b ) has six roots , thus with this vanishing sum we can build a vanishing sum of eight roots of the form @xmath494 where @xmath495 , @xmath293 , and @xmath496 and @xmath497 are further roots of unity yet to be specified .
the trivial root 1 has to occur in both partial sums in equation ( [ s1 ] ) .
therefore , without loss of generality , @xmath498 .
let us first assume @xmath499 .
then equation ( [ s1 ] ) gives @xmath500 the root @xmath501 should occur two times in this sum , but this is not the case and we can discard @xmath499
. we can deal similarly with @xmath502 .
next we assume @xmath503 , which leads to @xmath504 again , @xmath501 should occur twice , but it does not .
analogously , we can exclude all other possibilities like @xmath505 and so on . in summary
, we have found that sum b ) of conway and jones can not be part of the solution of equation ( [ lambda1 ] ) .
with sum a ) we have the vanishing sum of eight roots of unity @xmath506 since 1 must occur in both partial sums , we can choose without loss of generality @xmath507 .
indeed , this represents a solution of equation ( [ lambda1 ] ) with @xmath508 , @xmath509 , @xmath510 . according to the theorem of conway and jones , the remaining solutions of equation ( [ lambda1 ] )
can only contain powers of @xmath183 and @xmath501 .
one possibility is @xmath511 suppose the trivial root 1 occurs in each of the partial sums with @xmath183 .
then , without loss of generality @xmath512 .
this implies @xmath513 or @xmath514 .
thus we have a second solution of equation ( [ lambda1 ] ) : @xmath515 , @xmath516 . on the other hand
, we can also assume that the trivial root 1 occurs in the first and the third partial sum of equation ( [ s3 ] ) .
without loss of generality , this is achieved with @xmath517 . since , in this case
, @xmath501 occurs in the third partial sum , it must occur in the second partial sum too . without loss of generality , this implies @xmath518 , leading to another solution @xmath519 , @xmath520 , @xmath521 .
the last sum that needs to be discussed is of the form @xmath522 since the trivial root has to occur twice , we choose , without loss of generality , @xmath512 .
then @xmath523 with an undetermined root @xmath524 .
thus we have found another solution of equation ( [ lambda1 ] ) : @xmath519 , @xmath525 , @xmath526 .
the solution of the previous paragraph is special case of this one . in summary ,
equation ( [ lambda1 ] ) has only three solutions , which are displayed in equation ( [ abc ] ) .
here we determine all possible forms of the matrix @xmath528 .
as we will see , it lies in the nature of the method that we can determine @xmath122 only up to permutations , _
i.e. _ wherever we find a @xmath122 , then also the matrices @xmath529 , where @xmath228 and @xmath229 are @xmath131 permutation matrices , are admissible . in the present section ,
whenever we use `` unique '' for @xmath122 , it is always meant unique up to such permutations . according to equations ( [ t - lambda ] ) and ( [ abc ] ) , up to permutations
, every line and column of @xmath122 is identical to one of the three following possibilities : @xmath530 note that @xmath531 the angle @xmath127 in ( c ) is an arbitrary rational angle , comes from a root of unity , it must be @xmath532 times some rational number , but otherwise it is arbitrary . ]
but if we combine for instance a column of type ( c ) with a line of type ( a ) or ( b ) , then @xmath533 can assume only four values : @xmath534 later we will see in one instance that also the case @xmath535 occurs . in the following we derive the possible forms of @xmath122 , by exhausting all possible combinations of ( a ) , ( b ) and ( c ) .
to find all viable cases , we must take into account that @xmath122 stems from a unitary matrix @xmath10 .
firstly , an admissible @xmath122 must fulfill @xmath536 secondly , for such a @xmath122 , one must also check that it is possible to find phases @xmath537 such that the scheme of complex numbers @xmath538 fulfills the orthogonality relations of a unitary matrix .
it was shown in @xcite that for this purpose it is sufficient to check the validity of the inequality @xmath539 assuming the ordering of the first line as in equation ( [ abct ] ) , the first column can be either of type ( a ) or ( c@xmath540 ) .
[ [ first - column - in - t - of - type - a ] ] first column in @xmath122 of type ( a ) : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + then the second line can be of type ( a ) or ( b ) or ( c@xmath541 ) . if it is of type ( a ) , we obtain @xmath542 this @xmath122 is compatible with inequality ( [ ineq ] ) . here and in the following the capital letters to the right of the matrix
indicate the types of the lines .
if the second line is of type ( b ) , we arrive at @xmath543 this case is also compatible with inequality ( [ ineq ] ) . finally ,
if the second line is of type ( c@xmath541 ) , we find @xmath544 this case is _ not _ in agreement with inequality ( [ ineq ] ) .
[ [ first - column - in - t - of - type - c_1 ] ] first column in @xmath122 of type ( c@xmath540 ) : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let us assume @xmath545 and @xmath546 .
then the second line can be of type ( a ) or ( c@xmath540 ) .
the first possibility gives , upon reordering of the lines , @xmath547 this is a viable case with respect to inequality ( [ ineq ] ) .
if we have type ( c@xmath540 ) in the second line , then the third line must be ( c@xmath541 ) : @xmath548 however , this case does not comply with inequality ( [ ineq ] ) .
now we have exhausted all possibilities with one or more lines of type ( a ) . assuming the ordering of ( b ) as in equation ( [ abct ] ) , the first column can be either of type ( a ) or ( b ) or ( c@xmath541 ) . [
[ first - column - of - type - a ] ] first column of type ( a ) : + + + + + + + + + + + + + + + + + + + + + + + + + with the ordering of the first column as @xmath549 , the third line must be either of type ( a ) or of type ( c@xmath540 ) .
if it is of type ( a ) , then the second line must be of type ( b ) and we are lead to a permutation of ( abb ) of equation ( [ abb ] ) .
if it is of type ( c@xmath540 ) , then the column which has the zero must be of type ( c@xmath550 ) or ( c@xmath551 ) .
but then the second line can not be one of the types ( a ) , ( b ) , ( c ) .
so we come to the conclusion that , if the first column is of type ( a ) , no new case ensues . [ [ first - column - of - type - b ] ] first column of type ( b ) : + + + + + + + + + + + + + + + + + + + + + + + + + with @xmath552 , the second line can be ( b ) or ( c@xmath551 ) . in the first case we are lead to the viable matrix @xmath553
if the second line is of type ( c@xmath551 ) , the only way to fulfill equation ( [ 1 ] ) is @xmath554 however , this @xmath122 is not admissible because inequality ( [ ineq ] ) is not fulfilled .
[ [ first - column - of - type - c_2 ] ] first column of type ( c@xmath541 ) : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + assuming @xmath555 , then the second line must be of type ( c@xmath541 ) . in this case
we find @xmath556 and @xmath557 however , both possibilities must be discarded because inequality ( [ ineq ] ) is not fulfilled .
now we have exhausted all cases with lines of type ( a ) and ( b ) .
it remains to discuss matrices @xmath122 where _ all _ lines are of type ( c ) .
let us assume that the first line is of type ( c@xmath558 ) .
then we find @xmath559 all @xmath533 are in agreement with inequality ( [ ineq ] ) .
if no line of type ( c@xmath558 ) occurs in @xmath122 , _
@xmath560 , then one quickly finds that up to permutations the unique possibility is @xmath561 however , application of inequality ( [ ineq ] ) gives @xmath562 , which we had already excluded . in summary ,
equation ( [ abct ] ) , which we had derived using the theorem of conway and jones , leads to 11 matrices @xmath122 which comply with equation ( [ 1 ] ) .
however , checking these matrices against inequality ( [ ineq ] ) , we are left with only five viable cases : @xmath563 in the latter case , the angle @xmath127 is not restricted , apart from being a rational angle .
according to the procedure explained in the introduction to section [ external ] , we first compute @xmath564 since @xmath565 and having excluded the basic forms 1 and 4 , we know that @xmath566 has to be one of the basic forms 2 , 3 or 5 .
the @xmath566 above follows the pattern @xmath567 since @xmath568 occurs four times in @xmath566 , it follows from equation ( [ basic ] ) that @xmath569 , @xmath570 or @xmath75 , corresponding to @xmath571 respectively . in the latter case one finds @xmath572 , so there could be one , three or five elements @xmath570 in @xmath566 ; however , a glance at equation ( [ basic ] ) reveals that this is impossible , because @xmath570 appears repeated an even number of times in the basic forms 2 , 3 and 5 .
therefore , the viable possibilities are @xmath241 . proceeding in the same way as in the previous subsection , we find @xmath573 and @xmath574 follows the pattern @xmath575 because of @xmath576 there are , together with @xmath577 , at least five non - zero elements in @xmath574 . moreover , of the remaining four elements , not all can be zero at the same time .
thus there are more than five non - zero elements in @xmath574 , which excludes the basic form 5 .
thus only the basic forms 2 and 3 with all possible permutations come into question , with the boundary condition @xmath578 .
first of all , this gives @xmath579 secondly , there are eight possible forms of @xmath574 which are @xmath580 considering these matrices , we see that always one of the following two equalities holds : @xmath581 exploiting these equations , we obtain @xmath582 defining @xmath583 , the external phases are given by @xmath584 however , the first relation leads to @xmath585 or @xmath586 , which can not occur in the basic forms 2 and 3 .
this leaves us with @xmath587 finally , we are in a position to determine @xmath198 for all cases of @xmath136 : @xmath588 the phase factors @xmath268 and @xmath269 are defined in equation ( [ phi - rho0 ] ) . of course , for every @xmath198 also @xmath589 is a solution of equation ( [ ax ] ) , but in the light of the discussion in section [ equivalent ] this is a trivial variation .
as before , we consider @xmath230 , which in the present case is given by @xmath590 & \frac{1}{8}\left[-c\left(\kappa_{1}+\kappa_{3}\right)-d\kappa_{2}\right ] \\ \frac{1}{8}\left[c\left(\kappa_{1}+\kappa_{2}\right)+d\kappa_{3}\right ] & \frac{1}{4}\left(-\kappa_{1}+\kappa_{2}-\kappa_{3}\right ) & \frac{1}{8}\left[a\left(\kappa_{2}+\kappa_{3}\right)-b\kappa_{1}\right ] \\ \frac{1}{8}\left[a\left(\kappa_{1}+\kappa_{3}\right)-b\kappa_{2}\right ] & \frac{1}{8}\left[-d\kappa_{1}-c\left(\kappa_{2}+\kappa_{3}\right)\right ] & \frac{1}{4}\left(-\kappa_{1}-\kappa_{2}+\kappa_{3}\right ) \end{array}\right ) \hat\kappa,\end{aligned}\ ] ] where we have defined @xmath591 it is easy to show that none of the entries of @xmath566 can be as large as one , which rules out form 5 .
so it remains to consider forms 2 and 3 .
another observation is that phase factors can not be `` aligned , '' _
i.e. _ @xmath592 ( @xmath593 ) , because otherwise one or more entries at the diagonal of @xmath566 would be @xmath594 which is forbidden for all basic forms except form 5 which we have excluded before .
let us first envisage that @xmath566 is of form 3 .
a specific property of form 3 is that either all elements on the diagonal are the same or all three elements are different .
the latter case is impossible because @xmath595 , while one element on the diagonal would have to be @xmath76 which is larger than @xmath596 . since we know that all elements are the same on the diagonal , we deduce @xmath597
thus we conclude that either @xmath598 or @xmath599 . in the first case we would have alignment of phase factors , which we have excluded before .
therefore , equation ( [ kappa ] ) leads to the relations @xmath600 these relations are solved by @xmath601 with @xmath602 , since @xmath603 would lead to alignment of phase factors .
since @xmath250 is an arbitrary root of unity and complex conjugation of @xmath10 leads to a trivial variation , we obtain without loss of generality @xmath604 when @xmath566 is of form 3 .
all elements on the diagonal of @xmath566 are @xmath75 in this case .
next we discuss the possibility that @xmath566 is of form 2 , in which case there must be one zero in the matrix . because of @xmath605 , we have @xmath606 we further observe that @xmath607 which is impossible because no such element occurs in form 2 .
we conclude that the zero must be on the diagonal of @xmath230 : * @xmath608 @xmath609 @xmath610 @xmath609 @xmath611 .
this implies @xmath612 or @xmath613 , so we end up with @xmath614 or @xmath615 .
* @xmath616 @xmath609 @xmath617 @xmath609 @xmath618 .
this implies @xmath619 or @xmath613 , so we end up with @xmath620 or @xmath621 .
* @xmath622 @xmath609 @xmath623 @xmath609 @xmath624 .
this implies @xmath625 or @xmath613 , so we end up with @xmath626 or @xmath627 .
leaving out cases which emerge from each other by complex conjugation , we end up with the three solutions @xmath628 for the external phases , where @xmath251 and @xmath270 are arbitrary roots of unity .
the matrix @xmath629 is as follows : @xmath630 where @xmath631 , @xmath632 , @xmath74 , @xmath633 are defined in equation ( [ abcd ] ) .
the most obvious observation is that @xmath574 is symmetric .
if @xmath566 is if form 2 which has one zero , this zero can only be on the diagonal . however , @xmath634 rules out this possibility .
so it remains to consider @xmath566 being of form 3 and of form 5 .
if @xmath566 is of form 3 , then the elements on the diagonal are either all the same or all different .
since the first option leads to a non - symmetric matrix , we are left with the second option , namely having @xmath75 , @xmath635 and @xmath636 on the diagonal .
taking into account that @xmath637 we have three cases : * @xmath638 @xmath609 @xmath639 , * @xmath640 @xmath609 @xmath641 , * @xmath642 @xmath609 @xmath643 . if @xmath566 is of form 5 , we observe that @xmath644 therefore , the 1 has to be somewhere on the diagonal .
since @xmath645 holds , the phase factors necessarily fulfill @xmath646 , leading to @xmath647 , consistent with @xmath648 for the off - diagonal elements . in summary ,
the external phases are given by @xmath649 this case is quite similar to form 3a . with @xmath650 & \frac{1}{8}\left[c\left(\kappa_{2}-\kappa_{1}\right)-d\kappa_{3}\right ] & \frac{1}{4}\left(\kappa_{1}+\kappa_{2}+\kappa_{3}\right)\\ \frac{1}{4}\left(-\kappa_{1}+\kappa_{2}+\kappa_{3}\right ) & \frac{1}{8}\left[b\kappa_{2}+a\left(\kappa_{1}+\kappa_{3}\right)\right ] & \frac{1}{8}\left[c\left(\kappa_{2}-\kappa_{3}\right)-d\kappa_{1}\right]\\ \frac{1}{8}\left[-c\left(\kappa_{1}+\kappa_{3}\right)+d\kappa_{2}\right ] & \frac{1}{4}\left(\kappa_{1}+\kappa_{2}-\kappa_{3}\right ) & \frac{1}{8}\left[b\kappa_{3}+a\left(\kappa_{2}-\kappa_{1}\right)\right ] \end{array}\right ) \hat\kappa\end{aligned}\ ] ] and @xmath631 , @xmath632 , @xmath74 , @xmath633 defined in equation ( [ abcd ] ) , we find that none of its entries can be as large as 1 , which rules out form 5 .
furthermore , the phase factors @xmath208 can not be aligned , lest one or more entries of @xmath566 be @xmath594 .
first we consider form 3 .
the entries @xmath651 are either all the same or all different
. however , the latter case can not occur , because these entries are all smaller or equal @xmath596 , which is smaller than @xmath635 .
evaluating the requirement that the elements of equation ( [ ee ] ) have to be equal , we proceed along the same lines as for form 3a and obtain the relations @xmath652 from this we arrive at the external phases @xmath653 it remains to consider form 2 , which has one zero in some place .
the zero can not be on the diagonal of @xmath566 because @xmath654 .
moreover , proceeding as in equation ( [ if ] ) , we find * @xmath655 @xmath609 @xmath656 , * @xmath657 @xmath609 @xmath658 , * @xmath659 @xmath609 @xmath660 , which is impossible for form 2 .
thus we have the following three possibilities : * @xmath661 @xmath609 @xmath610 @xmath609 @xmath611 .
this implies @xmath612 or @xmath613 , so we end up with @xmath614 or @xmath615 . *
@xmath662 @xmath609 @xmath663 @xmath609 @xmath618 .
this implies @xmath619 or @xmath613 , so we end up with @xmath664 or @xmath665 . *
@xmath666 @xmath609 @xmath623 @xmath609 @xmath624 .
this implies @xmath625 or @xmath613 , so we end up with @xmath626 or @xmath627 . leaving out the complex conjugate solutions , we finally obtain @xmath667 for the external phases , with @xmath251 and @xmath270 being arbitrary roots of unity .
also this case is treated analogously to form 3a . with @xmath668 & \frac{1}{4}\left(\kappa_{1}-\kappa_{2}+\kappa_{3}\right ) & \frac{1}{8}\left[a\left(\kappa_{3}-\kappa_{1}\right)-b\kappa_{2}\right]\\ \frac{1}{8}\left[a\left(\kappa_{2}-\kappa_{1}\right)-b\kappa_{3}\right ] & \frac{1}{8}\left[c\left(\kappa_{3}-\kappa_{1}\right)+d\kappa_{2}\right ] & \frac{1}{4}\left(\kappa_{1}+\kappa_{2}+\kappa_{3}\right)\\ \frac{1}{4}\left(\kappa_{1}+\kappa_{2}-\kappa_{3}\right ) & \frac{1}{8}\left[b\kappa_{1}+a\left(\kappa_{2}+\kappa_{3}\right)\right ] & \frac{1}{8}\left[c\left(\kappa_{2}-\kappa_{1}\right)+d\kappa_{3}\right ] \end{array}\right ) \hat\kappa\end{aligned}\ ] ] and @xmath631 , @xmath632 , @xmath74 , @xmath633 defined in equation ( [ abcd ] ) , we find again that none of its entries can be as large as one , thus ruling out form 5 .
furthermore , the phase factors @xmath208 can not be aligned in order to avoid one or more entries of @xmath566 being @xmath594 .
first we consider form 3 .
the entries @xmath669 are either all the same or all different
. however , the latter case can not occur , because @xmath596 is an upper bound to these entries and @xmath670 .
evaluating the requirement that the elements of equation ( [ eee ] ) have to be equal , we obtain the relations @xmath671 which give @xmath672 assuming now that @xmath566 is of form 2 , there must be a zero in the matrix .
however , @xmath673 because of @xmath605 .
moreover , it follows that * @xmath608 @xmath609 @xmath674 , * @xmath616 @xmath609 @xmath675 , * @xmath622 @xmath609 @xmath676 , in the same vein as for form 3a in equation ( [ if ] ) , which is impossible because no such element occurs in form 2 .
thus we have to consider the remaining three possibilities : * @xmath655 @xmath609 @xmath677 @xmath609 @xmath678 .
this implies @xmath679 or @xmath613 , so we end up with @xmath680 or @xmath681 . *
@xmath657 @xmath609 @xmath682 @xmath609 @xmath618 .
this implies @xmath619 or @xmath613 , so we end up with @xmath683 or @xmath684 .
* @xmath659 @xmath609 @xmath685 @xmath609 @xmath624 .
this implies @xmath625 or @xmath613 , so we end up with @xmath686 or @xmath687 .
dropping the complex conjugate solutions , we arrive at @xmath688 with @xmath251 and @xmath270 being arbitrary roots of unity .
here we determine minimal flavour groups @xmath2 associated with the series @xmath247 .
the generators of the groups of this series are @xmath689 and @xmath20 ( @xmath33 ) .
we make a basis transformation with @xmath690 leading to @xmath691 and @xmath692 given by @xmath693 for the time being , we set @xmath694 , which entails @xmath695 and @xmath696 . at any rate , @xmath249 is an abitrary root of unity .
it is now useful to switch to another set of generators , as described in appendix f of @xcite . for this purpose
we compute @xmath697 instead of @xmath698 it is certainly admissible to use @xmath699 and @xmath700 as generators .
but then , with a further basis change , one can remove the phases from @xmath701 and @xmath702 : @xmath703 and @xmath704 in this final basis , our group @xmath2 has the generators @xmath353 , @xmath700 , @xmath698 and @xmath568 .
clearly , @xmath698 and @xmath568 are generators of permutations , so @xmath2 has the structure @xcite @xmath705 , where @xmath706 is the normal subgroup consisting of all diagonal matrices and @xmath163 consists of all @xmath131 permutations matrices .
this @xmath706 is generated by @xmath353 , @xmath700 , and matrices where the elements on the diagonal of @xmath353 and @xmath700 are permuted .
we can write @xmath249 as @xmath707 with integers @xmath453 and @xmath69 , and @xmath453 coprime to @xmath69 . if @xmath69 is
even , we define @xmath708 if @xmath69 is odd , we consider @xmath709 instead for the definition of @xmath710 . in this case
we have @xmath711 thus it is useful to define @xmath712 and @xmath713 , where @xmath714 for @xmath69 even and @xmath715 for @xmath69 odd .
note that @xmath713 is coprime to @xmath716 . since @xmath717
we can dispense with the generator @xmath353 . according to theorem 1 in @xcite ,
there is a procedure to find two generators which reveal the structure of @xmath706 and , as a consequence , the structure of @xmath2 .
we first need to find the element with the highest order in @xmath706 .
since @xmath713 is coprime to @xmath716 , there must be a positive integer @xmath631 such that @xmath718 this matrix has the highest possible order in @xmath706 .
in the next step we have to find the generator of the subgroup of @xmath706 which has 1 as the first entry .
it is easy to see that in our case every element in this subgroup is a power of @xmath719 if @xmath720 , there is a positive integer @xmath632 such that @xmath721 and we obviously arrive at the conclusion that the flavour group is @xmath722 . for the remaining cases we use again the results of
if @xmath723 divides @xmath69 , the flavour group is given by @xmath724 .
finally , we have to consider @xmath725 but @xmath456 . in this case we put @xmath726 with @xmath727 .
note that @xmath728 or @xmath729 .
now we proceed as before , but with the replacements @xmath730 in equation ( [ te ] ) and , due to equation ( [ ee ] ) , @xmath731 next we have a closer look at @xmath732 given that @xmath69 is divisible by @xmath733 but not by @xmath723 , we find that @xmath734 is divisible by @xmath733 .
so we end up with @xmath735/3}{n/3 } \right).\ ] ] this should be compared to @xmath707 defined before equation ( [ f1 ] ) . in effect , we replace @xmath69 by @xmath736 and we proceed as before , starting from equation ( [ f1 ] ) .
eventually , we arrive at the group @xmath737 , with @xmath716 being defined above .
we are dealing here with @xmath738 where @xmath57 is a rational angle , _
i.e. _ @xmath350 is a root of unity .
this case clearly involves only two - family mixing and we present it here only for completeness . because it is physically irrelevant
, we do not discuss all the details .
the phase factor @xmath250 and the product @xmath739 must be roots of unity . the only non - trivial part of @xmath10 is the 23-sector , so we focus exclusively on this sector .
there we have the two generators @xmath740 where @xmath46 is the 23-block of the generator @xmath741 of equation ( [ sss ] ) .
we search for possible values of the pair @xmath742 which lead to a finite group .
let us denote the eigenvalues of @xmath743 by @xmath744 , @xmath745 and those of @xmath746 by @xmath747 , @xmath748 .
then we find @xmath749 or @xmath750 taking the square of the absolute values of these relations , we obtain consistency only if @xmath751 is purely imaginary .
finally , we can express @xmath752 and @xmath753 as @xmath754 we can use equation ( [ cos ] ) to constrain the possible values @xmath755 and @xmath756 . using @xmath757 ,
we get the following vanishing sum of roots of unity : @xmath758 this sum is exactly of the form of equation ( [ lambda2 ] ) of section [ basic forms ] , with its solutions given in equation ( [ abc ] ) .
these solutions imply @xmath759 up to conjugations and permutations , where @xmath524 is some arbitrary root of unity . with this
we obtain the seven cases @xmath760 or the reversed order .
note that @xmath761 preserves @xmath762see equation ( [ cos])and @xmath763 is transformed to @xmath764see equation ( [ kappa23 ] ) . while the conjugation of this phase is irrelevant , in general this is not true for the sign reversal .
so , for each of the seven cases in equation ( [ st ] ) , one has to take into account the two signs of @xmath763 . in the following
, we will list the solutions for each of the seven cases in equation ( [ st ] ) and the two sign variations of @xmath763 , without going into detail .
for each of the seven cases we will display 1 .
@xmath752 and @xmath763 , 2 .
@xmath44 and the eigenvalues of @xmath10 for the positive sign , 3 .
@xmath148 for the positive sign , 4 .
the degenerate case of @xmath148 for the positive sign , 5 .
@xmath44 and the eigenvalues of @xmath10 for the negative sign , 6 .
@xmath148 for the negative sign , 7 .
the degenerate case of @xmath148 for the negative sign .
@xmath773 + this corresponds to @xmath774 this means that we can write the phases as @xmath775 , @xmath776 with a root of unity @xmath248 .
+ _ positive sign , non - degenerate case : _ @xmath777 @xmath778 _ positive sign , degenerate case : _ @xmath771 @xmath779 _ negative sign , non - degenerate case : _ @xmath780 @xmath781 _ negative sign , degenerate case : _
@xmath782 @xmath783 @xmath784 + this corresponds to @xmath785 _ positive sign , non - degenerate case : _ @xmath786 @xmath787 _ positive sign , degenerate case : _ @xmath788 @xmath789 _ negative sign , non - degenerate case : _ @xmath790 @xmath791 _ negative sign , degenerate case : _
@xmath771 @xmath792 @xmath793 + this corresponds to @xmath794 _ positive sign , non - degenerate case : _ @xmath795 @xmath796 _ positive sign , degenerate case : _ @xmath771 @xmath797 _ negative sign , non - degenerate case : _ @xmath798 @xmath799 _ negative sign , degenerate case : _
@xmath800 @xmath801 @xmath802 + this corresponds to @xmath803 _ positive sign , non - degenerate case : _ @xmath804 @xmath805 _ positive sign , degenerate case : _ @xmath806 @xmath807 _ negative sign , non - degenerate case : _ @xmath808 @xmath809 _ negative sign , degenerate case : _ @xmath800 @xmath810 @xmath811 + with @xmath121 unrestricted , except for being a root of unity ,
this corresponds to @xmath812 _ positive sign , non - degenerate case : _ @xmath813 @xmath814 _ positive sign , degenerate cases : _ @xmath815 @xmath816 @xmath817 @xmath818 _ negative sign , non - degenerate case : _ @xmath819 @xmath820 _ negative sign , degenerate case : _ @xmath821
@xmath822 at this point a note is in order .
the only restriction on the angle @xmath57 in @xmath823 is that it must be a rational multiple of @xmath128 , _
@xmath824 must be a root of unity .
it is therefore natural to ask whether or not the @xmath148 in @xmath345@xmath825 and @xmath346 are particular cases of @xmath823 .
it turns out that the mixing matrices associated with @xmath826 , @xmath825 and @xmath346 are indeed particular cases of the one of @xmath823 , with @xmath827 , @xmath828 and @xmath829 , respectively . for the remaining cases ,
theorem [ roots1 ] can be used to show that the associated angles @xmath57 are not rational .
@xmath830 + this corresponds to @xmath831 since @xmath524 is a generic root of unity , the sign is irrelevant here and @xmath250 , @xmath251 , @xmath270 are different roots of unity leading to @xmath832 the degenerate case has already been treated in @xmath833 .
p. ramond , _ group theory : a physicist s survey _ , ( cambridge university press , cambridge , uk , 2010 ) ; + g. altarelli and f. feruglio , _ discrete flavor symmetries and models of neutrino mixing _
* 82 * ( 2010 ) 2701 [ arxiv:1002.0211 [ hep - ph ] ] ; + h. ishimori , t. kobayashi , h. ohki , y. shimizu , h. okada and m. tanimoto , _ non - abelian discrete symmetries in particle physics _ , prog .
* 183 * ( 2010 ) 1 [ arxiv:1003.3552 [ hep - th ] ] ; + s. f. king and c. luhn , _ neutrino mass and mixing with discrete symmetry _ , rept .
* 76 * ( 2013 ) 056201 [ arxiv:1301.1340 [ hep - ph ] ] . c. s. lam , _ determining horizontal symmetry from neutrino mixing _ , phys .
* 101 * ( 2008 ) 121602 [ arxiv:0804.2622 [ hep - ph ] ] ; + c. s. lam , _ the unique horizontal symmetry of leptons _ , phys .
d * 78 * ( 2008 ) 073015 [ arxiv:0809.1185 [ hep - ph ] ] ; + w. grimus , l. lavoura and p. o. ludl , _
is @xmath449 the horizontal symmetry of tri - bimaximal lepton mixing ? _ , j. phys .
g * 36 * ( 2009 ) 115007 [ arxiv:0906.2689 [ hep - ph ] ] ; + c. s. lam , _ a bottom up analysis of horizontal symmetry _ , arxiv:0907.2206 [ hep - ph ] ; + s .- f .
ge , d. a. dicus , and w. w. repko , _
@xmath834 symmetry prediction for the leptonic dirac @xmath835 phase _ , phys .
b * 702 * ( 2011 ) 220 [ arxiv:1104.0602 [ hep - ph ] ] ; + h .- j .
he and f .- r .
yin , _ common origin of @xmath836@xmath837 and @xmath835 breaking in neutrino seesaw , baryon asymmetry , and hidden flavor symmetry _ , phys .
d * 84 * ( 2011 ) 033009 [ arxiv:1104.2654 [ hep - ph ] ] ; + s .- f .
ge , d. a. dicus , and w. w. repko , _ residual symmetries for neutrino mixing with a large @xmath838 and nearly maximal @xmath839 _ , phys .
* 108 * ( 2012 ) 041801 [ arxiv:1108.0964 [ hep - ph ] ] ; + h .- j .
he and x .- j .
xu , _ octahedral symmetry with geometrical breaking : new prediction for neutrino mixing angle @xmath838 and @xmath835 violation _ ,
d * 86 * ( 2012 ) 111301 ( r ) [ arxiv:1203.2908 [ hep - ph ] ] ; + d. hernandez and a. yu .
smirnov , _ lepton mixing and discrete symmetries _
d * 86 * ( 2012 ) 053014 [ arxiv:1204.0445 [ hep - ph ] ] ; + c. s. lam , _ finite symmetry of leptonic mass matrices _ , phys . rev .
d * 87 * ( 2013 ) 013001 [ arxiv:1208.5527 [ hep - ph ] ] ; + d. hernandez and a. yu . smirnov , _ discrete symmetries and model - independent patterns of lepton mixing _
d * 87 * ( 2013 ) 053005 [ arxiv:1212.2149 [ hep - ph ] ] ; + b. hu , _ neutrino mixing and discrete symmetries _ , phys . rev .
d * 87 * ( 2013 ) 033002 [ arxiv:1212.2819 [ hep - ph ] ] ; + d. hernandez and a. yu . smirnov , _ relating neutrino masses and mixings by discrete symmetries _ ,
d * 88 * ( 2013 ) 093007 [ arxiv:1304.7738 [ hep - ph ] ] .
m. holthausen and k. s. lim , _ quark and leptonic mixing patterns from the breakdown of a common discrete flavor symmetry _ , phys .
d * 88 * ( 2013 ) 033018 [ arxiv:1306.4356 [ hep - ph ] ] ; + t. araki , h. ishida , h. ishimori , t. kobayashi and a. ogasahara , _ ckm matrix and flavor symmetries _
d * 88 * ( 2013 ) 096002 [ arxiv:1309.4217 [ hep - ph ] ] .
d. v. forero , m. trtola , and j. w. f. valle , _ global status of neutrino oscillation parameters after recent reactor measurements _ , phys .
d * 86 * ( 2012 ) 073012 [ arxiv:1205.4018 [ hep - ph ] ] ; + g. l. fogli , e. lisi , a. marrone , d. montanino , a. palazzo , and a. m. rotunno , _ global analysis of neutrino masses , mixings and phases : entering the era of leptonic @xmath835 violation searches _ , phys .
d * 86 * ( 2012 ) 013012 [ arxiv:1205.5254 [ hep - ph ] ] ; + m. c. gonzalez - garcia , m. maltoni , j. salvado , and t. schwetz , _ global fit to three neutrino mixing : critical look at present precision _
, j. high energy phys .
* 1212 * ( 2012 ) 123 [ arxiv:1209.3023 [ hep - ph ] ] ; +
d. v. forero , m. trtola , and j. w. f. valle , _
neutrino oscillations refitted _ , arxiv:1405.7540 [ hep - ph ] . | assuming that neutrinos are majorana particles , we perform a complete classification of all possible mixing matrices which are fully determined by residual symmetries in the charged - lepton and neutrino mass matrices .
the classification is based on the assumption that the residual symmetries originate from a _
finite _ flavour symmetry group .
the mathematical tools which allow us to accomplish this classification are theorems on sums of roots of unity .
we find @xmath0 sporadic cases plus one infinite series of mixing matrices associated with three - flavour mixing , all of which have already been discussed in the literature .
only the infinite series contains mixing matrices which are compatible with the data at the 3 sigma level . |
twistor theory , created by penrose @xcite , establishes a close relationship between conformal riemannian geometry in dimension 4 , and ( almost ) complex geometry in dimension 3 .
in particular , to a riemannian manifold @xmath1 for which the part @xmath2 of the weyl tensor vanishes identically ( _ self - dual _ ) , one associates its _ twistor space _ @xmath3 , a complex 3-manifold containing rational curves with normal bundle @xmath4 , and admitting a real structure with no fixed points @xcite , @xcite , @xcite . the space of such curves is a complex 4-manifold @xmath5 @xcite with a holomorphic conformal structure and is , therefore , a conformal complexification of @xmath1 @xcite , @xcite , @xcite . as the conformal geometry of @xmath1
is encoded by the complex geometry of @xmath3 , we ask ourselves what holomorphic object on @xmath3 corresponds to @xmath6 , the weyl tensor of the self - dual manifold @xmath1 .
it seems that this question , although natural , has not been considered in the literature , and maybe a reason for that is that the answer appears to be a highly non - linear object .
this object is more easily understood in the framework of complex - riemannian geometry ( see section [ s.w2 ] ) : following lebrun @xcite , we ( locally ) introduce the space @xmath7 of complex null - geodesics of * m * ( _ ambitwistor space _ ) . for a self - dual ( complex ) 4-manifold * m * , its ( local ) twistor space
is then defined as the 3-manifold of @xmath0-surfaces ( some totally geodesic isotropic surfaces , see section [ s.w2 ] ) . the ambitwistor space @xmath7 and ( in the self - dual case )
the twistor space @xmath3 completely describe the conformal structure of * m*. in particular , a null - geodesic @xmath8 in * m * corresponds to the set of rational curves in @xmath3 tangent to a 2-plane .
the union of these curves , called the _
integral @xmath9-cone _ of @xmath8 ( see section [ s.w3 ] ) , is lifted to a ( linearized ) @xmath9-cone in @xmath10 .
our first result ( theorem [ t1 ] ) is that the weyl tensor of * m * is equivalent to the projective curvature ( see section [ s.w4 ] ) of the field of @xmath9-cones on @xmath7 .
in particular , if such a cone is _ flat _ , then @xmath6 vanishes on certain isotropic planes in * m*. we use theorem [ t1 ] to investigate global properties of a self - dual manifold * m * : if the integral @xmath9-cone of @xmath8 is part of a smooth surface in @xmath3 , then the linearized @xmath9-cone is flat ( theorems [ twist ] , [ 3 ] ) . in particular
, the space @xmath11 of rational curves of @xmath3 with normal bundle @xmath4 is compact iff @xmath12 . on the other hand ,
it is known , from a theorem of campana @xcite , that , for a compact twistor space @xmath3 , @xmath11 can be compactified within the space of analytic cycles iff @xmath3 is moishezon .
it appears then that the conformal structure does not extend smoothly to the compactification . a good illustration of what happens in the non - flat ( self - dual ) case is the khler - einstein manifold @xmath13 whose twistor space is known to be the manifold of flags in @xmath14 @xcite , see section [ s.w8 ] .
different methods allow us to generalize theorem 3 to non geodesically - connected self - dual manifolds : we show ( theorem [ compact ] ) that if a self - dual manifold admits a compact , simply - connected , null - geodesic , then it is conformally flat .
we also note that the rational curves in @xmath3 , corresponding to the points of * m * ( see section [ s.w2 ] ) are then geodesics of some _ projective structure _ of @xmath3 iff the latter is projectively flat ( corollary [ twfl ] ) .
the isotropic , totally geodesic surfaces ( called @xmath0-surfaces ) in a self - dual manifold * m * appear to have a projective structure , given by the null - geodesics of * m * contained in it ( section [ s.w6 ] ) .
we show that it is _ flat _
( i.e. locally equivalent to @xmath13 ) ( corollary [ prflat ] ) , and we obtain a classification of the compact @xmath0-surfaces of a self - dual 4-manifold ( theorem [ clasb ] ) . theorem [ compact ] can be adapted for conformal 3-manifolds : a conformal 3-manifold admitting a rational curve as a null - geodesic is conformally flat ( theorem [ comp3 ] ) .
the conformal geometry in dimensions 3 and 4 are related , as any geodesically convex 3-manifold @xmath15 can be realized as the _ conformal infinity _ of a self - dual 4-manifold * m * @xcite .
in particular , @xmath15 is umbilic in * m * , and we relate , in section [ s.w7 ] , the conformal invariants of the 2 manifolds : the cotton - york tensor of @xmath15 is identified to the derivative , in the normal direction , of the weyl tensor of * m * ( theorem [ umbilic ] ) .
this result can be equally stated in the real framework .
the paper is organized as follows : in section [ s.w2 ] we recall the classical results of the twistor theory ( especially for complex 4-manifolds ) , in section [ s.w3 ] we introduce the @xmath9-cones on the ( ambi-)twistor space , and , in section [ s.w4 ] , we prove the equivalence between the projective curvature of the latter and the weyl tensor @xmath6 of * m*. section [ s.w5 ] is devoted to the proof of some results of the type `` compactness implies conformal ( projective ) flatness '' : theorems [ twist ] , [ 3 ] and [ compact ] , mentioned above .
we study the projective structure of @xmath0-surfaces in section [ s.w7 ] , and we illustrate the above results on the special case of the self - dual manifold @xmath13 , in section [ s.w8 ] .
acknowledgements the author is deeply indebted to paul gauduchon , for his care in reading the manuscript and for his constant help during the research and redaction .
the content of this paper makes use of _ complex - riemannian geometry _ ( with the exception of theorem [ umbilic ] and corollary [ cumb ] , which hold also in the real framework ) .
complex - riemannian geometry is obtained from riemannian geometry by replacing the field @xmath16 by @xmath17 ( e.g. a complex `` metric '' becomes a non - degenerate symmetric complex - bilinear form on the tangent space ) , and all classical results hold , naturally with the exception of those making use of partitions of unity .
we will often omit the prefix `` complex- '' , when referring to geometric objects , and we will always consider them , unless otherwise stated , in the framework of complex - riemannian geometry .
let * m * be a 4-dimensional complex manifold .
conformal structure _ is defined , as in the real case @xcite , by a everywhere non - degenerate section @xmath18 of the complex bundle @xmath19 , where @xmath20 is a given line bundle of _ scalars of weight 1 _ , and @xmath21 , the anti - canonical bundle of * m*. ( while on an oriented real manifold such a line bundle always exists , being topologically trivial , in the complex case the existence of @xmath22 , a square root of the anti - canonical bundle , is submitted to some topological restrictions . ) from now on , only holomorphic conformal structures will be considered , thus @xmath20 is a holomorphic bundle and @xmath18 a holomorphic section of @xmath19 .
( in fact , all we need to define the conformal structure @xmath18 on the 4-manifold * m * is just the holomorphic bundle @xmath22 ; in odd dimensions the situation is different , see section [ s.w7 ] . ) as in the real case , @xmath18 is locally represented by symmetric bilinear forms on @xmath23 , but global representative metrics do not exist , in general . for each point @xmath24
, there is an isotropy cone @xmath25 in the tangent space @xmath26 , who uniquely determines the conformal structure @xmath18 . in the associated projective space , @xmath27 , the cone @xmath25 projects onto the non - degenerate quadratic surface @xmath28 , which is actually a ruled surface isomorphic to @xmath29 .
we thus get 2 families of complex projective lines contained in @xmath30 , that is , 2 families of isotropic 2-planes in @xmath31 , respectively called @xmath9-planes , the other @xmath0-planes .
this choice corresponds to the choice of an `` orientation '' of * m*. on a real 4-manifold an orientation is chosen by picking a class of volume forms ( which is not possible in this complex framework ) or by choosing one of the two possible hodge operators compatible with the conformal structure @xmath32 ( which can also be done in our complex case , @xcite ) . as @xmath33 is a symmetric involution
, @xmath34 decomposes in @xmath35 consisting in @xmath36-eigenvectors of @xmath33 , respectively called _ self - dual _ and _ anti - self - dual _ 2-forms ; the isotropic vectors in @xmath37 and @xmath38 are then exactly the _ decomposable _ elements @xmath39 , with @xmath40 .
an @xmath9-plane @xmath41 ( resp . a @xmath0-plane @xmath42 in @xmath23 is a 2-plane such that @xmath43 ( resp .
@xmath44 ) is a is a self - dual ( resp .
anti - self - dual ) isotropic line in @xmath34 .
the @xmath9- and @xmath0-planes can be interpreted in terms of spinors .
the structure group of the tangent bundle @xmath23 is restricted to the conformal orthogonal complex group , @xmath45 , where @xmath46 , by the given conformal structure of * m*. the choice of an orientation is the further restriction of this group to the connected component of the identity , @xmath47 , where @xmath48 .
consider a local metric @xmath49 in the conformal class @xmath50 .
we have then locally defined @xmath51 structures , and associated @xmath51 bundles @xmath52 , as in the real case @xcite,@xcite .
they are rank 2 complex vector bundles , and for each local section of @xmath20 ( i.e. a metric in @xmath18 ) , each of them is equipped with a ( complex ) symplectic structure @xmath53 , respectively .
then we locally have @xmath54 , and @xmath55 , for the fixed metric @xmath56 .
@xmath0- ) planes are then nothing but the isotropic 2-planes obtained by fixing the first ( resp . the second )
factor in the above tensor product : @xcite an @xmath9-plane , resp .
@xmath0-plane @xmath57 is a complex plane @xmath58 , resp @xmath59 , where @xmath60 , resp .
@xmath61 .
the @xmath9-planes in @xmath26 are , thus indexed by @xmath62 , and @xmath0-planes by @xmath63 , and these projective bundles are globally well - defined on * m * , @xcite .
it is obvious that a change of orientation interchanges the @xmath9 and @xmath0-planes ; the same is true for self - duality and anti - self - duality , to be defined below . for a local metric @xmath49 in @xmath18 ,
we denote by @xmath64 its riemannian curvature , and by @xmath65 the weyl tensor , i.e. the trace - free component of @xmath64 , which is known to be independent of the chosen metric within the conformal class @xcite .
it splits into two components @xmath66 , and the easiest way to see that is the spinorial decomposition of the space of the curvature tensors @xmath67 , @xcite,@xcite,@xcite , obtained from the relation @xmath68 and some of the clebsch - gordan identities @xcite .
@xmath69 where @xmath70 is the complex line of scalar curvature tensors , included in @xmath71 , @xmath72 is the space of trace - free ricci tensors , and @xmath73 , @xmath74 are the spaces of self - dual , resp .
anti - self - dual weyl tensors ( where @xmath75 denotes the @xmath76-symmetric power of @xmath77 ) .
the curvature @xmath64 restricted to any @xmath9-plane @xmath78 yields a weighted bilinear symmetric form @xmath79 on @xmath80 , i.e. a section in @xmath81 : @xmath82 the ( weighted ) bilinear form @xmath79 depends only on the self - dual weyl tensor , and this one is completely determined by the ( weighted ) values of @xmath79 for all @xmath9-planes @xmath78 .
we have the same result for @xmath0-planes .
let @xmath83 be an @xmath9-plane , and let @xmath84 , and suppose , for simplicity , that @xmath85 , so @xmath86 is identified to the element @xmath87 .
then it is easy to see that @xmath79 , evaluated on @xmath88 , is nothing but the evaluation of @xmath89 on @xmath90 , which depends only on the positive ( or self - dual ) part of the weyl tensor . to prove the second assertion
, we remark that @xmath6 , being a quadrilinear symmetric form on @xmath91 , can be identified with a polynomial of degree 4 on @xmath91 , which is determined by its values .
a conformal structure @xmath18 on a 4-manifold * m * is called self - dual ( resp .
anti - self - dual ) iff @xmath92 ( resp .
@xmath93 ) . in general ,
geodesics on a conformal manifold depend on the chosen metric , with the exception of the isotropic ones ( or _ null - geodesics _ ) . therefore the existence of totally geodesic surfaces tangent to @xmath9- ( resp .
@xmath0- ) planes is a property of the conformal structure alone .
an @xmath9-surface ( resp .
@xmath0-surface ) @xmath94 is a maximal , totally geodesic surface in * m * , whose tangent space in any point is an @xmath9-plane ( resp .
@xmath0-plane ) . on the other hand , any totally geodesic , isotropic surface in * m * is included in an @xmath9- or in a @xmath0-surface .
@xcite,@xcite if , in any point @xmath24 , and for any @xmath9- ( resp .
@xmath0- ) plane @xmath95 , there is a @xmath9- ( resp .
@xmath0- ) surface tangent to @xmath78 at @xmath96 , we say that the family of @xmath9- ( resp .
@xmath0- ) planes is integrable .
@xcite,@xcite the family of @xmath9- ( resp .
@xmath0- ) planes of a conformal 4-manifold @xmath97 integrable if and only if the conformal structure @xmath18 is anti - self - dual ( resp . self - dual ) .
the integrability of @xmath9-planes is equivalent to the integrability ( in the sense of frobenius ) of a distribution @xmath98 of 2-planes on the total space of the projective bundle @xmath99 .
more precisely , let @xmath49 be a local metric in the conformal class @xmath18 , and let @xmath100 be its levi - civita connection .
@xmath100 induces a connection in the bundle @xmath99 , thus a horizontal distribution @xmath101 , isomorphic to @xmath23 via the bundle projection .
let @xmath98 be the 2-dimensional subspace of @xmath102 where @xmath103 is an @xmath9-plane in @xmath26 which projects onto @xmath95 .
it can be easily shown ( as in @xcite , see also @xcite ) that the `` tautological '' 2-plane distribution @xmath98 is independent of the metric @xmath49 .
then @xmath9-surfaces are canonically lifted as integrable manifolds of the distribution @xmath98 .
for a geodesically convex open set of * m * , one can prove ( see @xcite ) that the space of these integrable leaves is a complex 3-manifold .
( this point of view is closely related to the one of @xcite , about the integrability of the canonical almost complex structure of the real twistor space . )
the same remark can be made about @xmath0-surfaces .
the existence , for any point @xmath24 , of an @xmath9-surface containing @xmath96 does not imply , in general , the integrability of the family of @xmath9-planes : in the conformal self - dual ( but not anti - self - dual ) manifold @xmath104 ( the complexification of @xmath13 , @xcite ) , the surfaces @xmath105 and @xmath106 are all @xmath9-surfaces , see section [ s.w8 ] . in the real framework , the twistor space of a real riemannian 4-manifold @xmath107 is the total space @xmath108 of the @xmath109-bundle of almost - complex structures on @xmath110 , compatible with the conformal structure and the ( opposite ) orientation ; it admits a natural almost - complex structure @xmath111 , equal , in @xmath112 , to the complex structure of the fibers on the _ vertical _ space @xmath113 , and to @xmath114 itself on the horizontal space ( induced by the levi - civita connection ) .
such a complex structure @xmath114 is equivalent to an isotropic complex 2-plane in @xmath115 , thus to an @xmath9- or @xmath0-surface ( depending on the conventions ) , which becomes then the space of vectors of type @xmath116 for @xmath114 ; as the integrability of the almost - complex structure @xmath111 can be expressed as the frobenius condition applied to @xmath117 , it is equivalent to the integrability of the family of @xmath9- , resp .
@xmath0-planes .
the _ penrose construction _ associates to an ( anti-)self - dual manifold * m * the space @xmath3 of @xmath9- ( resp .
@xmath0-)surfaces of * m * ; we have seen above that @xmath3 admits complex - analytic maps , but it may be non - hausdorff .
this is why we need to introduce the following condition , see also @xcite : an ( anti- ) self - dual manifold * m * is called civilized iff the space @xmath118 ( resp .
@xmath119 ) of integral leaves of the distribution @xmath98 ( resp .
@xmath120 ) in @xmath99 ( resp .
@xmath121 ) is a complex 3-manifold , and the projection @xmath122 ( resp .
@xmath123 ) is a submersion . in this case ,
the manifold @xmath118 ( resp .
@xmath119 ) which is the space of @xmath9-surfaces ( resp .
@xmath0-surface ) of * m * is called the @xmath9- ( resp .
the @xmath0-)twistor space of * m*. from now on , we suppose that @xmath97 is a self - dual complex analytic 4-manifold .
as any point @xmath24 has a geodesically convex neighborhood @xmath124 @xcite ( which is , therefore , civilized ) , we can construct @xmath125 , the @xmath0-twistor space ( for short , twistor space ) of @xmath124 .
for the infinitesimal results of this paper ( from sections [ s.w3],[s.w4],[s.w6 ] and [ s.w7 ] ) , we will suppose ( with no loss of generality ) that * m * is civilized ( for example , by replacing * m * by @xmath124 ) .
we recall now the correspondence between differential geometric objects on * m * and complex analytic objects on its twistor space , @xmath3 , @xcite,@xcite , see also @xcite,@xcite,@xcite .
@xmath0-surfaces @xmath126 correspond to points @xmath127 , by definition , and the set of @xmath0-surfaces passing through a point @xmath24 is a complex projective line @xmath128 , with normal bundle isomorphic ( non - canonically ) to @xmath129 ( where @xmath130 is the dual of the _ tautological _ bundle @xmath131 on @xmath132 ) @xcite,@xcite , see also @xcite .
in fact , this family of complex projective lines in @xmath3 permits us to recover * m * and its conformal structure , at least locally , by the _ reverse penrose construction _ : the normal bundle @xmath133 of a line @xmath128 in @xmath3 has the property @xmath134 , thus , by a theorem of kodaira @xcite , the space @xmath11 of projective lines in @xmath3 having the above normal bundle is a smooth complex manifold , whose tangent space at a point @xmath135 is canonically isomorphic to the space of global sections of the normal bundle @xmath133 of @xmath128 ( thus @xmath11 has dimension 4 ) .
the conformal structure of @xmath11 is described by its tangent cone , which corresponds to the sections of @xmath133 having at least one zero ( as such a section decomposes as 2 sections of @xmath136 , the vanishing condition means that they both vanish at the same point , which is a quadratic condition on the sections of @xmath133 ) .
we thus get a conformal diffeomorphism from * m * to an open set of @xmath11 .
we remark that @xmath121 is an open set of the projective tangent bundle of @xmath3 , as @xmath3 is the space of leaves of @xmath137 , but it is important to note that , in general , the reverse inclusion is not true ( i.e. not any direction in @xmath3 is tangent to a line corresponding to a point in * m * , or , equivalently , @xmath0-surfaces are not compact @xmath13 s , in general , see section [ s.w5 ] ) .
for example , if @xmath138 ( with the notations in section [ s.w8 ] ) , @xmath121 is an open subset in the @xmath13-bundle @xmath139 , consisting in the set of directions transverse to the _ contact structure _ of @xmath3 ( see section [ ss.w84 ] ) .
@xmath121 is , thus , in this case , a rank 2 affine bundle over @xmath3 .
another canonical @xmath13-bundle on @xmath3 , that is @xmath140 , leads to the _ ambitwistor space _ @xmath7 , by definition the space of null - geodesics of * m * @xcite .
it is an open set of the projective cotangent bundle of @xmath3 ( or , equivalently , the grassmannian of 2-planes in @xmath141 ) @xcite ( more precisely , a plane @xmath142 corresponds to a null - geodesic @xmath143 ( contained in @xmath0 ) if it is tangent to at least one projective line @xmath128 , corresponding to a point @xmath24 ) . to see that ,
let @xmath96 be a point in @xmath144 , @xmath0 a @xmath0-surface passing through @xmath96 , i.e. @xmath127 and @xmath128 contains @xmath145 ; let @xmath146 be a plane tangent to @xmath128 .
as small deformations of @xmath128 still correspond to points of * m * , we consider those rational curves which are tangent to @xmath78 .
they correspond to a ( continuous ) set of points on a curve @xmath147 , that will turn out to be a null - geodesic .
indeed , all we have to prove is @xmath148 , and @xmath149 corresponds to a section @xmath150 of @xmath133 , vanishing at @xmath151 ; as @xmath152 , @xmath150 is determined by its derivative at @xmath145 , which is a linear map @xmath153 ( the infinitesimal deformation of the direction of @xmath128 within @xmath78 ) .
as the points of @xmath8 correspond to lines tangent to @xmath78 , we have that @xmath154 corresponds to a section of @xmath133 collinear to @xmath150 , thus @xmath8 verifies the equation of a ( non - parameterized ) geodesic .
see @xcite , @xcite , and section [ s.w4 ] for details .
* example . *
the space of null - geodesics of @xmath155 is the total space of a @xmath156-bundle over @xmath157 , the flag manifold ( see section [ s.w8 ] ) ; a 2-plane @xmath158 which corresponds to a null - geodesic in * m * is identified either to a projective diffeomorphism @xmath159 ( section [ ss.w84 ] , case 3 ) , or to a point @xmath160 , resp .
a plane @xmath161 containing @xmath20 , and different from @xmath162 ( section [ ss.w84 ] , cases * 2 * and @xmath163 ) .
* conventions . * except for some results in section [ s.w5 ] , we will consider * m * to be a self - dual civilized 4-manifold , i.e. the ( twistor ) space @xmath3 of @xmath0-surfaces of * m * is a hausdorff smooth complex 3-manifold , and the projection @xmath164 is a submersion ( e.g. * m * is geodesically convex ) , see @xcite .
we will frequently identify , following the deformation theory of kodaira ( see @xcite ) , the vectors in @xmath26 with sections in the normal bundle @xmath165 of the projective line @xmath128 in @xmath3 .
we also consider the space of null - geodesics @xmath7 , as an open subsetset of @xmath166 . for a null - geodesic @xmath8 , resp .
a @xmath0-surface @xmath167 , we denote by @xmath168 , resp . @xmath145 , the corresponding point in @xmath7 , resp .
@xmath3 .
the vectors on @xmath7 can be expressed in terms of infinitesimal deformations of geodesics of * m * ( jacobi fields ) . more precisely : @xmath169 where , for a null - geodesic @xmath8 , @xmath170 is the space of jacobi fields @xmath114 such that @xmath171 , and @xmath172 is its subspace of jacobi fields `` along '' @xmath8 , i.e. @xmath173 in any point of the geodesic . a class in @xmath174 is represented by jacobi fields yielding the same local section of the normal bundle @xmath175 of @xmath8 in * m*. this is equivalent to the following obvious fact : the kernel of the natural application @xmath176 is @xmath172 .
as a consequence , jacobi fields on @xmath8 induce particular local sections in @xmath175 , which turn out to be solutions of a differential operator of order 1 on @xmath175 , see section [ s.w4 ] . the conformal geometry of * m * induces a particular structure on @xmath7 : we describe it in order to obtain an expression of @xmath6 in terms of the geometry of the ( ambi-)twistor space .
we have a canonical hyperplane @xmath177 in @xmath178 , defined by @xmath179 where @xmath180 is the set of jacobi fields @xmath114 everywhere orthogonal to @xmath181 ( i.e. @xmath182 _ and _ @xmath183 ) .
we deine now two fiels of cones in @xmath184 , both contained in @xmath185 : let @xmath8 be a null - geodesic in * m * , and , for each point @xmath186 , let @xmath187 be the @xmath0-plane containing @xmath149 .
the ( infinitesimal ) @xmath0-cone @xmath188 at @xmath189 is defined as follows : @xmath190 where @xmath191 is the set of jacobi fields @xmath114 on @xmath8 satisfying the condition @xmath192 the @xmath0-cone @xmath188 is _ flat _ , i.e. it is included in the 2-plane @xmath193 consisting of jacobi fields contained in the @xmath0-plane defined by @xmath181 in each point of it .
we have to prove that @xmath191 is included in @xmath194 , defined as follows : @xmath195 we will prove that @xmath196 , therefore it will follow that the latter is non - empty , and is a linear space of dimension 2 .
we denote by @xmath197 the parallel displacement , along @xmath8 , of a non - zero vector in @xmath187 , transverse to @xmath181 .
then @xmath198 , because @xmath8 is included in the is totally geodesic surface @xmath0 , thus we can characterize @xmath199 as the set @xmath200 , for any @xmath201 .
we then observe that @xmath202 because @xmath203 is in @xmath42 , thus @xmath204 so the scalar function @xmath205 satisfies to a linear second order equation , hence it it determined by its initial value and derivative .
it follows then that it is identically zero , thus @xmath206 everywhere , as claimed .
another subset in @xmath178 is the @xmath9-cone @xmath207 , defined as follows : let @xmath8 be a null - geodesic in * m * , and , for each point @xmath186 , let @xmath208 be the @xmath9-plane containing @xmath149 .
the ( infinitesimal ) @xmath9-cone @xmath207 at @xmath189 is defined as follows : @xmath209 where @xmath210 is the set of jacobi fields @xmath114 on @xmath8 satisfying the condition @xmath211 it is important to note that , in general , the projective curves @xmath212 and @xmath213 are non compact , as each of them corresponds to the set of points on @xmath8 , which is non - compact , in general .
the field of @xmath9-cones on @xmath7 is the object of main interest in this paper .
we may already guess that its flatness ( i.e. the situation when @xmath207 is a subset in a 2-plane ) can be related to some special property of the conformal structure of * m*. we have seen that @xmath188 is included in the 2-plane @xmath214 , i.e. the condition @xmath215 can be generalised to the linear condition @xmath216 , but there is no canonical way of supplying the `` missing '' points of @xmath8 with some appropriate jacobi fields in order to `` complete '' @xmath207 as in the @xmath0-cones case .
this would be possible , for example , if @xmath212 would be an open subset in a projective line .
but the defect of @xmath207 to be part of a 2-plane is measured by its _ projective curvature _ , and we will see in section [ s.w4 ] that the vanishing of the latter implies the vanishing of @xmath6 ( theorem [ t1 ] ) .
@xmath217 we study now the field of @xmath9-cones of @xmath7 in relation with @xmath3 and the canonical projection @xmath218 .
first , we note that there are complex projective lines in @xmath7 tangent to the directions in @xmath207 : let @xmath189 , @xmath186 a point on the null - geodesic @xmath8 ; let @xmath208 be the @xmath9-plane tangent to @xmath219 .
the rational curve @xmath217 in @xmath7 ( containing @xmath168 ) , is by definition the set of null - geodesics passing through @xmath96 and tangent to @xmath208 .
the curves @xmath220 are projected by @xmath221 onto the complex lines @xmath128 through @xmath145 ( corresponding to the @xmath0-surface @xmath0 containing @xmath8 ) tangent to the 2-plane @xmath222 . on the other hand , it is easy to see that the complex projective lines @xmath223 ( defined in an analogous way to @xmath217 ) , which are tangent to ( an open set of the directions of ) @xmath188 , are contained in the fibers of @xmath221 .
in fact , they coincide with some of the projective lines passing through the point @xmath224 .
the _ integral @xmath9-cones _ in @xmath7 , resp .
@xmath3 are defined by : @xmath225 we intend to prove that @xmath226 is the _
canonical lift _ of @xmath227 ( see proposition [ p4 ] ) .
we know that @xmath228 .
we have then the following : [ p1 ] except for the vertices @xmath229 and @xmath230 , the two integral cones @xmath226 and @xmath227 are smooth , immersed , surfaces of @xmath7 , resp .
@xmath3 .
the open set of @xmath7 which is the space of null - geodesics of * m * can be viewed as the space of integral curves of the _ geodesic distribution _
@xmath231 of lines in @xmath30 , the total space of the fibre bundle of isotropic directions in @xmath23 .
@xmath232 is defined as the horizontal lift ( for the levi - civita connection on * m * ) of @xmath233 , which is an isotropic line in @xmath26 .
this definition is independent of the chosen metric and connection @xcite , and , by integrating this distribution ( as * m * is civilized ) , we get a holomorphic map @xmath234 , where ( an open set of ) @xmath7 is the space of leaves of this foliation .
this map can be used to compute the normal bundle of @xmath217 , @xmath235 , see @xcite,@xcite,@xcite .
indeed , we have lines @xmath236 , such that @xmath237 , which project onto @xmath217 , thus we get the following exact sequence of normal bundles : @xmath238 where we have written the ambient spaces of the normal bundles on the second position .
the central bundle is trivial ( @xmath239 is trivially embedded in @xmath240 , which is trivially embedded in @xmath30 as a fibre ) , and it is easy to check that the left hand bundle is isomorphic to the tautological bundle over @xmath132 , @xmath131 .
this proves that @xmath241 , in particular the conditions in the completeness theorem of kodaira @xcite are satisfied . thus the lines in the integral @xmath9-cone @xmath226 form an analytic subfamily of the family @xmath242 , that correspond to the sections of the normal bundle of @xmath217 , vanishing at @xmath243 , or , equivalently , to the points @xmath96 of @xmath143 .
but , in order to prove the smoothness of @xmath244 , we first remark that the surface @xmath245 , defined as follows , is smooth : @xmath246 where @xmath247 is the @xmath9-plane containing @xmath181 . @xmath248 is smooth , and @xmath249 .
we note now that @xmath248 is everywhere , with the exception of the points of @xmath250 , transverse to the fibers of the submersion @xmath234 .
we may conclude that @xmath251 is a smooth analytic submanifold of @xmath7 ( not closed ) .
we can use similar methods to prove that @xmath252 is an immersed submanifold of @xmath3 ( by using the projection @xmath218 ) .
there is another argument for this latter claim , which gives the tangent space to @xmath227 in any point : we see @xmath227 as the `` trajectory '' of a 1-parameter deformation of @xmath128 : we fix @xmath145 and we `` turn '' @xmath128 around @xmath145 by keeping it tangent to @xmath222 .
the trajectory of this deformation is smooth in @xmath253 iff any non - identically zero section @xmath254 of the normal bundle @xmath165 corresponding to this 1-parameter deformation ) does not vanish at @xmath255 . in particular , the tangent space @xmath256 is spanned by @xmath257 and @xmath258 . but the sections @xmath254 generating this deformation are the sections of @xmath165 vanishing at @xmath145 , and they vanish at only one point ( and even there , only at order 0 ) unless they are identically zero , because @xmath259 .
the values of these sections in the points of @xmath128 other than @xmath145 , plus their derivatives in @xmath145 ( well - defined as they all vanish at @xmath145 ) , define a 1-dimensional subbundle of @xmath165 , which is isomorphic to @xmath130 .
in fact , we have a 1 - 1 correspondence between the subbundles of @xmath165 isomorphic to @xmath130 and the 2-planes in @xmath260 .
then , the space of holomorphic sections of such a bundle is a linear space of dimension 2 , consisting in a family of sections of @xmath165 vanishing on _ different _ points of @xmath128 .
thus we get a 2-plane @xmath41 of isotropic vectors in @xmath26 , which is easily seen to be an @xmath9-plane , as the @xmath0-plane @xmath261 consists in the set of all sections of @xmath165 vanishing at @xmath145 ( we have @xmath262 ) . the tangent space to @xmath227 , in a point @xmath263 ,
is spanned by the subbundle of @xmath165 ( isomorphic to @xmath130 see above ) , defined by the isotropic vectors @xmath264 .
if @xmath265 is the null - geodesic generated by @xmath266 , we conclude that @xmath256 is the 2-plane determined by @xmath265 , and that @xmath267 .
* example . * if @xmath155 , then the integral @xmath9-cone @xmath227 in @xmath3 , for @xmath268 ( where @xmath159 is a projective diffeomorphism ) , is the ( smooth away from the vertex @xmath269 ) surface @xmath270 .
its compactification ( by adding the _ special cycle _ @xmath271 ) is singular ( section [ ss.w84 ] ) .
as any smooth surface in @xmath3 has a canonical lift in @xmath272 , we get : [ p4 ] the integral @xmath9-cone @xmath226 is the canonical lift of the integral @xmath9-cone on @xmath3 , @xmath227 . basically , this lift can only be defined for @xmath252 , but , in this special case , it can be extended by continuity to @xmath145 .
of course , the smoothness of the lifted surface can only be deduced away from the vertex @xmath168 ( from the smoothness of @xmath252 ) .
as noted in section [ s.w3 ] , we intend to find a relation between the `` curvature '' of the @xmath9-cone @xmath207 ( its non - flatness ) and the weyl tensor @xmath6 of @xmath97 .
we begin by defining the _ projective curvature _ of @xmath273 : a projective structure on a manifold @xmath274 is an equivalence class of linear connections yielding the same geodesics . in such a space , we can define the _ projective curvature _ of a curve @xmath275 in a point @xmath276 as the linear application @xmath277 , with @xmath278(modulo @xmath279 ) , for @xmath100 any connection in the projective structure of @xmath274 .
in particular , we take for @xmath274 the projective space @xmath280 , with its canonical projective structure , and for @xmath275 we take @xmath212 , the projectivized @xmath9-cone in @xmath168 .
[ sigma ] the projective curvature of the @xmath9-cone @xmath207 at the generating line @xmath281 is the projective curvature of @xmath212 in @xmath276 , and is identified to a linear application @xmath282 where @xmath276 is the tangent direction to @xmath217 in @xmath168 , and @xmath283 . in order to compute the projective curvature of @xmath207
, we establish first some canonical isomorphisms between the spaces appearing in the above definition and some linear subspaces of @xmath26 .
we will fix now the geodesic @xmath8 , the point @xmath186 ( therefore also @xmath284 ) , and , thus , the @xmath9-plane @xmath285 containing @xmath149 , as well as @xmath286 , the orthogonal space to @xmath149 . for simplicity , in the following lemmas we will currently omit some indices referring to these fixed objects .
[ l1 ] there is a canonical isomorphism @xmath287 between the tangent space @xmath279 to the projective cone @xmath288 and the tangent space @xmath219 to the geodesic @xmath8 in the point @xmath96 corresponding to the direction @xmath289 .
let @xmath290 .
we will define @xmath291 as follows : recall that @xmath292 , where @xmath293 ( the tangent space in a point to a cone depends only on the line containing the point ) . we know that @xmath276 corresponds to @xmath294 , the space of jacobi fields on @xmath8 , vanishing at @xmath96 , and such that @xmath295 .
it will be shown in the proof of the next theorem that @xmath296 consists of classes of jacobi vector fields such that @xmath297 , ( [ * w ] ) .
then , on a representative jacobi field @xmath298 , we define @xmath291 to be the class of jacobi fields in @xmath299 , represented by the following jacobi field @xmath300 on @xmath8 , which is given by @xmath301 .
we remark that @xmath302 is what we usually note @xmath303 , when the parameter on @xmath8 is understood .
it is straightforward to check that @xmath304 induces an isomorphism @xmath305 for each non - zero @xmath306 .
we remark that @xmath307 , the 4-dimensional subspace represented by jacobi fields @xmath114 , such that @xmath308 .
we further introduce the subspace @xmath309 , represented by jacobi fields @xmath114 as before , with the additional condition @xmath310 .
it is a 3-dimensional subspace , and it contains @xmath311 .
the curvature of @xmath207 will take values in @xmath312 , and we will show ( [ * * w ] ) in the proof of the next theorem that it takes values in a smaller space , @xmath313 .
@xmath314 is just the normal space of @xmath275 in @xmath315 at @xmath276 , and @xmath316 is the subspace of @xmath317 consisting in elements represented by @xmath318 .
[ l2 ] there is a canonical isomorphism @xmath319 as @xmath101 is a subbundle of the normal bundle @xmath320 , @xmath316 is isomorphic to@xmath321 . as in lemma
[ l1 ] , we will construct the inverse isomorphism @xmath322 : let @xmath323 be a linear application .
let @xmath324 be a representant of @xmath325 ( it involves a choice of a complementary space to @xmath41 in @xmath326 ) .
we define @xmath327 as being induced by the following linear application between spaces of jacobi fields on @xmath8 : @xmath328 , where the second space corresponds to @xmath329 , i.e. it contains jacobi fields @xmath114 such that @xmath330 . consider a parameterization of @xmath8 around @xmath96 , and let @xmath331 .
we define @xmath332 by @xmath333 , and it is easy to check that the class of @xmath334 in @xmath335 is independent of the representant @xmath336 , such that @xmath322 is well - defined .
it is also obviously invertible .
we are now in position to translate the projective curvature of @xmath273 in terms of conformal invariants of @xmath97 : [ t1 ] let @xmath96 be a point in a null - geodesic @xmath8 . then the projective curvature @xmath337 of the @xmath9-cone @xmath207 at @xmath276 ( corresponding to @xmath96 , see definition [ sigma ] ) , which is a linear map @xmath338 takes values in @xmath339 ( see above ) , and is canonically identified to the linear map @xmath340 defined by the self - dual weyl tensor of * m * : @xmath341 consider the following analytic map , which parameterizes , locally around @xmath186 , the deformations of the geodesic @xmath8 that correspond to points contained in the integral @xmath9-cone @xmath226 : @xmath342 where @xmath124 is a neighborhood of the origin in @xmath14 , and @xmath343 is a deformation of the null - geodesic @xmath8 , such that @xmath344 where the parameterization of the geodesic @xmath8 satisfies @xmath345 , and @xmath346 is the @xmath9-plane in @xmath347 containing @xmath348 . * convention * we know that @xmath349 is defined around the origin in @xmath14 , so there exists a polydisc centered in the origin included in @xmath124 , therefore all the relations that we will use are true for values of the variables @xmath350 sufficiently close to 0 . for simplicity
, we will omit to mention these domains .
the geodesics @xmath343 correspond to points in @xmath351 , and the jacobi fields @xmath352 on @xmath8 , defined as @xmath353 correspond to vectors in @xmath207 tangent to the above mentioned lines .
we suppose that the deformation @xmath349 is _ effective _ , i.e. @xmath354 and @xmath355 , which is equivalent to @xmath356 . in order to compute the projective curvature of @xmath207
, we need thus to study the ( second order ) infinitesimal variation of these jacobi fields on @xmath8 .
as they are determined by their value and first order derivative in @xmath345 , we need to evaluate @xmath357 for the first derivative of @xmath352 at @xmath358 , and @xmath359 for the second .
dots mean , as before , covariant differentiation with respect to the `` speed '' vector @xmath181 , thus correspond to the operator @xmath360 . as the covariant derivation @xmath361 has no torsion
, we can apply the usual commutativity relations between the operators @xmath362 and use them to differentiate the following equation , which follows directly from the definition of @xmath349 and @xmath352 : @xmath363 we get then @xmath364 we recall now that , besides ( [ eq : jt ] ) , we have @xmath365 , thus @xmath366 is isotropic , which implies that : @xmath367 as @xmath368
. equations ( [ eq : djt ] ) and ( [ eq : djt2 ] ) prove that @xmath369 which completes the proof of lemma [ l1 ] . from ( [ eq : djt2 ] ) , it equally follows that @xmath370 is isotropic , and , by differentiating ( [ eq : djt2 ] ) , we get @xmath371 from , ( [ eq : djt ] ) we have that @xmath372 is isotropic , and also @xmath373 which , together with ( [ eq : djt2 ] ) , implies that @xmath374 .
we have then @xmath375 which proves that the curvature @xmath337 of the @xmath9-cone takes values in @xmath316 , as it is represented by @xmath376 . in view of the lemmas [ l1 ] and [ l2 ] , it is clear now that the projective curvature @xmath337 is represented by the following application : @xmath377 from ( [ eq : djt3 ] ) , as @xmath378 and @xmath379 , we get @xmath380 the right hand side actually involves only @xmath6 , as the other components of the riemannian curvature vanish on this combination of vectors , thus we can replace @xmath381 with @xmath6 inn the above relation . on the other hand ,
the class of @xmath382 modulo @xmath41 is determined by its scalar product with @xmath303 , which represents a non - zero generator of @xmath383 .
the proof of the theorem is now complete .
we may ask whether the projective lines in @xmath3 are the geodesics of some projective structure . indeed , in the conformally flat case , when * m * is the grassmannian of 2-planes in @xmath384 ( the complexification of the moebius 4-sphere ) , @xmath385 , and the complex lines are given by the standard ( flat ) projective structure .
but there are two reasons ( related to each other , as we will soon see ) why @xmath3 can not carry a canonical projective structure : first , we do not necessarily have projective lines @xmath386 in every direction of @xmath260 ( this would mean that @xmath387 , see next section for a treatment of this problem ) , and second , the lift of a 2-plane @xmath388 would be a 2-plane in @xmath178 , so @xmath207 would be a flat cone : [ twfl ] the projective lines @xmath128 in the twistor space @xmath3 are geodesics of a projective structure iff it is projectively flat , and * m * is conformally flat .
if @xmath3 admits a projective structure , some of whose geodesics are the lines @xmath128 , then we have , for a fixed @xmath389 , a linear connection around @xmath145 , whose geodesics in the directions of @xmath390 coincide , locally , with @xmath128 .
this means that the integral @xmath9-cone @xmath227 , for @xmath391 a null - geodesic , is part of a complex surface ( namely @xmath392 , where @xmath388 is the 2-plane corresponding to @xmath8 ) .
then the integral @xmath9-cone @xmath226 , the lift to @xmath7 of @xmath227 , is also a complex surface , thus @xmath207 is a subset of the tangent space @xmath393 , thus a flat cone .
as this is true for all points of @xmath3 and for all null - geodesics @xmath8 , theorem [ t1 ] implies that * m * is flat . on the other hand ,
it is well - known that the twistor space of a conformally flat manifold admits a flat projective structure , for which the projective lines @xmath128 are geodesics , @xcite .
we have given , in the preceding section , a way to measure the projective curvature of the @xmath9-cone in @xmath7 ; we shall see now what happens in the special case when this cone is _ complete _ in a point @xmath168 , i.e. when @xmath212 is a compact submanifold in @xmath394 .
this situation appears for example if , for any direction in @xmath388 , there are projective lines in @xmath3 tangent to it .
[ twist ] let @xmath3 be the twistor space of the connected civilized self - dual 4-manifold @xmath97 , and suppose that , for a point @xmath395 and for a 2-plane @xmath396 , there are projective lines @xmath128 tangent to each direction of @xmath41
. then @xmath97 is conformally flat .
the idea is to prove that the integral @xmath9-cone @xmath227 is a smooth surface .
we know that this holds in all its points except for the vertex @xmath145 ( proposition [ p1 ] ) .
the fact that all direction in @xmath222 admits a tangent line is a necessary condition for this cone to be a smooth surface , as it needs to be well - defined around @xmath145 .
we choose an auxiliary hermitian ( real ) metric @xmath397 on @xmath3 .
its restrictions @xmath398 to the lines @xmath399 yield khlerian metrics on these lines ; in fact these metrics are deformations of one another , just like the lines @xmath128 are .
this means that the metrics @xmath398 depend continuously on @xmath400 , a parameter in a compact set .
we can therefore find a lower bound @xmath401 for the injectivity radius of all @xmath402 at @xmath145 , and a finite upper bound @xmath381 for the norm of all the second fundamental forms @xmath403 we can also suppose that @xmath404 is smaller than the injectivity radius of @xmath405 at @xmath145 . the first step is to prove that @xmath227 is a submanifold of class @xmath406 .
as its tangent space is everywhere a complex subspace of @xmath141 , it will follow that it is a complex analytic submanifold .
consider now the exponential map @xmath407 , defined for the metric @xmath397 ; if we restrict it to a ball of radius less than @xmath404 , it is a diffeomorphism into @xmath3 .
the image of the complex plane @xmath41 is then a smooth 4-dimensional real submanifold @xmath275 of @xmath3 , and there exists a positive number @xmath408 such that the exponential map in the directions normal to @xmath275 , @xmath409 restricted to the vectors of length less than @xmath408 , is a diffeomorphism .
the image of this diffeomorphism is a tubular neighborhood of @xmath275 , and we will denote by @xmath410 such a tubular neighborhood of `` width '' @xmath411 , for @xmath412 . the existence of an upper bound @xmath381 for the second fundamental forms of @xmath413 implies the following fact : for any @xmath412 , there is a neighborhood @xmath414 of the origin such that @xmath415 is contained in @xmath410 , and is transverse to the fibers of the orthogonal projection @xmath416 , where @xmath417 .
this is standard if @xmath227 is a submanifold ; but it is also true in our case , where @xmath227 is a union of submanifolds @xmath128 .
now it is easy to prove that @xmath227 is a @xmath406 submanifold of @xmath3 ( the projection @xmath418 yields a local @xmath406 diffeomorphism from a neighborhood of @xmath145 in @xmath275 to a neighborhood of @xmath145 in @xmath227 ; it is @xmath406 in @xmath145 because @xmath275 is tangent to @xmath227 at @xmath145 ) .
so @xmath227 is a @xmath406 submanifold of @xmath3 ; its tangent space is complex in each point , thus @xmath227 is a complex - analytic surface immersed in @xmath3 .
we have then that @xmath419 , being the lift of @xmath227 , is a smooth analytic surface immersed in @xmath7 , in particular the @xmath9-cone @xmath207 is a complex plane .
theorem [ t1 ] implies that @xmath6 vanishes on the @xmath9-plane @xmath420 which contains @xmath149 , for every point @xmath186 .
now , the plane @xmath388 is not the only one admitting projective lines @xmath128 tangent to any of its directions : all planes `` close '' to @xmath222 have the same property .
then @xmath6 vanishes on a neighborhood of @xmath8 , hence on the whole connected manifold * m*. there is a more general situation where the integral @xmath9-cone @xmath227 through @xmath0 is smooth in @xmath0 : [ 3 ] suppose that , for each direction @xmath421 , there is a smooth ( non - necessarily compact ) curve @xmath422 tangent to @xmath276 , such that : \(i ) if @xmath276 is tangent to a projective line @xmath128 , then @xmath423 ; \(ii ) @xmath422 varies smoothly with @xmath424 .
then @xmath425 is a smooth surface around @xmath0 , containing the @xmath9-cone @xmath227 and @xmath426 , where @xmath427 is the @xmath9-plane containing @xmath181 .
the proof is similar to the one of the previous theorem .
note that , if there is a direction @xmath276 which is not tangent to a projective line @xmath128 , we can not apply the deformation argument in theorem [ twist ] to conclude that @xmath6 vanishes everywhere .
* example . * if @xmath155 , then @xmath157 and there are some particular planes for which the conditions in theorem [ 3 ] are satisfied , although theorem [ twist ] never applies to @xmath3 : for a generic 2-plane @xmath222 , the @xmath9-cone @xmath273 is not flat .
the above mentioned particular planes in @xmath141 correspond to the vanishing of @xmath6 on some particular @xmath9-planes , but * m * is not anti - self - dual ( see sections [ ss.w83 ] , and also [ ss.w87 ] , [ ss.w88 ] ) .
theorem [ twist ] suggests that the existence of a compact null - geodesic diffeomorphic to @xmath132 yields strong constraints upon the conformal structure of * m*. in fact , we have : [ compact ] if a connected complex self - dual 4-manifold @xmath97 admits a compact null - geodesic diffeomorphic to @xmath132 , the conformal structure of * m * is flat .
a similar result has been proven by y .-
g . ye using algebraic geometry techniques : if a _ projective _ complex manifoldadmits a conformal structure having a compact null - geodesic diffeomorphic to @xmath132 , it is conformally flat @xcite .
note that we do not need * m * to be compact in theorem [ compact ] ; on the other hand , we assume it to be self - dual .
we first remark that the main difficulty is the definition of @xmath7 , the space of null - geodesics , and of @xmath3 , the twistor space of @xmath97 , as * m * is not necessarily civilized .
this is only possible on small open sets , but , in general , we can not expect to have any global construction of this kind .
thus , things that were almost obvious in the twistorial framework ( like the existence of compact deformations of the null - geodesic @xmath8 ) , seem much more difficult to prove directly . the idea is to prove that all null - geodesics close to @xmath8 are diffeomorphic to @xmath132 .
then , we show that , conversely , every projective line which is a deformation of @xmath8 as a compact curve is a null - geodesic . in particular , sections in the normal bundle @xmath175 are induced by ( local ) jacobi fields .
we obtain then directly that @xmath93 .
[ p5 ] let @xmath8 be an immersed null - geodesic , diffeomorphic to a projective line @xmath132 .
then any local jacobi field @xmath114 with @xmath428 induces a global normal field @xmath429 on @xmath8 . if @xmath8 is just a compact geodesic , it may have points of self - intersection , but it is always an immersed curve .
it is more convenient then to think of @xmath8 as a projective line immersed in * m * rather than the image of this immersion . the tangent , normal bundles , etc .
are also to be thought as bundles over this projective line , still denoted by @xmath8 .
tubular neighborhoods of @xmath8 are then neighborhoods of the zero section in the normal bundle @xmath175 , small enough to be immersed ( non - injectively ) in * m * as a neighborhood of the image of @xmath8 .
we first notice that @xmath8 may be decomposed in the union of a two open sets @xmath430 , both biholomorphic to the unit disk in @xmath17 , and such that @xmath431 is connected .
then , for any local metric in @xmath18 , we have a jacobi equation around a point @xmath186 , and a jacobi field @xmath114 corresponding to prescribed @xmath432 .
it is easy to prove that the ( local ) normal field induced by @xmath114 is independent of the chosen metric .
moreover , this normal field is the unique solution , for the prescribed 1-jet in @xmath96 induced by @xmath432 , of a second order differential equation on @xmath175 : the jacobi equations for null - geodesic induce a second order linear differential operator @xmath433 on @xmath175 , depending only on the conformal structure @xmath18 of * m*. for a levi - civita connection @xmath361 of a local metric on * m * , we locally define the following differential operator on @xmath434 : @xmath435 by @xmath436 .
it obviously induces a ( local ) differential operator on @xmath175 , and all we need to show is that , for a different connection @xmath437 , the corresponding operator @xmath438 induces the same one on @xmath175 .
first we write @xmath439+\nb_{[x , y]}x-[\nb_xx , y],\ ] ] then we recall that another levi - civita connection @xmath437 is related to @xmath361 by the formula @xcite : @xmath440 so we directly obtain : @xmath441 thus they induce the same operator on @xmath175 .
this one is , therefore , globally defined ( the topology of @xmath8 is not important ) .
now , for any @xmath186 , we have an unique solution @xmath254 of @xmath433 , for a prescribed _ 1-jet _ @xmath442 ( which consists in the values in @xmath96 of @xmath254 and of his first - order derivative ) , globally defined on every contractible open set @xmath443 in @xmath8 .
this is because on any such contractible set the equation @xmath444 becomes a second order ordinary linear equation on a disk in @xmath17 , which admits global holomorphic solutions ( unique if we fix the initial conditions ) .
take now @xmath445 .
then , the two solutions @xmath446 and @xmath447 , defined on @xmath448 , resp .
@xmath449 , coincide on the connected @xmath450 , so they yield a global solution @xmath254 with the prescribed initial conditions in @xmath96 .
in particular , this solution is a global section of the normal bundle @xmath175 . after the infinitesimal result , the local one : [ tub ] small deformations of @xmath8 are also compact immersed projective lines .
the tubular neighborhoods considered below are always seen as images , by a local diffeomorphism , of subsets which are , generally , fiber bundles over @xmath132 of the normal bundle of @xmath8 , resp .
@xmath451 , the lift of @xmath8 to @xmath30 .
we need this because of the possible self - intersections of @xmath8 ; @xmath451 is always embedded .
consider an auxiliary hermitian metric @xmath397 on * m*.then @xmath452 induces the same topology like a round metric @xmath453 on the sphere @xmath109 .
we can define a tubular neighborhood @xmath454 of @xmath8 , as the open set of points @xmath455 with @xmath456 .
we choose @xmath408 small enough for @xmath454 to be a fiber bundle over @xmath8 ( the fiber @xmath457 , for @xmath186 , being the image of the real 4-plane of @xmath26 , @xmath397-orthogonal to @xmath219 , by the exponential of @xmath397 ) .
take now a finite number of contractible open sets whose union covers @xmath8 , and we choose holomorphic metrics in @xmath18 on each of this open sets .
then , on these sets we have connections , and it is well - known that any point in such a set has a basis of geodesically connected neighborhoods @xcite , @xcite .
we choose a finite number of such geodesically convex open sets , that cover @xmath8 , and such that they are all included in @xmath454 .
we have then @xmath458 ( of course , they are geodesically convex only with respect to some particular metric in @xmath18 , but we are interested only in the implications involving the null - geodesics , which are independent of the metric . )
it is immediate , @xcite , that a geodesically convex set @xmath459 has the following property : all maximal geodesics are closed submanifolds of @xmath459 and are contractible as topological spaces .
this is particularly true for null - geodesics included in @xmath459 .
we refine now the covering by another one , @xmath460 , such that @xmath461 , where @xmath462 , such that @xmath463 .
in fact , we ask for @xmath464 to be restrictions to @xmath465 of a tubular neighborhood @xmath466 , for @xmath467 sufficiently small ( @xmath464 are `` cylindrical '' neighborhoods ) .
we can easily imagine how to find such a refinement of the initial covering .
the proof of the proposition now follows two ideas : first , we consider a very special covering by disks ( for the round metric @xmath453 ) of @xmath8 ; second , we define a neighborhood of @xmath451 in @xmath30 of isotropic directions for which , using the special covering of @xmath8 , we can extend the associated null - geodesics and eventually get compact ones .
we call @xmath451 the canonical lift of a null - geodesic in @xmath30 ; it is an integral curve of the geodesic distribution of lines in @xmath30 .
the first idea has nothing to do with complex analysis ; it is just a matter of metric topology on the round sphere @xmath468 .
[ ldisc ] let @xmath469 be a covering of @xmath470 by open sets .
then there exists a positive number @xmath404 and a finite set of points @xmath471 , for @xmath472 , such that : \(i ) all disks @xmath473 are contained in at least one of the open sets @xmath465 ; \(ii ) the disks @xmath474 cover @xmath475 ; \(iii ) @xmath476 , where @xmath162 is the diameter of @xmath475 .
an important property of this covering is that all sets , as well as the intersections of a finite number of them , are convex for the round metric , thus contractible .
we intend to extend null - geodesics which can be projected diffeomorphically onto @xmath8 by means of the fiber projection @xmath477 .
we will do that step - by - step , extending it over disks @xmath478 of increasing radius .
but before that , we need to restrict ourselves to some particularly `` close to @xmath8 '' null directions . we need two things : the extensions need to remain within @xmath466 , and they should also be transverse to the fibers of @xmath479 , otherwise the projection into @xmath8 would not be an invertible diffeomorphism .
first , we consider the following compact subset of @xmath480 : @xmath481 where we say that a complex line @xmath20 is tangent to a real manifold @xmath482 if it contains a non - zero ( real ) vector tangent to this real submanifold . the hermitian metric @xmath397 on * m * induces a metric on @xmath480 , and also one on @xmath30 .
we can , then , evaluate the distance between @xmath451 and @xmath275 : @xmath483 as they are disjoint compact sets .
following lebrun @xcite , we can define the complex 5-manifolds @xmath484 as the spaces of null - geodesics of @xmath485 , equivalently the space of integral curves of the geodesic distribution in @xmath486 .
the projections @xmath487 , which send an isotropic direction to the null - geodesic tangent to it , is a submersion and the ( closed ) fibers are precisely the lifts of the null - geodesics of @xmath459 .
this construction is possible because @xmath459 are geodesically convex ( for a particular local holomorphic metric in @xmath18 ) , see @xcite for details .
we first consider a tubular neighborhood @xmath488 of @xmath451 in @xmath30 which projects , by @xmath489 _ inside _
@xmath466 , and such that @xmath490 .
this second condition ensures that all directions in @xmath488 are transverse to the fibers of @xmath479 .
consider then the following neighborhoods of @xmath491 : @xmath492 is an open application , so we define @xmath493 to be @xmath494 , where @xmath495 is an open neighborhood of @xmath496 contained in @xmath497 .
it is important to note that @xmath493 have the following property : for any point @xmath498 , the null - geodesic @xmath499 tangent to @xmath500 , contained in @xmath459 , lies into @xmath466 and is always transverse to the fibers of @xmath501 .
hence , its restriction to the points of @xmath464 projects diffeomorphically onto @xmath502
. moreover , all the points of @xmath503 that lie over @xmath504 are in @xmath493 . obviously , the crucial property of these open sets is that every null - geodesic starting there is totally contained in @xmath493 , at least the part that `` lies over '' ( in the sense of the projection @xmath479 ) @xmath505 . after constructing these sets @xmath506 ,
we define @xmath507 small enough for the tubular neighborhood @xmath508 , restricted to @xmath504 , to be contained into @xmath493 , for all @xmath509 . for each @xmath509 , this means that @xmath510 has to be less than the minimum of the following continuous functions defined on the compact set @xmath504 : @xmath511 this neighborhood @xmath512 of @xmath451 has the following property : for each @xmath513 , and for each @xmath509 such that @xmath514 , we have @xmath498 , and thus the whole null - geodesic @xmath515 , contained in @xmath459 , is lifted to @xmath516 .
the disadvantage of @xmath512 is that it does not necessarily contain @xmath517 .
but we know that the latter is contained in @xmath518 , which contains the union of all @xmath493 .
we recall now that the idea of proof is to extend a null - geodesic @xmath499 close to @xmath8 over the disks @xmath519 .
every extension over a disk brings @xmath499 from @xmath508 to the larger set @xmath488 . as we have a finite , well - determined , number of disks @xmath520 , all we need now to apply our extending idea is a sequence of open sets @xmath521 such that @xmath522 to do that , we construct @xmath523 as we have done for @xmath493 , and then @xmath524 by repeating the same procedure .
we stop after @xmath520 ( the number of disks covering @xmath525 , see lemma [ ldisc ] ) steps and claim : @xmath526 , the null - geodesic @xmath499 extends to a compact curve which projects ( via @xmath479 ) diffeomorphically onto @xmath8 .
fix @xmath527 .
we can define @xmath499 inside @xmath459 , where @xmath459 is an open set containing @xmath96 .
in particular , @xmath499 is well - defined over @xmath528 , where @xmath529 , see lemma [ ldisc ] .
( of course , this is because @xmath530 is transverse to the fibers of @xmath479 .
we intend to extend it over disks centered in @xmath531 .
consider the domains @xmath532 which are `` quadrilaterals '' contained in @xmath533 and containing @xmath534 , as in the following picture ( the `` vertical '' parts of the border of @xmath535 are segments of circles centered in @xmath531 ) : we change , if necessary , the order of the indices @xmath509 of @xmath536 and @xmath535 such that it coincides with the ordering of increasing distances @xmath537 .
we define than the open sets @xmath538 the closed disk @xmath539 is included in @xmath540 as soon as @xmath541 , where @xmath542 is the point of @xmath543 opposed to @xmath531 .
then , because of the specific geometry of the domains @xmath535 ( see lemma [ ldisc ] ) , we easily conclude that the domains @xmath540 are contractible ( along the geodesics of the sphere passing through @xmath531 ) and so are the intersections @xmath544 , too .
we prove , then , by induction , that @xmath499 can be extended over @xmath540 , and that all the corresponding points of @xmath503 are contained in @xmath545 .
this is obvious for small values of @xmath546 .
when we add @xmath547 to @xmath548 , we consider a point @xmath549 in the connected ( see above ) intersection @xmath550 .
it is contained in the @xmath464 that contains @xmath551 and , as @xmath552 is contained in @xmath553 , the connected piece @xmath554 is contained in @xmath555 , and thus the whole extension @xmath517 in @xmath459 is contained in @xmath555 , hence in @xmath545 .
the connectedness of the considered piece implies that @xmath556 coincides with @xmath554 .
thus we have obtained an extension of @xmath499 to @xmath540 , such that the corresponding lift lies in @xmath545 , as claimed . for large values of @xmath546 , the @xmath540 are all identical to @xmath557 , which contains @xmath558 .
we have thus proven that there is an extension of @xmath499 over this disk , such that the corresponding points of @xmath503 are in @xmath508 .
consider then the disk @xmath559 ; it is contained in some @xmath465 : the intersection @xmath560 of this disk with @xmath557 is a connected open subset of @xmath465 , and we know that @xmath561 is contained in @xmath508 .
this implies that we can extend in a unique way @xmath562 to @xmath465 , in particular to @xmath559 , and the corresponding points in @xmath563 are in @xmath564 .
we have proven that @xmath499 extends over @xmath8 , i.e. there is a maximal extension ( obviously unique ) of the null - geodesic tangent to @xmath500 , such that it projects ( via @xmath479 ) diffeomorphically onto @xmath8 .
the projection is @xmath565 , but the extended null - geodesic is clearly an analytic submanifold of * m*. the proof of proposition [ tub ] is now complete .
generic deformations of @xmath8 are embedded projective lines .
indeed , let @xmath566 be the nodes ( self - intersection points ) of @xmath8 , and consider the manifold @xmath567 , obtained by blowing - up the points @xmath566 .
then , generically , any null - geodesic @xmath568 close to @xmath8 avoids these points , hence is diffeomorphic to its lift to @xmath567 , which is a deformation of the lift of @xmath8 , thus it is an immersed projective line .
but the lift of @xmath8 is embedded , and so must be its deformations , hence @xmath568 is embedded . the next step is to prove that all deformations of @xmath8 as a compact curve are null - geodesics , by a dimension - counting argument ; we need to compute the normal bundle of @xmath8 .
we ask now if the family of projective lines in * m * defined as the deformations of @xmath8 is _ locally complete _ in the sense of kodaira @xcite . for this
, we need to prove that the dimension of the space of global sections in @xmath175 is equal to 5 , i.e. to the dimension of @xmath7 .
the extensions of the null - geodesics close to @xmath8 yield local diffeomorphisms between neighborhoods of @xmath8 in @xmath484 , resp . @xmath569 .
in fact , we have a projection @xmath570 onto the space of integral curves of the geodesic distribution in @xmath571 .
@xmath572 is the space of complex null - geodesic _ close to @xmath8 _ in * m*. but essential for us is that @xmath76 is a submersion , fact that has important consequences for the normal bundle of @xmath8 in * m*. the normal bundle of @xmath8 in *
m * is isomorphic to @xmath573 .
it is well - known that all holomorphic bundles over @xmath132 are direct sums of line bundles , all of which are isomorphic to @xmath574 .
we have the subbundle @xmath575 of @xmath175 , represented by vectors orthogonal to @xmath181 .
we have the following exact sequence : @xmath576 the right hand term of this sequence is a line bundle , and it admits global non - zero sections ( extensions of jacobi fields @xmath114 such that @xmath577 is a non - zero constant , see proposition [ p5 ] ) .
@xmath578 is then isomorphic to @xmath579 , with @xmath580 .
we denote by @xmath581 the subbundle of the normal bundle represented by vectors in @xmath582 .
it admits global sections , namely the extensions of jacobi fields contained in @xmath583 , see proposition [ p5 ] .
it is a line bundle , thus isomorphic to @xmath584 , as it contains global sections with prescribed 1-jet in a point . in general
, @xmath585 is the line bundle over @xmath132 admitting global sections for any prescribed @xmath546-jet in a point @xmath96 .
this section is unique , and it gives a unique value of the @xmath586-jet in @xmath96 .
we have the following exact sequence : @xmath587 it is easy to check that the right hand term admits local sections represented by jacobi fields for any prescribed 1-jet in a point @xmath96 . hence , @xmath588 .
all we can obtain now is that @xmath589 , with @xmath590 .
actually @xmath591 , otherwise all sections of @xmath592 that vanish somewhere would be contained in a line subbundle , which would contradict proposition [ p5 ] .
we have then @xmath593 . we want to prove that there is equality in all these three inequalities . we know
, from proposition [ tub ] , that there is a tubular neighborhood @xmath594 of @xmath451 in @xmath30 such that the null - geodesic distribution yields a foliation with compact leaves , and such that the projection onto the space @xmath572 of these compact curves is a submersion .
( of course , this space is nothing but the space of complex null - geodesics _ close to _ @xmath8 . )
it is obvious then that the normal bundle of @xmath451 in @xmath30 is trivial , as @xmath451 is e fiber of a submersion .
we have now the following exact sequence of bundles , related to the projection @xmath595 : @xmath596 where @xmath597 is the normal subbundle of @xmath451 represented by vectors tangent to the fibers of @xmath221 . in a point
@xmath598 , the fiber of @xmath221 is equal to @xmath599 , so the tangent space to it is isomorphic to @xmath600 , for the projective variety @xmath601 .
thus @xmath602 the central bundle in the exact sequence ( [ eq : opi ] ) is trivial .
the equation above then implies that the chern number of @xmath175 is subject to the following constraint : @xmath603 thus , as @xmath591 and @xmath604 , we have @xmath605 and @xmath606 . as observed above , generic , compact , simply - connected null - geodesics are embedded . from now on ,
we suppose @xmath8 is one of them .
then @xmath607 , so we can apply the theory of kodaira to deform @xmath8 , @xcite , so the dimension of the space of global sections of @xmath175 is @xmath608 , the same as the space of complex null - geodesic close to @xmath8 , hence the deformations of @xmath8 as a compact curve are null - geodesics in @xmath97 .
this means that any global section in @xmath175 can be represented , locally , by the jacobi fields that yield the same element in @xmath609 , the space of jets of order 1 in @xmath175 .
recall now the exact sequence ( [ eq : o1 ] ) ; we conclude that @xmath610 we have the canonical isomorphism @xmath611 , coming from the restriction of the projection @xmath612 ( we denote by @xmath613 , resp . @xmath581 , the subbundle of the normal bundle of @xmath8 , such that its fiber at @xmath186 is @xmath614 , resp .
@xmath615 ) .
we have thus @xmath616 , and all the 1-jets of @xmath613 yield global sections , thus local jacobi fields .
but the existence of a jacobi field in @xmath617 implies , by the jacobi equation @xmath618 , that @xmath619 , the self dual weyl tensor vanishes on @xmath41 , the @xmath9-plane generated by @xmath181 .
we recall now that , for a fixed point @xmath96 , @xmath620 is a polynomial of order 4 , and it is zero on @xmath41 for the compact null - geodesic @xmath8 .
the same is true for the compact deformations of @xmath8 , which implies that @xmath621 .
this holds for all the points of @xmath8 , and also for all points covered by the deformations of @xmath8 .
as these deformations cover at least an open set around @xmath8 , we conclude that @xmath6 vanishes on a non - empty open set , thus , being holomorphic , @xmath93 on the whole ( connected ) manifold * m*. this completes the proof of theorem [ compact ]
the null - geodesics contained in a @xmath0-surface @xmath0 define a projective structure on the totally - geodesic surface @xmath0 , which is also given by any connection on @xmath0 induced by a levi - civita connection on * m*. we claim that this projective structure is _ flat _ ,
i.e. locally equivalent to @xmath13 .
* example . * if @xmath155 , then a @xmath0-surface indexed by @xmath622 is @xmath623 , and the null - geodesics in @xmath624 are identified to the affine lines in @xmath625 ( see section [ ss.w85 ] ) . to prove the projective flatness of a 2-dimensional manifold @xmath0 , we need to prove that the thomas tensor @xmath626 vanishes identically @xcite .
this tensor is an analog of the cotton - york tensor in conformal geometry ( there is also a weyl tensor of a projective structure , but it only appears in dimensions greater than 2 ) . for a connection @xmath361 in the projective class of @xmath0 ,
the thomas tensor is defined as follows @xcite : for @xmath627 @xmath628 where the derivation involves only the curvature tensor @xmath381 , and the @xmath629 is the trace of the endomorphism @xmath630 .
the thomas tensor is independent of the connection @xmath361 , therefore we will consider that @xmath361 is induced by a levi - civita connection on * m*. [ p9 ] the thomas tensor of a @xmath0-surface can be expressed in terms of the anti - self - dual cotton - york tensor of * m * , thus it is identically zero . we need first to define the _ anti - self - dual _ cotton - york tensor as an irreducible component of the cotton - york tensor of * m*. * convention * we note @xmath631 the cotton - york tensor of @xmath97 ; we will not use this letter for the isotropic cone in this section , nor in the following one .
the cotton - york tensor is not conformally invariant ; its definition depends on a ( local ) metric @xmath49 in the conformal structure , which is supposed to be fixed @xcite:@xmath632 where @xmath397 is the normalized ricci tensor of * m * , @xmath633 @xmath634 being the trace - free ricci tensor , resp . the scalar curvature of the metric @xmath49 , and @xmath635 . in our case , @xmath636 , but the formula applies in all dimensions greater than 2 @xcite .
the cotton - york tensor @xmath631 of * m * is a 2-form with values in @xmath637 , thus it has two components @xmath638 , and @xmath639 .
@xmath631 satisfies a _ first _ bianchi identity , as @xmath397 is a symmetric tensor , and also a _ contracted _ ( second ) bianchi identity , coming from the second bianchi identity in riemannian geometry , @xcite : @xmath640 that means that @xmath641 , and is orthogonal on @xmath642 and on @xmath643 , which is identified with the image in @xmath644 by the metric adjoint of the contraction ( [ ar2 ] ) .
now , the hodge operator @xmath645 induces a symmetric endomorphism of @xmath644 , which maps the two above spaces isomorphically into each other .
this implies that @xmath646 and @xmath647 satisfy ( [ ar1 ] ) and ( [ ar2 ] ) ( note that these two relations are equivalent in their case ) .
the cotton - york tensor is related to the weyl tensor of * m * by the formula @xcite : @xmath648 where @xmath649 is induced by the codifferential on the second factor , and by the levi - civita connection @xmath361 .
then , @xmath646 has to be the component of @xmath650 in @xmath651 , and we know that the restriction of @xmath2 to @xmath652 is identically zero .
this means that @xmath653 hence , as * m * is self - dual , @xmath647 vanishes identically .
we can prove now that the thomas tensor of a @xmath0-surface @xmath0 is identically zero : first we prove @xmath654 we recall that the _ suspension _
@xmath655 , viewed as an endomorphism of @xmath34 , is defined by @xcite : @xmath656 where @xmath397 is identified with a symmetric endomorphism of @xmath23 .
we have then the following decomposition of the riemannian curvature @xcite : @xmath657 of course , if * m * is self - dual , @xmath92 and @xmath658 if @xmath659 ( in fact , the elements in @xmath44 , for any @xmath0-plane @xmath660 , correspond to the isotropic vectors in @xmath661 ) , because @xmath662 .
then , if we choose the basis @xmath663 in @xmath582 , we get @xmath664 which proves ( [ eq : thomh ] ) .
the thomas tensor of the projective structure of @xmath0 has the following expression ( see ( [ eq : thomas ] ) ) : @xmath665 and , as @xmath666 vanishes on the anti - self - dual 2-form @xmath667 , we conclude @xmath668 as the flatness of the projective structure on a 2-dimensional manifold is equivalent to the vanishing of its thomas tensor @xcite , we get [ prflat ] the projective structure of the @xmath0-surfaces of a self - dual complex manifold * m * is flat . from the classification of projectively flat compact complex surfaces ( @xcite , see also @xcite ) , we get then a classification of compact @xmath0-surfaces in * m * : [ clasb ] a compact @xmath0-surface of a self - dual complex 4-manifold belongs ( up to finite covering ) to the following classes : 1 .
@xmath13 ; 2 .
a compact quotient of the complex - hyperbolic plane @xmath669 ; 3 . a compact complex surface admitting a ( flat ) affine structure : \(i ) a kodaira surface ; + ( ii ) a properly elliptic surface with @xmath670 odd ; + ( iii ) an _ affine _ hopf surface ; + ( iv ) an inoue surface ; + ( v ) a complex torus .
see @xcite , @xcite , @xcite for details .
it has been shown by lebrun @xcite that , for a given geodesically connected 3-manifold with conformal structure @xmath671 , there is a ( germ - unique ) self - dual 4-manifold @xmath97 , such that @xmath671 is an umbilic hypersurface of @xmath97 .
we can therefore note by @xmath18 the conformal structure of @xmath672 , as it coincides with the restriction of the conformal structure of * m*. the technical tool used in the proof of this result @xcite is the _ twistor space _
@xmath3 of @xmath673 , which is the space of complex null - geodesics of this manifold .
it has been shown by lebrun @xcite that @xmath3 is a 3-manifold with a _ contact structure _ , and containing projective lines with normal bundle @xmath4 .
conversely , for any such manifold @xmath3 , the space of these lines _ tangent _ to the distribution of planes induced by the contact structure , is a conformal 3-manifold . on the other hand ,
@xmath3 can be identified with the twistor space of a self - dual 4-manifold * m * , in which @xmath672 is an umbilic hypersurface ( the _ conformal infinity _ of a einstein metric on * m * , @xcite ) .
@xmath3 has an additional structure , namely a _ contact structure _ , represented by a ( non - integrable ) distribution of 2-planes @xmath674 , which corresponds to the space of jacobi fields @xmath675 , where @xmath676 and @xmath3 is considered as the twistor space of @xmath672 , or , equivalently , to the null - geodesic @xmath677 , if @xmath3 is considered as the twistor space of * m*. we also remark that the above contact structure yields a section in the bundle @xmath678 , which is never tangent to the @xmath9-cones ( as @xmath672 is transverse to all @xmath9-planes ) .
a holomorphic contact structure on a twistor space @xmath3 does not necessarily determine a conformal 3-manifold : if @xmath157 with its contact structure ( see section [ ss.w84 ] ) , there is no rational curve in @xmath3 , with normal bundle @xmath4 , tangent to the corresponding distribution of planes .
on the other hand , this contact structure yields an einstein metric on @xmath155 , which admits the smooth 3-manifold @xmath679 as `` infinity '' .
however , this `` infinity '' is not conformal ( see section [ ss.w86 ] ) . in the 3-dimensional case ,
the conformally invariant tensor `` measuring '' the non - flatness of @xmath673 is the cotton - york tensor , defined in general by @xmath680 where @xmath397 is the _
ricci tensor @xcite , ( [ eq : rich ] ) .
for the 3-manifold @xmath681 , we have @xmath682 .
it is natural to ask how is the cotton - york tensor of @xmath672 related to the weyl tensor of * m*. we first recall a few facts of conformal geometry in dimensions 3 and 4 . for the 3-dimensional manifold @xmath672
, the riemannian curvature has the following expression : @xmath683 as there is no weyl tensor ( in general , @xmath684 is the ricci component of the curvature , @xcite ) .
if we introduce the hodge operator @xmath685 , then @xmath686 is equivalent to the symmetric 2-tensor @xmath687 .
a straightforward application of the above formula yield @xmath688 for the 4-dimensional manifold * m * , the components of the riemannian curvature can also be expressed as eigenspaces of @xmath33-type operators .
namely , regarding @xmath689 as a symmetric endomorphism of @xmath690 , @xmath6 is the trace - free component of @xmath381 in @xmath691 , and @xmath2 is the trace - free component of @xmath381 in @xmath692 @xcite .
let @xmath681 be a hypersurface , such that the restriction of the conformal structure @xmath18 of * m * to @xmath672 is non - degenerate ( equivalently , @xmath693 is nowhere tangent to an isotropic cone ) .
we call @xmath18 the induced structure on @xmath672 .
we suppose that @xmath672 is umbilic . there is no possible choice of an `` orientation '' in this case . indeed , the group of conformal transformations of @xmath694 , @xmath695 is non - connected if @xmath696 is even , and a choice of an orientation is a restriction of the frame bundle of a @xmath696-dimensional conformal manifold to the connected component of @xmath697 .
but if @xmath696 is odd , @xmath698 is connected , so all @xmath698-frames can be connected to each other by continuous paths .
therefore , although a complex - riemannian 3-manifold admits , locally , 2 possible orientations , they are conformally equivalent , fact that makes impossible a canonical way to associate an orientation to a metric in the conformal class .
there is another way to see this difference between the even- and odd - dimensional conformal manifolds : let @xmath699 be a @xmath696-dimensional conformal manifold .
then a ( local ) metric @xmath49 in the conformal class @xmath18 is a global section in @xmath700 , where @xmath20 is the bundle of weighted scalars .
we can canonically associate to @xmath18 a global section of @xmath701 , which is the induced metric on the canonical bundle .
for a given metric , there are only 2 ( local ) sections of @xmath702 of `` norm '' @xmath703 .
we can pick one of them if we have a given section in @xmath704 ( an `` orientation '' ) , and if we have a canonical way to associate to @xmath49 , which is a section in @xmath22 , a section of @xmath705 .
this can be done if @xmath696 is even , namely @xmath706 .
we also remark that , if @xmath696 is even , all we need to define a conformal structure on @xmath274 is just the bundle @xmath22 ( which is a @xmath707-th root of @xmath702 ) , but if @xmath708 is odd , @xmath22 automatically gives us @xmath709 .
we can canonically identify @xmath710 and @xmath661 , restricted to @xmath681 , to @xmath711 , by : @xmath712 [ umbilic ] let @xmath672 be an umbilic hypersurface of a self - dual manifold * m*. then : \(i ) the weyl tensor of * m * , restricted to @xmath672 , is identically zero ; \(ii ) the cotton - york tensor of @xmath672 is related to the self - dual weyl tensor of * m * by the formula : @xmath713 where @xmath714 , @xmath715 is unitary for the metric @xmath49 , and the hodge operator @xmath716 is induced by @xmath49 and the orientation on @xmath672 admitting @xmath254 as an exterior normal vector .
[ cumb ] if @xmath97 is self - dual and @xmath681 is an umbilic hypersurface , then the cotton - york tensor of @xmath672 , @xmath717 , is identified to the restriction to @xmath672 of the self - dual cotton - york tensor @xmath646 of * m * : @xmath718 proof of the theorem .
the claimed identity is conformally invariant : if @xmath719 is a @xmath49-orthonormal oriented basis of * m * , then @xmath720 is a @xmath49-orthonormal basis on @xmath672 giving the orientation as above .
then @xmath721 , and , if we take @xmath722 , the claimed identity becomes @xmath723 where angle brackets denote the scalar product induced by @xmath49 .
the tensors @xmath724 , in the above form , are independent of the chosen metric @xmath49 @xcite , which depends on the normal vector @xmath254 , supposed to be @xmath49-unitary . if @xmath725 , for @xmath726 , then the corresponding metric @xmath727 , and also @xmath728 , thus the identity ( [ eq : cyw ] ) for @xmath729 is equivalent to the one for @xmath730 . as @xmath6 is the trace free component of the riemannian curvature contained in @xmath731 , and
is symmetric , it is enough to evaluate it on pairs @xmath732 which are unitary and orthogonal for the metric @xmath49 , therefore the check of the equation ( [ eq : cyw ] ) will prove the theorem . as @xmath733 are @xmath734-eigenvectors in @xmath735 ( the space of trace - free endomorphisms of @xmath34 ) , they are determined by the following formulas , where @xmath720 is any oriented orthonormal basis of @xmath693 : @xmath736 where @xmath719 is supposed to be a local extension , around a region of @xmath672 , of the @xmath49-orthonormal frame used in ( [ eq : cyw ] ) . as *
m * is supposed to be self - dual , @xmath2 is identically zero , thus , in the points @xmath737 , we have @xmath738 it is a standard fact that , if @xmath672 is umbilic , there is a local metric @xmath49 in the conformal class @xmath18 of * m * , such that , for @xmath49 , @xmath672 is totally geodesic .
we fix such a metric
. then we have @xmath739 which , together with [ eq : w1 ] , implies that @xmath740 . on the other hand , ( [ eq : tgeod ] ) , together with ( [ eq : w1 ] ) and ( [ arw- ] ) , yield @xmath741 let us compute now the normal derivative of @xmath6 in a point @xmath737 ; we suppose that @xmath719 are locally extended by an orthonormal frame , and that they are parallel at @xmath96 ( we omit , for simplicity of notation , the point @xmath96 in the following lines : @xmath742 from ( [ eq : w1 ] ) .
this is then equal to : @xmath743 from the second bianchi identity .
then we have @xmath744 from analogs of ( [ eq : rq ] ) .
then @xmath745 from ( [ eq : tgeod ] ) @xmath746 from ( [ h3 ] ) .
finally , from ( [ ar2 ] ) , we get @xmath747 this proves equation ( [ eq : cyw ] ) .
the corollary [ cumb ] now easily follows from the above theorem and ( [ ar+ ] ) .
the main results in section [ s.w5 ] hold also in the case of a conformal ( complex ) 3-manifold : [ compact3 ] let @xmath3 be a twistor space of a conformal 3-manifold @xmath748 ; let @xmath749 be its contact structure .
suppose there is a point @xmath750 such that , to any direction in @xmath751 , there is a tangent rational curve in @xmath3 with normal bundle @xmath4 .
then @xmath3 is projectively flat , and @xmath672 is conformally flat .
this follows directly from theorem [ twist ] , as @xmath672 is umbilic in * m * , the space of all projective lines in @xmath3 with the above normal bundle @xcite , and from theorem [ umbilic ] .
let @xmath672 be a conformal 3-manifold containing an immersed rational curve as null - geodesic .
then @xmath672 is conformally flat .
we can not use directly theorem [ umbilic ] , as the `` ambient '' self - dual manifold * m * can only be defined for a _ civilized _ ( e.g. geodesically connected ) 3-manifold .
hence , we follow the steps in proposition [ tub ] and prove [ tub3 ] null - geodesics close to a compact , simply - connected one are also compact and simply - connected , and they are embedded . still using the same arguments as in section [ s.w5 ] ,
we get an embedded null - geodesic @xmath752 diffeomorphic to @xmath132 , and we have : the deformations of @xmath8 as a compact curve coincide with the null - geodesics close to @xmath8 .
we cover @xmath8 with geodesically convex open sets @xmath753 , such that : @xmath754 where @xmath755 is still geodesically convex ( with respect to a particular levi - civita connection ) .
this is possible by choosing @xmath756 , small enough .
then we choose a relatively compact tubular neighborhood @xmath757 of @xmath8 , such that its closure is covered by the @xmath459 s .
we consider then the twistor spaces @xmath758 , the spaces of null - geodesics of @xmath459 .
the compact , simply - connected , null - geodesics close to @xmath8 identify ( diffeomorphically ) the neighborhoods of @xmath759 with the space @xmath3 of the deformations of @xmath8 as a compact curve .
we can see then @xmath3 as an open set common to all the @xmath758 s : following lebrun , we define the self - dual manifolds @xmath760 as the space of projective lines in @xmath758 , with normal bundle @xmath4 .
then @xmath459 is an umbilic hypersurface in @xmath760 .
the local twistor spaces @xmath758 admit contact structures , which coincide on @xmath3 , and contain projective lines @xmath761 corresponding to points @xmath762 .
if we denote by @xmath763 the twistor space of @xmath755 , then @xmath758 and @xmath764 are identified to open sets in @xmath763 , in particular the lines @xmath761 and @xmath765 are identified , thus their intersections with the common set @xmath3 coincide .
we denote by @xmath128 this ( non - compact ) curve in @xmath3 , and by @xmath78 the canonical contact structure of @xmath3 ( restricted from the ones of @xmath758 ) .
we already have obtained that the @xmath9-cone corresponding to @xmath751 is a part of a smooth surface : the union of the lines @xmath128 , @xmath186 , thus , from theorem [ 3 ] , the weyl tensor @xmath766 of the self - dual manifold @xmath760 vanishes on the @xmath9-planes generated by @xmath767 .
but this is nothing new : we know , from theorem [ umbilic ] , that @xmath766 vanishes on @xmath459 . we intend to apply theorem [ 3 ] to prove that @xmath766 vanishes on points close to @xmath459 , but in @xmath768 .
we do that by showing that the integral @xmath9-cones corresponding to planes @xmath769 are parts of smooth surfaces , then we conclude using theorem [ 3 ] .
first we choose hermitian metrics @xmath770 on @xmath758 , such that they coincide ( with @xmath397 ) on @xmath3 .
we have a diffeomorphism between @xmath8 and @xmath771 , so we choose relatively compact open sets in @xmath772 , covering @xmath771 , with the following properties : as the metrics @xmath770 induce metrics on @xmath760 , we first choose a small enough distance @xmath773 such that 1 .
@xmath774 , there is a sub - covering @xmath775 of @xmath8 such that the `` tubular neighborhoods '' @xmath776 are compact ( @xmath777 is the `` orthogonal projection '' for the hermitian metric from @xmath760 to @xmath778 ; it is well defined because of the condition below ) ; 2 .
@xmath408 is less than the bijectivity radius of the ( hermitian ) exponentials for the points of @xmath504 in @xmath760 , and for the points of @xmath779 in @xmath780 ( if @xmath781 ) .
we have then for any @xmath782 , @xmath783 such that the curves @xmath784 , @xmath785 are tangent to the same direction in @xmath750 , the respective curves @xmath786 coincide .
we first note that the projection @xmath777 from @xmath760 is equivalent to the @xmath397- orthogonal projection of the direction of @xmath787 to a direction in @xmath751 , so @xmath788 ; thus @xmath789 belongs to both @xmath459 and @xmath790 , and we use again the twistor space @xmath763 to conclude that @xmath791 and @xmath792 are `` restrictions '' to @xmath3 of the same projective line ( as they both have the same tangent space at @xmath168 ) @xmath793 , for a point @xmath794 .
now we have a tubular neighborhood @xmath795 of @xmath771 , of radius @xmath796 , such that , for any 2-plane @xmath797 , the conditions in theorem [ 3 ] are satisfied ( considering any of the local twistor spaces @xmath758 ) . we conclude that the weyl tensor @xmath766 of @xmath760 vanishes along all null - geodesics of @xmath760 , close ( in @xmath798 ) to @xmath8 and included in the @xmath0-surface @xmath799 , determined by @xmath8 .
this means that @xmath6 vanishes everywhere on @xmath799 . by deforming @xmath8
, we obtain that @xmath800 vanishes on a neighborhood of @xmath459 in @xmath760 , hence @xmath760 , as well as @xmath459 , are conformally flat ( by theorem [ umbilic ] ) .
it follows from theorem [ umbilic ] that @xmath672 is conformally flat .
the first example is the `` flat '' case : @xmath801 , with its canonical projective structure , and its space of projective lines @xmath802 .
( @xmath3 is equally the twistor space of the riemannian round 4-sphere , which is , therefore , a real part of @xmath803 . )
if @xmath395 , then the @xmath0-surface associated to it is the set @xmath804 . in this flat case
, we can equally define the @xmath9-twistor space @xmath805 , which is the dual projective 3-space @xmath806 , and an @xmath9-surface @xmath807 is the set @xmath808 .
a null - geodesic @xmath8 is then determined by a pair of _ incident _ @xmath9- , resp .
@xmath0-surface @xmath809 : @xmath810 @xmath9-surfaces and @xmath0-surfaces are diffeomorphic to @xmath13 , null - geodesics to @xmath132 , and the ambitwistor space @xmath7 is the `` partial flag '' manifold @xmath811 the flag manifold , of dimension 7 , is isomorphic to total space of the projective cone bundle over * m * , @xmath30 .
another example is when @xmath3 is the twistor space of the real riemannian manifold @xmath13 , with the fubini - study metric .
then @xmath3 is the manifold of flags in @xmath812 , @xmath813 , ( @xmath814 , resp .
@xmath815 are viewed as the space of lines , resp .
2-planes , in @xmath296 ) @xcite .
a projective line @xmath128 in @xmath3 is a set @xmath816 where @xmath817 belongs to @xmath818 , which is , therefore , the space * m * of such lines , and a conformal self - dual 4-manifold .
it can be naturally compactified within the space of analytic cycles of @xmath3 to @xmath819 , which is obviously a smooth manifold , but it carries no global conformal structure , as its canonical bundle has no square root .
this means that the conformal structure on @xmath820 is smooth on * m * , and _ singular _ on @xmath821 .
the cycles of @xmath3 corresponding to a point @xmath822 in this subset are pairs of complex projective lines in @xmath3 : @xmath823 a @xmath0-surface in * m * , corresponding to a point @xmath824 , is the set @xmath825 and can be naturally compactified to @xmath826 a vector in @xmath829 is a pair of vectors @xmath830 , with @xmath831 and @xmath832 , which satisfy a linear condition ( as @xmath833 ) . actually , there is a duality between @xmath815 , the grassmannian of 2-planes in @xmath296 , and @xmath834 , the projective space of @xmath835 , and an analogous one between @xmath814 and @xmath836 : @xmath837 then , the flag manifold @xmath827 is defined , as a submanifold of @xmath838 , by the equation @xmath839 the vector @xmath840 is a homomorphism in @xmath841 . by duality , @xmath842 .
then the vector @xmath843 lies in @xmath827 iff : @xmath844 or , equivalently , @xmath845 where @xmath846 is the projection ( as @xmath847 ) .
* @xmath848 is the `` degenerate '' 2-plane tangent to the 2 special curves @xmath849 , @xmath850 whose union is the special cycle @xmath851 corresponding to @xmath852 .
there are no projective lines @xmath853 , tangent to it ; only the special cycles @xmath854 and @xmath855 are tangent to @xmath856 , actually only to the two privileged directions of @xmath849 , resp . @xmath850 .
the special curves @xmath849 , @xmath850 have trivial normal bundle , being fibers of the projections from @xmath827 to @xmath814 , resp .
@xmath815 , so these special curves form two complete families of analytic cycles in @xmath827 , isomorphic to @xmath814 , resp . @xmath815 .
two such curves are incident iff they are of different types ( @xmath849 is _ of type _
@xmath296 , @xmath850 is _ of type _
@xmath857 ) , so they can only form `` polygons '' with an even number of edges .
but there are no quadrilaterals , as one can easily check , using the fact that @xmath849 and @xmath850 are incident iff @xmath847 , thus iff @xmath162 is a line in @xmath814 containing @xmath20 .
but there are hexagons , corresponding to the 3 vertices and 3 sides of a triangle in @xmath858 : the hexagon above is not `` flat '' , i.e. there is no canonical submanifold of @xmath827 containing it .
this , and the fact that there are no quadrilaterals made of @xmath859-type curves , is just a consequence of the fact that the distribution @xmath860 on @xmath157 is _ non integrable _ ; in fact it is the holomorphic _ contact structure _ induced by the fubini - study _
einstein metric _ on @xmath13 , @xcite , see also section [ ss.w86 ] .
* @xmath861 , for @xmath862 , @xmath863 .
this is a 2-plane that is tangent to only one of the special curves @xmath864 .
the projective lines tangent to @xmath865 at @xmath828 are @xmath866 , @xmath867 , hence the corresponding null - geodesic is @xmath868 thus it is diffeomorphic to @xmath17 , and its closure is @xmath869 the `` limit '' curve is @xmath870 , so it is non - singular at @xmath269 . actually , the points of @xmath866 _ close _ to @xmath269 converge , when @xmath871 , to some points in @xmath849 , which is tangent to @xmath865 .
we can , then , apply the same method as in theorem [ twist ] to conclude that the integral @xmath9-cone associated to @xmath865 is a smooth manifold _ around _
@xmath269 , thus , from theorem [ t1 ] , the weyl tensor @xmath6 of * m * vanishes on the @xmath0-planes generated , along @xmath872 , by its own direction .
we will see that the vanishing of @xmath6 on these @xmath9-planes leads to the existence of some @xmath9-surfaces , see below .
of course , the deformation argument in theorem [ twist ] does not hold in the present case , as the normal bundle of @xmath849 is trivial , thus different from the one of the rest of the rational curves @xmath866 ( as we will see below , generic 2-planes through @xmath269 do not admit projective lines tangent to all their directions ) .
@xmath878 , where @xmath879 is a projective diffeomorphism such that @xmath880 . indeed , the tangent spaces @xmath881 and @xmath882 are isomorphic to @xmath883 , resp . to @xmath884 , and
a generic 2-plane @xmath78 in @xmath829 is the graph of a linear isomorphism @xmath885 satisfying a linear condition ( [ eq : tflag ] ) or ( [ eq : tflag1 ] ) .
actually , the graph is determined by the projective application @xmath886 induced by @xmath887 from @xmath888 to @xmath889 : the condition @xmath880 is implied by ( [ eq : tflag1 ] ) .
the null - geodesic associated to the 2-plane @xmath890 is @xmath891 and its closure in @xmath820 is @xmath892 hence the `` limit '' point is @xmath893 , corresponding to the special cycle @xmath271 , none of whose components is tangent to @xmath890 .
the integral @xmath9-cone associated to @xmath890 looks like suggested in the picture below : the application @xmath886 has the following interpretation in terms of projective geometry on @xmath894 : a direction @xmath895 in @xmath882 is identified to the point @xmath896 and a direction @xmath897 is identified to a direction ( thus a projective line @xmath161 ) through @xmath898 .
@xmath886 is , thus , a _
homography _ that associates to @xmath899 ( we identify @xmath162 with the projective line @xmath900 ) the line @xmath901 . as @xmath902 , we have , then , that three points @xmath903 belong to the same null - geodesic iff @xmath904 i.e. the cross - ratio of the points @xmath905 equals the cross - ratio of the lines @xmath906 through @xmath20 ( the dotted lines , and their intersections with the lines @xmath907 are the points in the integral @xmath9-cone ) : we can now describe the null - geodesics passing through a point @xmath908 and contained in a @xmath0-surface @xmath909 , whose closure @xmath145 is isomorphic to @xmath29 : they coincide with the rational curves in @xmath145 , containing @xmath910 ; except the `` horizontal '' ( @xmath911 ) and `` vertical '' ( @xmath912 ) ones , all these curves contain @xmath269 : we remark that , in the usual affine coordinates on @xmath913 , these null - geodesics are the affine lines containing @xmath910 , thus the _ projective structure _ on @xmath0 is ( locally ) isomorphic to a flat affine structure .
we have seen , in section [ s.w6 ] ( corollary [ prflat ] ) , that this is true for all @xmath0-surfaces of a self - dual manifold .
let us study now the conformal structure of @xmath914 directly ; actually * m * has a complex metric @xmath49 .
let @xmath908 , then @xmath915 is transverse to @xmath161 , thus we have the isomorphisms @xmath916 and @xmath917 .
then , a vector @xmath918 is identified to a pair of homomorphisms @xmath919 and @xmath920 .
then the metric @xmath49 is given by @xmath921 let @xmath397 be a hermitian metric on @xmath296 .
then we have a real - analytic embedding of @xmath922 into * m * , given by : @xmath923 a vector @xmath924 is tangent to @xmath11 iff @xmath925 then one easily checks that @xmath926 , hence , up to a constant , the restriction of @xmath49 to @xmath927 is the fubini - study metric of @xmath928 . an isotropic vector in * m * is @xmath918 , such that @xmath929 , viewed as an endomorphism of @xmath915 ( see above ) , or , equivalently , such that @xmath930 let us see which is the limit of the isotropic cone in the points of @xmath827 : from the relation above , it follows that the isotropic cone in a point @xmath931 is @xmath932 so the conformal structure of * m * is _ singular _ at the `` infinity '' @xmath827 .
the situation @xmath833 is very similar to the one treated in section [ s.w7 ] , see also @xcite : @xmath838 has an einstein self - dual metric @xmath49 , singular at the `` infinity '' , and this einstein structure yields a contact structure on the twistor space @xmath157 ; the field of 2-planes determined by this contact structure corresponds to the `` infinity '' @xmath933 . but
these planes do not admit tangent rational curves , with normal bundle @xmath4 : the conformal structure does not extend to the `` infinity '' ( which is , therefore , not a _
conformal infinity _ ) .
we consider the isotropic planes in @xmath934 ( @xmath935 ) : for a fixed isotropic direction , represented by a generic vector @xmath936 , the line @xmath937 and the plane @xmath938 are fixed .
the linear space of all vectors @xmath939 satisfying @xmath940 is isotropic and orthogonal to @xmath830 : they form a @xmath0-plane .
the @xmath9-plane @xmath41 containing @xmath830 corresponds to the isotropic vectors @xmath941 , orthogonal to @xmath830 , with @xmath942 . as a plane transverse to all the @xmath0-planes
( whose projection onto @xmath943 or @xmath944 is never injective ) , @xmath41 is determined by a linear isomorphism @xmath945 , whose graph in @xmath946 is @xmath41 ; @xmath886 induces the application @xmath947 between the projective spaces of @xmath943 , resp . @xmath944 .
the plane @xmath948 , the graph of @xmath886 , is isotropic iff @xmath949 , @xmath950 , i.e. @xmath951 is the homography that sends a point @xmath274 in @xmath161 into the projective line through @xmath915 and @xmath274 .
we can extend @xmath886 to a projective isomorphism @xmath952 : for example , @xmath953 contains @xmath943 as an affine open set .
then @xmath954 is defined as follows : @xmath955 actually @xmath956 and @xmath957 .
we have then : a generic @xmath9-plane @xmath948 in @xmath934 is the graph of a linear isomorphism @xmath958 , which is determined by a projective isomorphism @xmath959 such that @xmath960 and @xmath961 , for all @xmath962 .
the exponential @xmath963 has an interpretation in terms of projective geometry : each direction @xmath964 is determined by the point @xmath965 in @xmath966 and the line through @xmath915 and @xmath965 , and a homography @xmath967 from the points @xmath7 of the projective line @xmath968 to the space of lines @xmath969 through @xmath965 ( see next picture and the convention below ) . as
this homography is the restriction of @xmath954 to the appropriate spaces , it follows that it is related to the homography @xmath970 , where @xmath971 is another direction in @xmath890 : the points @xmath972 , @xmath973 and @xmath915 are collinear : of course , this implies that @xmath433 determines a homography @xmath974 between the lines @xmath968 and @xmath975 , such that @xmath976 and @xmath977 .
then , for any other points @xmath978 , the lines @xmath979 intersect on the line @xmath980 ( see the right hand side of the picture above ) .
* convention . * in the framework of plane projective geometry , we identify a point in @xmath815 with a line in @xmath814 ( we note , for example @xmath981 ) .
the lines determined by the distinct points @xmath7 and @xmath631 will be denoted by @xmath982 ( thus @xmath983 ) . the null - geodesic tangent to @xmath830 at @xmath910 is the set @xmath984 , and the null - geodesic tangent to @xmath941 is the analogous set of the pairs @xmath985 . thus @xmath986 where @xmath987 , and @xmath988 , as in the picture above ( where @xmath989 , @xmath990 ) .
this gives the exponential of the @xmath9-plane determined by the isotropic vector @xmath830 .
we remark that the point @xmath910 has a privileged position in @xmath991 : @xmath992 @xmath993 ; on the other hand @xmath994 in general ( see the picture above ) , which means that the points @xmath995 and @xmath985 are not _ null - separated _
( i.e. they do not belong to the same null - geodesic ) .
that means that @xmath991 is not totally isotropic , thus there is no @xmath9-surface tangent to a generic @xmath9-plane ; not surprising as the corresponding @xmath9-cone is not flat ( see section [ ss.w84 ] ) .
but there are @xmath9-surfaces tangent to the two @xmath9-planes @xmath996 and @xmath997 : the `` slices '' @xmath998 and @xmath999 .
( it is easy to see that these planes are isotropic , and that they are not @xmath0-planes , as these project on _ lines _ in @xmath943 , resp . @xmath944 . )
thus @xmath914 is a conformal self - dual manifold , not anti - self - dual , that admits @xmath9-surfaces passing through any point . | we study complex 4-manifolds with holomorphic self - dual conformal structures , and we obtain an interpretation of the weyl tensor of such a manifold as the projective curvature of a field of cones on the ambitwistor space . in particular , its vanishing is implied by the existence of some compact , simply - connected , null - geodesics .
we also relate the cotton - york tensor of an umbilic hypersurface to the weyl tensor of the ambient . as a consequence , a conformal 3-manifold or a self - dual 4-manifold admitting a rational curve as a null - geodesic is conformally flat .
we show that the projective structure of the @xmath0-surfaces of a self - dual manifold is flat .
[ 1 ] # 1 . _ _ [ 1 ] # 1 . _ _ [ 1 ] # 1 . _ _ [ 1 ] # 1 . _
_ l _ 3 ^ 3 |
the atomic gases in many cold atom experiments are confined in harmonic traps . an important consequence of this kind of confinement is that , in the absence of any additional external perturbation , the centre of mass of the system oscillates about the centre of the trap in simple harmonic motion without dissipation .
this particular collective oscillation is referred to as the centre of mass or dipole mode . according to the generalized kohn theorem @xcite ,
this behaviour is a generic property of a harmonically - confined system in which the interactions between particles depend only on their relative separation , and is independent of other intrinsic properties such as dimensionality , quantum statistics and the state of internal excitation . for these reasons ,
the undamped dipole oscillation can in fact be used to accurately determine the trapping frequencies @xcite in situations where the experimental parameters defining the trapping potential are not known precisely .
an additional , but more subtle , implication of such confinement is the content of the so - called harmonic potential theorem ( hpt ) @xcite .
in essence , the hpt demonstrates the existence of a class of dynamical many - body states for which the probability density moves without change in shape .
this theorem imposes important constraints on the form of approximate theories which deal with the dynamics of harmonically - confined many - body systems @xcite .
when the harmonicity of the confining potential is compromised , however , the centre of mass is coupled to the internal degrees of freedom and its dynamics becomes sensitive to the intrinsic properties of the system , including the specific form of the particle interactions .
for this reason , the dipole oscillation can serve as an experimental diagnostic of various perturbations acting on the system .
for instance , several experiments @xcite have used dipole oscillations to study the transport of a bose - condensate through a disordered medium or past a localized impurity . although the motion of the condensate in these experiments does not lose its collectivity
, dissipation does occur and leads to the damping of the centre of mass motion .
another experimental example is provided by the dipole oscillation of a trapped bose gas in the presence of an optical lattice potential @xcite .
here it was found the that dimensionality of the bose gas plays a critical role in determining the way in which the centre of mass behaves as a function of time . in all of these experiments ,
the dipole oscillation of the atomic system is initiated by an abrupt displacement of the trapping potential along a certain axial direction .
if the displacement is large , the system begins its evolution in a highly non - equilibrium initial state .
it is partly for this reason that much of the theoretical work dealing with the collective dynamics of bose - condensed systems relies on numerical simulations of the time - dependent gross - pitaevskii ( gp ) equation @xcite .
one of the goals of this paper is to show that this nonequilibrium dynamics in the presence of the external perturbation can be considered from a different point of view when the system is harmonically confined . by means of an appropriate transformation , one can equivalently think of the system as being driven out of an initial _ equilibrium _ state by a _ dynamic _ external perturbation oscillating sinusoidally at the frequency of the trap .
the availability of this alternate point of view is a consequence of what we refer to as the _ extended _ hpt .
its advantage is that the external perturbation can be treated by conventional linear response theory , at least when the perturbation is sufficiently weak .
this approach was used effectively in an earlier paper @xcite to determine the damping of the centre of mass motion due to a disorder potential .
a second purpose of this paper is to study the response of a harmonically - confined system to an external potential which is made to oscillate at an _ arbitrary _ frequency .
our discussion is motivated by a recent proposal @xcite to probe the optical conductivity of a cold atomic gas in an optical lattice by shaking the lattice periodically along a certain direction .
this is an interesting idea since it provides a method of addressing experimentally the optical conductivity of a system consisting of _ neutral _ atoms .
however , the authors of ref .
@xcite only considered bosons within a _ uniform _ lattice , while in most experiments the atoms are also subjected to a harmonic potential . in this paper
, we show that one can also probe directly the optical conductivity of a gas that experiences a _ combination _ of a harmonic trapping potential and an arbitrary external potential when the latter is made to oscillate sinusoidally with a small amplitude .
this generalization provides a precise link between theoretical calculations of the optical conductivity and possible experimental measurements on harmonically - confined gases in the presence of various external perturbations .
the rest of the paper is organized as follows . in sec . 2
, we provide a basic discussion of the dipole modes of a harmonically - confined system . in sec . 3
, we consider the response of a harmonically - confined system to a time - dependent homogeneous force .
an explicit expression for the evolution operator of the system is obtained which motivates the introduction of a rather useful unitary displacement operator .
these results are then used to provide an alternative derivation of the hpt .
we next consider in sec .
4 situations in which the system is perturbed by an additional external potential that couples the centre of mass and internal degrees of freedom . here
we present a derivation of the extended hpt . in sec . 5
, we consider the energy absorption rate and centre of mass dynamics of a harmonically - confined gas that is subjected to an oscillating external potential and demonstrate that both aspects serve to probe the optical conductivity of the system .
all of our findings are summarized in sec .
as a preliminary to the the development of the hpt and its extension , we discuss in this section the dipole modes of a harmonically - confined system and the underlying physics for the existence of such modes .
the dipole modes are the low - lying collective excitations that have frequencies equal to the frequencies of the trap . unlike other low - lying excitations , the frequencies of these modes are independent of the total number of trapped atoms and the atomic interactions . for bose - condensed systems ,
these modes are often discussed in the context of mean - field theory , namely as solutions to the time - dependent gp equation .
however , it can be shown rigorously that such dipole modes exist for any harmonically - confined ( bosonic or fermionic ) system in which interactions depend only on the relative coordinates of the particles .
the fundamental reason behind this is that the centre of mass degree of freedom is separable from all the internal degrees of freedom , which implies that there are excitations associated solely with motion of the centre of mass . to demonstrate this
, we consider a harmonically - confined many - body system described by the generic hamiltonian h_0 = _ i=1^n ( + v_trap ( _ i))+_i < jv(_i-_j ) , [ h5 ] where the harmonic trapping potential is v_trap ( ) = m_=x , y , z_^2 r_^2 .
we define the centre of mass co - ordinate @xmath0 and the total momentum operator of the system @xmath1 .
since these two operators satisfy the commutation relation @xmath2=i\hbar \delta_{\mu\nu}$ ] , they are canonically conjugate variables . introducing the relative variables @xmath3 and
@xmath4 , we observe that the hamiltonian @xmath5 can be written as h_0 = h_cm+h_int . here
h_cm=+m_=x , y , z^2_r_^2 [ h_cm ] and h_int=_i=1^n ( + v _
trap(_i))+_i < jv(_i-_j ) , where @xmath6 is the total mass of the system .
the hamiltonian for the centre of mass degree of freedom @xmath7 is that of a harmonic oscillator ; @xmath8 is the hamiltonian determining the internal dynamics of the system .
one can check that the centre of mass and relative variables commute , namely = 0;=0 .
[ rp_com ] it follows from these results that the centre of mass hamiltonian @xmath7 commutes with @xmath8 .
this means that the motion of the centre of mass decouples from the internal dynamics of the system .
one simple implication of this decoupling is that the centre of mass exhibits simple harmonic motion at the frequencies of the trap . using eq .
( [ h_cm ] ) and ( [ rp_com ] ) , the heisenberg equations of motion for the centre of mass coordinate and the total momentum are @xmath9=\frac { \hat { p}_{\mu,\rm i}(t)}{m } , \\
\frac{d\hat{p } _
{ \mu,\rm i}(t)}{dt}&=\frac{1}{i\hbar}[\hat{p}_{\mu,\rm i}(t),\hat{h_0}]=-m\omega_\mu^2\hat{r}_{\mu,\rm i}(t ) , \label{eqp}\end{aligned}\ ] ] where @xmath10 and @xmath11 .
equations ( [ eqr ] ) and ( [ eqp ] ) lead to the simple harmonic motion equation + _
^2 r _ , i(t)=0 .
the formal solution of this equation is @xmath12 from this we see that the expectation value @xmath13 for an arbitrary state @xmath14 evolves in time according to the equation r_(t)=r_(0)_t + _ t = a(_t+_0 ) , [ cm_osc ] where the amplitude @xmath15 and phase angle @xmath16 are determined by the initial conditions @xmath17 and @xmath18 .
this undamped harmonic oscillation of the centre of mass coordinate is the dipole oscillation we have been referring to .
the centre of mass excitations are conveniently described by defining the centre of mass annihilation and creation operators a_=(r _ + p _ ) ; a^_=(r _ -p _ ) .
[ a_cm ] in terms of these operators , the hamiltonian @xmath5 takes the form h_0= _ _ ( a^_a_+ ) + h_int . from eq .
( [ rp_com ] ) and the definition in eq .
( [ a_cm ] ) , we observe that @xmath19 and @xmath20 commute with @xmath21 .
taking @xmath22 to be an eigenstate state of @xmath5 with energy @xmath23 , we find that the state @xmath24 satisfies h_0 |=(e_+n_)| , that is , @xmath25 remains an eigenstate of @xmath5 with energy @xmath26 ; the application of @xmath20 creates a quantum of excitation of the centre of mass oscillation with energy @xmath27 . finally , we mention the well known fact @xcite that a simple harmonic oscillator has wave packet quantum states that move harmonically without any change in shape .
these states are the so - called coherent states .
since the centre of mass degree of freedom is effectively a harmonic oscillator , analogous states also exist for the many - body system described by the hamiltonian @xmath5 in eq .
( [ h5 ] ) .
this property is encapsulated by the hpt discussed in the next section .
the separability of the centre of mass and internal degrees of freedom leads to excitations which are associated solely with the motion of the centre of mass .
we now discuss a further implication of this property , namely the existence of a class of dynamical many - body states for which the probability density moves without change in shape . as shown by dobson in his proof of the harmonic potential theorem @xcite
, this behaviour can occur not only for systems described by the hamiltonian in eq .
( [ h5 ] ) , but also when the system is subjected to an arbitrary time - dependent , but spatially homogeneous , force .
in this situation , the system is described by the hamiltonian h_f(t)=h_0-(t)_i=1^n_i= h_0-n(t ) .
[ hwf ] in the presence of the force @xmath28 , the system is still subjected at any instant of time to a purely harmonic , albeit time - dependent , confining potential .
it is in this sense that we use the phrase purely harmonic confinement " to distinguish this situation from those we consider later in which additional perturbing potentials are also present . before dealing with these situations ,
we first determine the dynamical state @xmath14 which evolves in time according to the hamiltonian @xmath29 .
this result will then be used to provide an alternative derivation of the hpt .
the time evolution of @xmath14 is formally given by @xmath30 satisfies the equation i(t)=h_f(t)(t ) with the initial condition @xmath31 . to determine @xmath32
, we go to the interaction picture and define _ i(t)=e^ih_0 t/(t ) , [ evoopint ] which also has the initial condition @xmath33 .
this evolution operator satisfies the equation i_i(t)=-n_f_(t ) r_,i ( t ) _
i(t ) , [ evolopeq ] where @xmath34 is given explicitly in eq . ( [ solr ] ) .
the formal solution of eq .
( [ evolopeq ] ) can be written as @xmath35 where @xmath36 .
the second line follows from the fact that @xmath34 and @xmath37 commute when @xmath38 . the product of operators in eq .
( [ uifsol ] ) can be evaluated recursively .
starting from the right and making use of the baker - hausdorff formula if @xmath39 and @xmath40 commute with @xmath41 $ ] . ]
, we find @xmath42\delta t\right \}\exp\left \{\frac{in}{2m\hbar\omega_\mu}f_\mu(\delta t)f_\mu ( 0)\sin\omega_\mu\delta t(\delta t)^2\right \}.\end{aligned}\ ] ] to arrive at this result we have noted that @xmath43=\frac{i\hbar}{nm\omega_\mu}\sin\omega_\mu(t'-t ) , \label{r_commu}\end{aligned}\ ] ] which is obtained using eq .
( [ solr ] ) . after repeating these steps @xmath44 times
, we must consider in the next step @xmath45 it is clear that a phase factor of the kind given by the second exponential appears at each step of the process .
accumulating these phase factors and converting the summation into an integral , we find that eq .
( [ uifsol ] ) becomes @xmath46 this result for @xmath47 defines the dynamic state in eq .
( [ dystate ] ) for an arbitrary initial state @xmath48 . with @xmath49 given by eq .
( [ solr ] ) , the first exponential factor in eq .
( [ u_i ] ) takes the form of the unitary displacement operator ( , ) \{(- ) } , [ t_ope ] where @xmath50 is a position vector and @xmath51 is a momentum vector .
this operator can be viewed as a generalization of the usual translation operator @xmath52 @xcite . when applied to some arbitrary state , the operator @xmath53 has the effect of shifting the state by @xmath50 in position space and by @xmath54 in momentum space . to see this , we make use of the baker - hausdorff formula to obtain @xmath55 defining the state @xmath56 and using eq .
( [ txpbh1 ] ) , one finds @xmath57\psi(\br_1-\bx,\cdots,\br_n-\bx ) , \label{psi'r } \end{aligned}\ ] ] where @xmath58 .
we thus have @xmath59 .
similarly , by interchanging the final two exponentials in eq .
( [ txpbh1 ] ) , one can show that the momentum - space wavefunction is @xmath60\tilde\psi(\bp_1-\bp / n,\cdots,\bp_n-\bp / n ) , \label{psi'p } \end{aligned}\ ] ] where @xmath61 . thus latexmath:[$|\tilde
\psi'(\bp_1, ... ,\bp_n)|^2 = _ total _ momentum of the state is boosted by @xmath51 .
furthermore , it is straightforward to demonstrate the operator displacement properties @xmath63 we will make use of these transformation properties in the following .
we now consider the dynamical evolution of the system when prepared in the initial state |(0)=(,)| , [ psi_0_a ] where @xmath64 is an arbitrary many - body state . as discussed above , this initial state is simply the state @xmath64 translated rigidly in position space through the vector @xmath50 and given a total momentum boost of @xmath51 . at the end of this section
we shall explain how such an initial state can be realized in cold atom experiments .
we next show how eq .
( [ u_i ] ) together with eq .
( [ psi_0_a ] ) leads to the hpt .
we have @xmath65 here , the product @xmath66 can again be evaluated using the baker - hausdorff formula .
i(t)t(,)=e^-is(t)/ \ { } , [ ut ] where the phase @xmath67 is s(t)=-(t ) -__0^t dt_0^tdt f_(t)f_(t ) _ ( t-t ) , with ( t ) = _ _ 0^t dt nf_(t)substituting eq .
( [ ut ] ) into eq .
( [ dystateii ] ) , we find @xmath68
\right \}e^{-i\hat h_0 t/\hbar}|\phi\ra \nn \\ & = e^{-is(t)/\hbar}\hat t\left ( \bx(t),\bp(t)\right ) e^{-i\hat h_0 t/\hbar}|\phi\ra , \label{dynamicstate}\end{aligned}\ ] ] where eqs .
( [ solr ] ) and ( [ solp ] ) are used to obtain the final result . here
@xmath69 and @xmath70 are given by @xmath71 where @xmath72 from these expressions we see that @xmath69 is simply the solution of the forced harmonic oscillator equation mx_(t)+m_^2 x_(t)=nf_(t ) [ cdoeq ] with the initial conditions @xmath73 and @xmath74 .
likewise , @xmath75 is the solution in the absence of the force , again with the initial conditions @xmath76 and @xmath77 . in terms of the solution of eq .
( [ cdoeq ] ) , the phase @xmath67 can be simplified as s(t)=-_0^t dt ( t)(t ) .
equivalently , this can be written as s(t)=_0^t dt _ , which is the classical action of a harmonic oscillator .
equation ( [ dynamicstate ] ) shows that the evolution of the initial state in eq .
( [ psi_0_a ] ) can be considered as taking place in two steps .
first , the state @xmath64 evolves _ freely _ for a time @xmath78 , that is , in the absence of @xmath28 .
the state @xmath79 is then displaced by @xmath80 to generate , apart from a phase factor , the final dynamical state of interest . since the operator @xmath81 shifts a state in position space by @xmath82 , the wave function corresponding to the state @xmath83 in eq .
( [ dynamicstate ] ) is @xmath84 , where @xmath85 is some time - dependent phase angle .
dobson s hpt now follows by choosing @xmath64 to be @xmath22 , an eigenstate of @xmath5 . in this case one
has @xmath86 , which implies that the probability density simply moves rigidly , following the trajectory of the centre of mass motion given by @xmath82 in eq .
( [ xt ] ) .
this property does not apply to an arbitrary initial state which evolves according to eq .
( [ dystate ] ) .
so far we have only discussed the evolution of a many - body eigenstate .
more generally , we can consider a system described by a statistical density matrix of the form _ 0=_k_k|_k_k| .
a density matrix of this form encompasses the case of a system in thermal equilibrium .
we now imagine that all of the states are displaced by @xmath53 at time @xmath87 . the resulting density matrix is then given by @xmath88 the time evolution of this density matrix is given by ( t)=(t)(0)^(t ) = ( t ) t ( , ) _ 0 t^ ( , ) ^(t ) . comparing eqs .
( [ dystateii ] ) and ( [ dynamicstate ] ) , we have the operator identity ( t ) t ( , ) = e^-s(t ) ( ( t),(t ) ) e^-ih_0 t/ [ identity ] where eqs .
( [ xt ] ) and ( [ pt ] ) define the time evolution of the displacement operator .
we thus find ( t ) = ( ( t),(t))e^-ih_0 t/ _ 0e^ih_0 t/^((t),(t ) ) .
this density matrix corresponds to each state @xmath89 evolving freely for a time @xmath78 and then being displaced along the forced oscillator trajectory . if the states @xmath90 in @xmath91 are in fact eigenstates @xmath22 of @xmath5 , we have the simpler result ( t)=((t),(t))_0^((t),(t ) ) . with this density matrix ,
the time - dependent density of the system is @xmath92 \nn \\ & = \text{tr}[\hat { t}(\bx(t),\bp(t))\hat \rho_0\hat { t}^{\dag}(\bx(t),\bp(t))\hat n(\br ) ] \nn \\ & = \text{tr}[\hat{\rho}_0\hat t^\dag(\bx(t),\bp(t))\hat{n}(\br)\hat t(\bx(t),\bp(t ) ) ] .
\end{aligned}\ ] ] recalling that @xmath93 and using eq .
( [ tft ] ) , we find @xmath94 \nn \\ & = n_0(\br-\bx ( t)),\end{aligned}\ ] ] where @xmath95 $ ] is the density of the system before its displacement .
we thus see that the density of the system experiences the same kind of rigid motion in the density matrix description as it does for a pure state . in the rest of this section
, we consider the special case in which the force is absent .
we then find from eq .
( [ dynamicstate ] ) that the dynamical state of the system at time @xmath78 is given by @xmath96 and @xmath97 are given by eqs .
( [ x0 t ] ) and ( [ p0 t ] ) .
this result is of course consistent with the general result in eq .
( [ cm_osc ] ) for the dynamics of the centre of mass coordinate .
( [ psi_t_alpha ] ) also implies that the displacement operator evolves according to t ( _ 0(t),_0(t))=e^-ih_0 t/ t ( , ) e^ih_0 t/ [ emt ] which is the force - free analogue of eq .
( [ identity ] ) .
as we alluded to earlier , one can also understand the result in eq .
( [ psi_t_alpha ] ) from the perspective of coherent states . for simplicity
, we take @xmath98 and @xmath99 in eq . ( [ psi_0_a ] ) . using eq .
( [ a_cm ] ) , the initial state in eq .
( [ psi_0_a ] ) can be written as |(0)=e^-a_z+^ * a^_z| _ , where @xmath100 .
we recognize this state as the analogue of a coherent state of a simple harmonic oscillator @xcite .
its dynamics is then given by @xmath101 where a_z , i(t)e^ih_0 t / a_z e^-i h_0t/=e^-i_z ta_z .
[ azt ] substituting eq .
( [ azt ] ) into eq .
( [ psitco ] ) and using eqs .
( [ a_cm ] ) , we recover eq .
( [ psi_t_alpha ] ) where @xmath102 and @xmath103 .
finally we discuss the experimental realization of the initial state in eq .
( [ psi_0_a ] ) .
the basic idea is to initiate the oscillatory motion of a harmonically - confined system by a sudden displacement of the trapping potential . to be specific ,
we take the state of the undisplaced potential to be the eigenstate @xmath22 . if the trapping potential is then displaced in the @xmath104-direction by an amount @xmath105 , the state of the system _ relative to _ the shifted potential is given by |(0)=t ( = z,=0)|_. at a later time @xmath106 , the state of the system according to eq .
( [ psi_t_alpha ] ) is |(t_0)=e^-e_t_0 t ( _ 0(t_0),_0(t_0))| _ , where , from eqs .
( [ x0 t ] ) and ( [ p0 t ] ) , @xmath107 and @xmath108 . with an appropriate choice of the initial displacement @xmath104 and time @xmath106 , we can achieve the initial conditions @xmath109 and @xmath110 .
a more elaborate sequence of displacements of the trap in different directions can in principle be used to achieve arbitrary initial conditions .
the theoretical development in this section is motivated by several recent experiments @xcite which studied the centre of mass dynamics of trapped bose condensates in the presence of a disorder potential .
the disorder in these experiments is an example of an external perturbation which couples the centre of mass and internal degrees of freedom . as a result of this coupling , the energy associated with the centre of mass motion
is transferred to internal excitations , in other words , mechanical energy is converted into ` heat ' .
one can think of the external perturbation as effectively exerting a drag force on the centre of mass which leads to a damped oscillation .
although the external potential acting on the system can be quite arbitrary , for illustration purposes we will occasionally visualize it as a disorder potential in the following .
a harmonically - confined system in the presence of an additional external potential @xmath111 is governed by the hamiltonian h = h_0 + _ i=1^nv_ext(_i ) h_0 + v_ext .
[ h_static ] to investigate the dynamics of the centre of mass , we consider the heisenberg equations of motion ( the heisenberg operators here are defined with respect to the full hamiltonian @xmath112 ) @xmath113=\frac { \hat { p}_{\mu}(t)}{m } , \\ \frac{d\hat{p}_{\mu}(t)}{dt}&=\frac{1}{i\hbar}[\hat{p}_{\mu}(t),\hat{h } ] = -m\omega_\mu^2\hat{r}_\mu(t)+\hat f_\mu(t ) , \label{eqpdis}\end{aligned}\ ] ] where f_=-_i=1^n is the @xmath114-component of the external force operator .
equations ( [ eqrdis ] ) and ( [ eqpdis ] ) then lead to + _ ^2
r_(t ) = . [ rf ] taking the expectation value of both sides of eq . ( [ rf ] ) with respect to the initial state @xmath48
, we find that the @xmath104-component of the centre of mass position satisfies the equation + _
z^2 z(t ) = , [ cofmeq1 ] where @xmath115 and f(t)=(0)| f_z(t ) |(0 ) = ( t)| f_z oscillator considered earlier .
however here , @xmath116 is not given explicitly but is defined by the dynamical state @xmath14 .
the dynamical evolution of the state @xmath14 is governed by the time - dependent schrdinger equation i|(t)=h |(t ) .
[ psi_evol ] the physical situation of interest is one in which the dynamics is initiated by suddenly shifting the harmonic trapping potential at some instant of time as discussed at the end of the previous section . for this reason
, we will consider an initial state of the form given in eq .
( [ psi_0_a ] ) .
the state @xmath64 being displaced , will be specified later when we consider various experimental protocols for its preparation . the following development does not depend on the specific choice of the state @xmath64 . in the interaction picture , the state @xmath117 satisfies the equation i|_i(t)=e^ih_0
t/ v_exte^-ih_0 t/ |_i(t ) .
[ psi_i_evol ] we now define the state |_i(t)= ^(,)|_i(t)which has the initial value @xmath118 .
this state satisfies the equation @xmath119 ) and ( [ tft ] ) .
the above equation can be interpreted as the interaction picture evolution of the state @xmath120 which satisfies the equation i|(t)=(h_0 + _
i=1^nv_ext(_i + _ 0(t)))|(t)h(t)|(t ) , [ tilde_psi_evol ] with the initial condition @xmath121 .
the states @xmath122 and @xmath14 are related by |(t)= ( _ 0(t),_0(t ) ) |(t ) .
[ transf ] we thus see that the state @xmath14 of interest , which evolves from @xmath123 according to the stationary hamiltonian @xmath124 , can be obtained by a displacement of the state @xmath125 via eq .
( [ transf ] ) . the latter state corresponds to a different physical situation in which the system starts in the state @xmath64 and then evolves in the presence of a _ dynamic _ potential oscillating at the trap frequency .
equation ( [ transf ] ) is the main result of this section .
it in fact reduces to the result given in eq .
( [ psi_t_alpha ] ) obtained in the context of the hpt when @xmath126 . for this reason
, we refer to it as the _ extended hpt_. its utility will become clear in the subsequent discussion . , the condensate is in its ground state with respect to the trapping potential indicated by the dashed curve . at @xmath127 , the trapping potential
is suddenly shifted to the origin , initiating the free oscillation of the condensate .
( b ) at @xmath87 , the condensate arrives at @xmath128 with the velocity @xmath129 and the external potential is suddenly switched on . for @xmath130 ,
the condensate evolves according to the hamiltonian in eq .
( [ h_static ] ) .
( c ) the external potential has a velocity @xmath131 at @xmath87 and subsequently oscillates according to the free centre of mass motion .
the condensate , initially stationary , begins to respond to the _ dynamic _ external potential.,title="fig : " ] , the condensate is in its ground state with respect to the trapping potential indicated by the dashed curve . at @xmath127 , the trapping potential
is suddenly shifted to the origin , initiating the free oscillation of the condensate .
( b ) at @xmath87 , the condensate arrives at @xmath128 with the velocity @xmath129 and the external potential is suddenly switched on . for @xmath130 ,
the condensate evolves according to the hamiltonian in eq .
( [ h_static ] ) .
( c ) the external potential has a velocity @xmath131 at @xmath87 and subsequently oscillates according to the free centre of mass motion .
the condensate , initially stationary , begins to respond to the _ dynamic _ external potential.,title="fig : " ] to give a concrete example of these general ideas , we consider the effect of a disorder potential on the dynamics of a harmonically - confined bose - condensed gas .
[ oscillation ] illustrates two possible protocols for the initiation of the dynamics @xcite . in the left panel
( a ) , we start with the condensate initially in its ground state with respect to the trapping potential ( dashed curve ) and the external ( disorder ) potential . at @xmath87 , the trapping potential
is suddenly shifted to the origin ( solid curve ) which initiates the centre of mass oscillation .
the initial state @xmath48 is the ground state of the condensate in the total potential v_trap(- ) + v_ext ( ) , where @xmath132 .
this state can be expressed as state in the total potential v_trap ( ) + v_ext(+ ) .
this latter potential together with the state @xmath133 is illustrated in ( b ) of the left panel .
the initial state @xmath133 evolves into the state @xmath122 according to eq .
( [ tilde_psi_evol ] ) . in the situation illustrated in the right panel of fig .
[ oscillation ] , the condensate starts off in the ground state of the harmonic potential which is shifted to the origin at some instant of time ( a ) .
the state is then allowed to evolve freely for some interval of time after which the state of the condensate is given by |(0)= t ( , ) |_0 , where @xmath133 is now the ground state of @xmath5 ( with the harmonic potential @xmath134 centred on the origin ) and the displacement operator @xmath53 determines the position and velocity of the condensate at the time @xmath87 as shown in ( b ) of the right panel of fig .
[ oscillation ] . at this instant ,
the disorder potential is switched on and the system evolves according to the hamiltonian @xmath112 .
we thus see that in both scenarios illustrated in fig .
[ oscillation ] , the initial state can be expressed in the form shown in eq .
( [ psi_0_a ] ) .
although the two initial states are different , the subsequent evolution for @xmath130 ( left panel ( a ) and right panel ( b ) ) takes place in both cases according to the hamiltonian @xmath112 to generate the state @xmath14 .
this state is related to the state @xmath135 through eq .
( [ transf ] ) .
this relationship implies that the force appearing in eq .
( [ cofmeq1 ] ) can be expressed as @xmath136 where we have used eq .
( [ tft ] ) to obtain the force operator f_z(t ) = -_i=1^n [ ftilde ] corresponding to the oscillating disorder potential . that the cloud experiences the same force due to the disorder in these two situations is by no means obvious and is a consequence of the validity of the extended hpt .
this equivalence was exploited in our earlier work @xcite to determine the disorder - induced damping in the limit of a weak disorder potential .
since the state @xmath122 starts off at @xmath87 in the ground state @xmath133 ( either with or without the disorder potential ) , the effect of a weak disorder potential can be accounted for using conventional linear response theory . on the other hand ,
the initial state in the @xmath14 dynamics is a highly excited state of the harmonic trapping potential and conventional linear response theory can not be applied in this case .
the relationship in eq .
( [ transf ] ) shows that there is an intimate connection between the two very distinct physical situations depicted in fig .
[ oscillation ] ( left panel , ( a ) or right panel , ( b ) and left panel , ( b ) or right panel , ( c ) ) . in the first
, the system starts in a highly excited state in which the condensate is displaced from the minimum of the harmonic trap .
the energy of this state is given by e(0 ) = ( 0)|h operator in eqs .
( [ tft ] ) and ( [ tft2 ] ) , we have t^(,)h t ( , ) = h_0 + _
i=1^n v_ext(_i+ ) + + _
m_^2 x_r_+ e_cm , where the centre of mass energy is defined as e_cm = + _ m_^2 x_^2 .
thus the initial energy is e(0 ) = e_0 + e_cm + _
m_^2 x__0|r_|_0 , with @xmath137 .
we again note that @xmath133 denotes different states for the two scenarios depicted fig .
[ oscillation ] , but for both we have @xmath138 .
the initial state then evolves in the presence of the static external potential according to eq .
( [ psi_evol ] ) and during this evolution , the total energy of the system is conserved . in the alternative point of view ( fig .
[ oscillation ] : left panel , ( b ) or right panel , ( c ) ) , the condensate starts out near the minimum of the harmonic trap and is driven by a _
dynamic _ external potential which oscillates at the trap frequency . in view of eq .
( [ transf ] ) , the energy is given by @xmath139 here , the dynamic perturbation continually excites the condensate and the total energy increases as a function of time from the initial value @xmath140 .
provided the external potential provides a coupling between the centre of mass and internal degrees of freedom , one would expect on physical grounds that the expectation value of the total momentum @xmath141 in the state @xmath14 should tend to zero at long times . by the same token
, the expectation value of @xmath142 should tend to some constant limiting value . in this case
, the time average of @xmath143 tends to a finite limiting value of @xmath144 .
the important conclusion is that the energy does not increase indefinitely as a result of the dynamic perturbation and suggests that the system approaches a steady state .
this interpretation of the long time behaviour is supported by the time dependence of the centre of mass coordinate . in view of eq .
( [ transf ] ) , we have @xmath145 where @xmath146 is given by eq .
( [ x0 t ] ) . in fig .
[ oscillation ] ( left panel ( b ) or right panel ( c ) ) we have placed a marker ( filled dot ) tied to the external potential that initially coincides with the origin .
the position of this marker is @xmath147 and the position of the centre of mass relative to it is @xmath148 . with the assumption that @xmath149
, we thus conclude that the condensate in fig .
[ oscillation ] ( left panel ( b ) or right panel ( c ) ) moves synchronously with the the external potential at long times with the centre of mass located at @xmath150 relative to the marker . the final position @xmath151 of the centre of mass in fig .
[ oscillation ] ( left panel ( a ) or right panel ( b ) ) depends on the details of the external perturbation . for the example of a _ weak _ disorder potential
, one expects @xmath14 to approach a quasi - equilibrium state is a highly excited state which is a complex superposition of all eigenstates of @xmath112 .
however , physically we expect that this state is essentially equivalent to a thermal equilibrium state in the sense that the expectation values of various macroscopic physical quantities obtained using the dynamical state will be very similar to the averages calculated using a thermal equilibrium ensemble . ] which is centred on the minimum of the harmonic potential , i.e. , @xmath152 .
however , if the disorder is very strong , the initial state can be localized by the disorder potential @xcite and @xmath153 . in this case , the centre of mass is pinned to the position of the marker in fig .
[ oscillation ] ( left panel ( b ) or right panel ( c ) ) so that @xmath154 .
this motion gives a centre of mass energy @xmath155 .
the oscillatory term in this result corresponds to the @xmath156 contribution from the last term in eq .
( [ tildee ] ) .
its time average is of course zero .
we conclude this section with a few general comments .
the dissipation of the harmonically - confined condensate in the presence of an external perturbation is analogous to the dissipation experienced by a uniform superfluid moving past an impurity with a _ constant _ velocity .
galilean invariance allows one to consider the latter situation from the equivalent point of view of the impurity moving with a _ constant _ velocity through a stationary superfluid @xcite .
although a galilean transformation does not apply to the harmonically trapped gas , the situation depicted in fig .
[ oscillation ] ( left panel ( b ) or right panel ( c ) ) is analogous to the moving impurity in that the external ( disorder ) potential is moving relative to the gas which , at least initially , is stationary .
however , it would be incorrect to think of these situations as having arisen by means of a transformation to a frame of reference in which the centre of mass , say , is at rest .
such a frame would not oscillate freely at the trap frequency as the external potential is required to do according to the extended hpt .
although the extended hpt was motivated by other considerations , it is worth pointing out that eqs .
( [ psi_evol ] ) and ( [ tilde_psi_evol ] ) are in fact related by means of a coordinate transformation .
we interpret the state vector @xmath122 satisfying eq .
( [ tilde_psi_evol ] ) as the state in the ` laboratory ' frame of reference .
we now imagine making a transformation to the non - inertial frame of reference in which @xmath157 is stationary @xcite . in this frame of reference ,
the trapping potential is of course non - stationary .
nevertheless , when the harmonic trapping potential is transformed , one finds that the state vector in the non - inertial frame is the solution of eq .
( [ psi_evol ] ) _ provided _
@xmath96 is given by eq .
( [ x0 t ] ) .
it is only in this circumstance that the non - inertial forces generated by the transformation from the laboratory to the non - inertial frame are eliminated , with the evolution of the state vector in the non - inertial frame being governed simply by @xmath112 .
that eq .
( [ transf ] ) provides the relation between the state vectors in the two frames of reference is a direct consequence of the fact that the trapping potential is harmonic .
in the course of the derivation of the extended hpt , we encountered the state @xmath122 which satisfies the schrdinger equation ( [ tilde_psi_evol ] ) .
this state begins in the state @xmath64 at @xmath87 and is subjected to a dynamic perturbing potential .
one would usually expect a dynamic perturbation of this kind to lead to a continuous energy absorption , but as discussed earlier , for the special case in which the perturbation oscillates at the trap frequency , the time - averaged energy absorption rate eventually goes to zero
. this would _ not _ be the case if the perturbing potential oscillated at some arbitrary frequency . in this section
, we examine this more general situation . as a concrete example
, one might consider the effect of an oscillating optical lattice on a harmonically - confined system .
oscillations of the lattice in a certain range of frequency can lead to suppression of the tunnelling and the so - called dynamically induced phase transition @xcite . here , we show that a calculation of the energy absorption or centre of mass position provides a probe of the optical conductivity of the system when the amplitude of the oscillating perturbing potential is small .
this analysis extends the results of earlier work @xcite to the case of a _ harmonically - confined _ system subjected to an _ arbitrary _ oscillating external potential .
the physical problem of interest is described by the hamiltonian h(t)=h_0 + _ i=1^nv_ext(_i - _ 0(t ) ) , [ h_shaking ] where @xmath158 is an arbitrary , time - dependent displacement vector .
we refer to this kind of external potential as a ` shaking ' perturbation .
if @xmath158 is small on the length scale of variations of the external potential , the hamiltonian can be expanded as h(t)h + h(t ) , [ h_h ] where @xmath112 is given in eq .
( [ h_static ] ) and the perturbation is h(t)=
-_i=1^n v_ext(_i)_0(t ) = -dv_ext()_0(t ) n ( ) . [ h ]
although the perturbation is seen to couple to the density operator , the coupling has the very special form of the gradient of the external potential .
this allows the perturbation to be expressed in terms of the commutator relation h(t)= _ r_0(t ) [ j_,_i=1^n v_ext(_i ) ] , where @xmath159 is the total current operator . expressing @xmath160 in terms of @xmath112 and noting that @xmath161 commutes with the total kinetic energy and interatomic interactions , we find h(t)=
_ r_0(t ) [ j_,h]+nm__^2 r_0(t)r_. the second term on the right hand side comes from the commutator of @xmath161 and @xmath162 .
this form of the perturbation provides a direct route to the current response of the system . in the following
, it is convenient to write the perturbation as h(t ) _
ir_0(t ) a_i , where [ opa1 ] a_1= [ j_,h ] , and a_2= nm_^2r_. [ opa2 ] the total energy of the system in the presence of the perturbation is @xmath163 , where @xmath164 is the state of the system at time @xmath78 .
the tilde notation is used since the hamiltonian in eq .
( [ h_shaking ] ) is analogous to the hamiltonian in eq .
( [ tilde_psi_evol ] ) .
the rate of energy absorption is given quite generally by = ( t)| | ( t)if we assume that the perturbation is turned on at @xmath165 and that the system starts out in the ground state @xmath133 of @xmath112 , the energy absorption rate in the linear response regime is given by @xmath166 where we have defined the retarded response functions @xmath167|\phi_0\ra.\end{aligned}\ ] ] the heisenberg operators appearing in the response function are @xmath168 . by introducing the correlation function _
j(t - t ) _ 0|[a_i(t),a_j(t)]|_0 , [ phi_t ] the response function can be written as _
j(t - t)= 2i(t - t ) _ i , j(t - t ) .
we now consider the energy absorption rate for the special case of a monochromatic displacement , namely @xmath169 . taking the limit @xmath170 and averaging eq .
( [ de / dt ] ) over one period @xmath171 , we find @xmath172 where @xmath173 , the fourier transform of eq .
( [ phi_t ] ) , has the spectral representation ( here , @xmath174 ) _ i , j()=_n\{_0|a_i|_n_n|a_j|_0(-e_n 0)- _ 0|a_j|_n_n|a_i|_0(+e_n 0)}. [ phiab ] since @xmath175 , it is clear that eq . ( [ edrinter ] ) is manifestly real .
furthermore , time - reversal symmetry implies that @xmath176 and hence that @xmath173 is real . in this case ,
@xmath177 and @xmath178 is obtained from @xmath173 by a kramers - kronig relation @xcite . from eqs .
( [ opa1 ] ) and ( [ opa2 ] ) , we find _
m|a_1|_n = e_nm_m| j_|_n[amele ] and _ m|a_2|_n=_m|j_|_n .
[ bmele ] the latter equation follows from the identity = j_. in the above equations , @xmath179 .
we see that all the required matrix elements can be expressed in terms of those of the total current operator . using eqs .
( [ amele ] ) and ( [ bmele ] ) in eq .
( [ phiab ] ) , we obtain @xmath180 where _ ( ) = _
n\{_0|j_|_n_n|j_|_0(-e_n 0)- _ 0|j_|_n_n|j_|_0(+e_n 0)}. [ imchiab ] this quantity is the imaginary part of the fourier transform of the current - current response function _
( t - t)=(t - t)_0| [ j_(t),j_(t)]|_0 .
[ current_corr_1 ] it defines the real part of the optical conductivity according to @xcite _ ( ) = _ ( ) .
we emphasize that this is the optical conductivity of the system described by the hamiltonian @xmath112 in eq .
( [ h_static ] ) which includes both the harmonic and external potentials . inserting eqs .
( [ phi11])-([phi22 ] ) into eq .
( [ edrinter ] ) , we thus find @xmath181 this is a general result valid for any harmonically - confined system in the presence of an arbitrary oscillating external potential .
we thus see that the time - averaged energy absorption rate in the presence of harmonic confinement is proportional to the optical conductivity .
if the oscillation is restricted to the @xmath104-direction , i.e. @xmath182 , the energy absorption rate becomes = m^2 ^ 3z_0 ^ 2 ( 1-)^2 i m _ zz ( ) .
[ ear ] in the absence of harmonic confinement ( @xmath183 ) , eq .
( [ ear ] ) reduces to the result given in ref .
@xcite which was derived for the special case of an oscillating uniform optical lattice .
an alternative derivation of this limiting result is provided in appendix a ; this derivation points out a shortcoming of the original derivation in ref .
@xcite .
it is interesting to observe that the energy absorption rate in eq .
( [ ear ] ) vanishes if the external perturbation is oscillating at the trapping frequency @xmath184 .
this in fact is an exact result and is not a consequence of the perturbative analysis . if we take @xmath185 where @xmath146 is given in eq .
( [ x0 t ] ) , the hamiltonian in eq .
( [ h_shaking ] ) takes the form of the hamiltonian in eq .
( [ tilde_psi_evol ] ) . according to the extended hpt
, the time - averaged energy absorption rate goes to zero at long times ( see eq . ( [ tildee ] ) and the discussion thereafter ) .
this result applies for any amplitude of the oscillating potential and in particular , accounts for the second order result in eq .
( [ ear ] ) .
an alternative proof is given in appendix b where we consider the oscillation of the external potential to be turned on _
adiabatically_. if the oscillation occurs at the frequency @xmath184 , we show that the initial state of the system , @xmath133 , evolves into a state which moves together with the oscillating potential .
this dynamical state has a vanishing time - averaged energy absorption rate which is consistent with the @xmath184 result in eq .
( [ ear ] ) .
( [ edrf ] ) or ( [ ear ] ) shows that information about the current - current correlation function can in principle be accessed via the energy absorption rate .
however , the latter is not a quantity that is easily measured . on the other hand ,
a measurement of the centre of mass motion is relatively straightforward and provides an alternative means of probing the current - current correlation function .
for simplicity , we consider a perturbation shaking in the @xmath104-direction .
the relevant centre of mass coordinate is @xmath186 which , in linear response , is given by @xmath187.% \nn \\ % & = \frac{z_0}{nm\omega_z^2}{\sqrt{\left [ { \rm % re}\chi_{ba}(\omega ) % + { \rm re}\chi_{bb}(\omega)\right ] ^2 % + \left [ { \rm im}\chi_{ba}(\omega ) % + { \rm im}\chi_{bb}(\omega)\right ] ^2}}\sin\left ( \omega t % -\phi\right ) , \label{comtr}\end{aligned}\ ] ] we see that @xmath188 oscillates at the frequency @xmath189 with some phase lag relative to the oscillating external potential .
if we compare this with the expected experimental centre of mass trajectory @xmath190 , we see immediately that _ j im_2z , jz ( ) = nm_z^2 . [ imsum ] using eqs .
( [ phi21 ] ) and ( [ phi22 ] ) , we find _ zz ( ) = .
this result indicates that the current - current correlation function can be obtained experimentally through a measurement of the centre of mass oscillation amplitude @xmath191 and phase lag @xmath192 .
this kind of measurement is thus a direct probe of the optical conductivity .
in this paper we have studied various aspects of the dynamics of harmonically - confined atomic systems .
the results we have obtained are of a general nature and have a broad applicability to trapped atomic gases .
we have focused , in particular , on the effect that various external perturbations have on the dynamical evolution of the many - body wavefunction . in the case of a perturbation that couples solely to the centre of mass , we were able to obtain an explicit expression for the schrdinger evolution operator ; dobson s harmonic potential theorem follows naturally from this result . for more general external perturbations ,
the centre of mass and internal degrees of freedom are coupled and dissipation of the centre of mass motion sets in .
we have proved an extension of the hpt which demonstrates that this dissipative dynamics can be considered from two distinct points of view . in the first
, the evolution of an initial nonequilibrim state takes place in the presence of a _
static _ external potential . on the other hand
, one can equivalently think of the evolution as taking place in the presence of an external potential that itself moves according to the trajectory of a harmonically - confined particle . here
, the trapped atomic cloud starts off in an initial equilibrium state and is then continually excited by a _
dynamic _ perturbation .
this latter point of view has the advantage that the calculation of the damping of the centre of mass motion can be addressed by means of linear response theory when the perturbation is weak @xcite .
we next considered the response of a harmonically - confined system to a ` shaking ' potential . for a weak perturbation , this response
is directly related to current - current correlations and hence the optical conductivity .
our result for the energy absorption is a generalization of one obtained previously @xcite .
we have also shown that the optical conductivity can be probed by measuring the trajectory of the centre of mass itself
. this may in fact be the most feasible way of determining the optical conductivity experimentally .
this work was supported by a grant from the natural sciences and engineering research council of canada . in this appendix
we give a different derivation of the result in eq .
( [ ear ] ) for the special case of no harmonic confinement ( @xmath193 ) .
the hamiltonian in this case is h(t)=_i=1^n+_i < j v(_i-_j ) , [ h(t ) ] where we allow the displacement in the @xmath104-direction to have an arbitrary time dependence .
a hamiltonian of this kind was considered in ref .
@xcite for the case in which @xmath194 corresponds to a uniform optical lattice . in this context
, the displacement @xmath195 provides a phase modulation of the lattice potential .
we observe that the hamiltonian can be expressed as h(t)=u^(t ) h u ( t ) , [ phuhu ] where @xmath112 is the hamiltonian in eq .
( [ h(t ) ] ) with @xmath196 , and @xmath197 is the translation operator u ( t ) = \{i z_0(t)p_z/ } = \{imz_0(t)j_z/}. [ transoperator ] the dynamic state of the system evolves according to the schrdinger equation @xmath198 defining the state |(t)=u ( t ) |(t ) , we find that @xmath122 satisfies the equation @xmath199 where h ( t ) = h -mz_0(t ) j_z . [ ap5tildeh ] we see that the hamiltonian governing the evolution of the state @xmath122 contains a perturbation proportional to the total current operator . it should be emphasized that @xmath122 is _ not _ the state of the system as seen in the non - inertial frame of reference in which the external potential is stationary . to obtain the state in this frame of reference
one must apply a momentum boost in addition to the spatial displacement provided by @xmath197 @xcite .
the total energy of the system is given by @xmath200 using eq .
( [ schtildepsi ] ) and eq .
( [ aveenergy ] ) , we find that the energy absorption rate is given by @xmath201|\tilde \psi(t)\ra \nn \\ & = -\frac{m\dot z_0(t)}{i\hbar}\la \tilde \psi(t ) | [ \hat h , \hat j_z]|\tilde \psi(t)\ra.\end{aligned}\ ] ] introducing the interaction picture state vector @xmath202|\tilde \psi_{\rm i}(t)\ra , \label{er}\end{aligned}\ ] ] where @xmath203 .
the state @xmath204 evolves according to i =- m z_0(t ) j_z(t)|_i(t ) .
first order perturbation theory gives |_i(t)|_0- _
t_0^t dt z_0(t ) j_z(t ) |_0 , [ tpsi ] where we assume that @xmath205 for @xmath206 and that @xmath133 is the ground state of @xmath112 . substituting eq . ( [ tpsi ] ) into eq .
( [ er ] ) we have @xmath207| \phi_0\ra \nn \\ & + \left ( \frac{m}{i\hbar}\right ) ^2 \dot z_0(t)\int_{t_0}^t dt ' \dot z_0(t')\la
\phi_0|[[\hat h,\hat j_z(t ) ] , \hat j_z(t')]|\phi_0\ra .
\label{edrphasemn}\end{aligned}\ ] ] the first term on the right hand side of this equation vanishes since @xmath133 is the ground state of @xmath112 . using ij_z(t)=[j_z(t ) , h ] , eq .
( [ edrphasemn ] ) can be written as @xmath208|\phi_0\ra\nn\\ & = m^2\dot z_0(t)\frac{d}{dt } \int_{-\infty}^\infty dt ' \pi_{zz}(t - t')\dot z_0(t ' ) , \label{edrpmsys}\end{aligned}\ ] ] where @xmath209 is defined in eq .
( [ current_corr_1 ] ) .
this is a general result for any displacement @xmath195 that vanishes for @xmath206 . for the case of a sinusoidal perturbation , @xmath210
, we can take the limit @xmath170 and obtain = m^2z_0 ^ 2 ^ 3 t [ -t re_zz ( ) + t im_zz ( ) ] .
[ edrpmsin ] averaging this expression over one period , we obtain @xmath211 which is the result given in eq .
( [ ear ] ) for the case of @xmath212 .
we now point out that the above is _ not _ in fact the derivation given in ref .
@xcite . instead of the correct expression for the energy given in eq .
( [ aveenergy ] ) , the authors of ref .
@xcite take the energy of the system to be @xmath213 , where @xmath214 is the solution of eq .
( [ schtildepsi ] ) . the energy absorption rate in ref .
@xcite is then defined to be @xmath215 with @xmath216 given by eq .
( [ ap5tildeh ] ) , one has @xmath217 which differs from the correct result in eq .
( [ edrpmsys ] ) . for the sinusoidal displacement ,
we have = m^2z_0 ^ 2 ^ 3 t , [ eartokuno ] which clearly has a different time dependence from @xmath218 in eq .
( [ edrpmsin ] ) . however , when averaged over one period , the energy absorption rate is = m^2z_0 ^ 2 ^ 3 im_zz ( ) , which is the same as @xmath219 .
thus the approach adopted in ref .
@xcite does indeed yield the correct time - averaged energy absorption rate for a sinusoidal displacement .
however , this is not true for other forms of the displacement .
for example , for the displacement @xmath220 , which corresponds to the external potential moving with a constant velocity @xmath221 for @xmath130 , eq .
( [ edrtokuno ] ) gives @xmath222 since the initial state is one in which there is no current .
this conclusion also follows from eq .
( [ dtildee_dt ] ) . on the other hand , eq .
( [ edrpmsys ] ) gives @xmath223 which is a non - zero result .
this shows that eq .
( [ dtildee_dt ] ) can not be the correct result for the energy absorption rate in general . in this appendix
we provide an alternative explanation for why the energy absorption rate in eq .
( [ ear ] ) vanishes when @xmath224 . to this end
, we write the hamiltonian in eq .
( [ h_shaking ] ) as h_ad(t ) = h + e^t h(t , ) where h(t , ) = _ i=1^n .
the perturbation @xmath225 is turned on adiabatically via the parameter @xmath226 . for @xmath227
, @xmath228 reduces to @xmath112 , while for @xmath229 one recovers the hamiltonian @xmath230 in eq .
( [ h_shaking ] ) .
the parameter @xmath231 is introduced as an ordering parameter in the perturbation analysis of @xmath225 and is set to unity at the end of the calculation .
the displacement will be taken to have the specific form @xmath232 .
we observe that @xmath225 can be written as @xmath233 where @xmath234 .
although the required displacement of @xmath235 can be generated with any @xmath236 , we make the choice @xmath237 to ensure that the displacement operator evolves in time according to eq .
( [ emt ] ) .
this property will be shown to be crucial in the derivation of our final result .
we now construct the dynamical state in the interaction picture which reduces to @xmath133 , the ground state of @xmath112 , in the @xmath227 limit .
this state is given by @xmath238 here @xmath239 where @xmath240 and @xmath241 with a ( t)=(t)-(t ) .
[ aop ] we now evaluate @xmath242 explicitly . expanding this operator in powers of @xmath231
, we have _
i(t,)=_n=0^ ^(n)_i(t,0)^n , [ uexp ] where ^(n)_i(t,0)._i(t,)|_=0 .
similarly we have h_i(t,)= _
n=0^ h_i^(n)(t,0 ) ^n , [ hpexp ] where h_i^(n)(t,0 ) .
[ hin ] using eq .
( [ hpmlam ] ) in eq .
( [ hin ] ) , we obtain h_i^(0)(t,0)=0 , @xmath243 , \label{hpm1}\end{aligned}\ ] ] and for @xmath244 , h_i^(n)(t,0)=[a_i(t ) , h_i^(n-1)(t,0 ) ] . substituting eqs .
( [ uexp ] ) and ( [ hpexp ] ) into eq .
( [ evolop ] ) and comparing like powers of @xmath231 , we find ^(n ) _
i(t,0)=_m=0^n_-^t dt_1 e^t_1h_i^(m)(t_1,0 ) ^(n - m)_i(t_1,0 ) .
[ calun ] let us first consider @xmath245 . from eq .
( [ hpm1 ] ) we have @xmath246 e^{-i\hat h t/\hbar}\nn \\ & = -\frac{imz_0(t)}{\hbar}e^{i\hat h t/\hbar}\left [ \hat j_{z } , \hat h-\sum_{i=1}^n \frac{\hat \bp_i^2}{2m}-\sum_{i < j}^nv(\hat \br_i-\hat \br_j)-\sum_{i=1}^n v_{\rm tr}(\hat \br_i)\right]e^{-i\hat h t/\hbar}\nn \\ & = mz_0(t)\left \ { \frac{1}{i\hbar}[\hat j_{z,\rm i}(t ) , \hat h]+
n\omega_z^2\hat r_{z,\rm i}(t)\right \ } \nn \\ & = nmz_0(t)\left \ { \frac{\pa^2 } { \pa t^2}\hat r_{z,\rm i}(t)+ \omega_z^2\hat r_{z,\rm i}(t)\right \}. \label{hip1}\end{aligned}\ ] ] using eq . ( [ aop ] ) and @xmath247
, we find h_i^(1)(t,0)=-a_i(t ) .
[ hpm1f ] this result is _ only _ true when @xmath248 , as assumed . inserting eq .
( [ hpm1f ] ) into eq .
( [ calun ] ) for @xmath249 , we have @xmath250 where @xmath251 denotes terms that vanish in the @xmath252 limit . repeating this calculation for @xmath253
, we find @xmath254 + 2 \frac{\pa
\hat a_{\rm i}(t_1)}{\pa t } \hat a_{\rm i}(t_1)\right \ } \nn \\ & = -\frac{1}{\hbar^2}\int_{-\infty}^t dt_1 e^{\eta t_1}\frac{\pa \hat a ^2_{\rm i}(t_1 ) } { \pa t_1 } \nn \\ & = \left ( \frac{i}{\hbar}\hat a_{\rm i}(t)\right ) ^2 + { \cal o}(\eta)\end{aligned}\ ] ] the results for @xmath249 and @xmath253 suggest that ^(n)_i(0,t)= ( a_i(t ) ) ^n+ o ( ) [ caluna ] for all @xmath255 .
this in fact can be proven by induction . without presenting the details ,
we thus find that in the @xmath229 limit , @xmath256 the dynamic state of interest is thus given by @xmath257 we have thus proved that , after the perturbation is switched on adiabatically , the final state is the ground state of @xmath112 oscillating together with the external potential .
the energy of the system in the state @xmath14 is e(t,)=(t)|h_0+_i=1^nv_ext(_i - z_0(t))|(t ) .
this energy can be obtained from eq .
( [ tildee ] ) with the transcription @xmath258 , giving e(t,)=e_0 + ^2 e_cm+_0|p_z|_0z_0(t ) + m_z^2_0|r_z|_0z_0(t ) .
[ e_lambda ] the terms in eq .
( [ e_lambda ] ) linear in @xmath231 oscillate harmonically at the frequency @xmath259 and do not contribute to the time - averaged energy absorption rate .
we thus find = 0 .
this result is true to all orders in @xmath231 and in particular , demonstrates that the linear response energy absorption rate ( of order @xmath260 ) vanishes when @xmath261 . | in this paper we consider the dynamics of harmonically - confined atomic gases .
we present various general results which are independent of particle statistics , interatomic interactions and dimensionality . of particular interest
is the response of the system to external perturbations which can be either static or dynamic in nature .
we prove an extended harmonic potential theorem which is useful in determining the damping of the centre of mass motion when the system is prepared initially in a highly nonequilibrium state .
we also study the response of the gas to a dynamic external potential whose position is made to oscillate sinusoidally in a given direction .
we show in this case that either the energy absorption rate or the centre of mass dynamics can serve as a probe of the optical conductivity of the system . |
a great success of nuclear astrophysics was the demonstration that radioactivity powers the light curves of type ia supernovae ( sn ia , most likely thermonuclear ) and , at least for some events at late times , type ib / c and type ii ( most likely core collapse ) supernovae . shortly after its explosion a supernova enters a state of homologous expansion .
the temperature drops during the expansion until nuclear fusion reactions cease ; radioactive decays , however , still take place , since the matter is only slightly ionized .
it is now widely accepted that the energy liberated in the decay chain of radioactive @xmath1 is the most important nuclear source for re - heating the supernova ejecta to temperatures high enough for the spectrum to peak at optical wavelengths @xcite . at first
, the bulk of the heating is produced by the energetic @xmath0-rays which thermalize and deposit their energy via compton scattering and photoelectric absorption . in the homologous expansion ,
the column density ( and therefore also approximately the compton opacity ) decreases with time as @xmath7 , and the ejecta become more and more transparent to these high energy photons .
once @xmath0-rays escape , the positrons produced in the decay of @xmath2 and @xmath6 were thought to be the main heating sources . here , we draw attention to often overlooked additional leptonic heating channels : auger and internal conversion electrons . in section [ sec : decay ]
we review the physics of nuclear decays relevant for supernova light curves . in section [ sec : lc ] we demonstrate the impact of internal conversion electrons from the decay of @xmath4 on different supernova light curves .
we conclude in section [ sec : conc ] with an outlook how this effect could be used to constrain supernova explosion models in the future . for the published refereed journal article that first pointed out the significance of internal conversion electrons on supernova light curves
, please see @xcite .
the time dependence of @xmath8 nuclide abundances @xmath9 in a ecay chain is governed by the bateman equations : @xmath10 for @xmath11 and initial abundances @xmath12 and @xmath13 we get the solution @xmath14 + n_2(0)\exp(-\lambda_2 t ) \hspace{8pt}.\end{aligned}\ ] ] the decay constants @xmath15 are related to the half - lives @xmath16 and the mean life - time @xmath17 via @xmath18 the rate of energy deposition by decays of nucleus @xmath19 is given by the activity multiplied by the energy deposited per decay : @xmath20 where the number @xmath9 is given by eq .
[ eq : sol1 ] or eq . [ eq : sol2 ] and the energy deposited
, @xmath21 , is a function of time due to the increasing escape fraction of @xmath0-rays and possible late time escape of positrons
. 9 cm [ cols="^,^,^,^,^",options="header " , ] most of the nuclei synthesized in supernova explosions have atomic masses in the range @xmath22 . the isotopes are generally either stable or on the proton - rich side of the valley of stability . for these unstable nuclei ,
radioactive decay occurs along an isobar towards neutron richness , either by electron capture or positron emission .
electron capture proceeds via the capture of an atomic ( typically k or l shell ) electron by a proton in the nucleus and corresponding emission of an electron neutrino .
positron emission proceeds via the decay of a proton in the nucleus into a neutron and the corresponding emission of a positron and an electron neutrino . for both processes ,
the transition to the daughter nucleus is a statistical process to a distribution of ( excited ) nuclear states .
following electron capture , the daughter is formed with a hole in its atomic electron cloud .
higher lying atomic electrons transition to fill the gaps in the lower lying atomic shells , which results in characteristic x - rays being emitted from the cascade . for every such electron transition
, there is also a probability that instead of an x - ray one or more higher lying atomic electrons are ejected .
these electrons are known as auger electrons .
r2.5 in analogous to the x - ray cascade of the atom , the excited daughter nucleus typically undergoes a series of transitions towards the ground state accompanied by emission of characteristic @xmath0-ray photons .
if the nucleus is surrounded by an electron cloud , then there is again a probability that instead of a @xmath0-ray one or more atomic electrons are ejected .
this process , in which the energy difference of the nuclear levels is carried away by the ejection of an inner atomic electron and there is no @xmath0-ray photon emitted , is called internal conversion .
the probability for internal conversion to occur for a given level is measured by the internal conversion coefficient @xmath23 .
@xmath24 is normally small for nuclei with low atomic number , but increases for levels close to the ground state . for the decay of @xmath4
the probability for production of internal conversion electrons is very large , due to the existence of a low - lying nuclear level in the daughter nucleus @xmath25 ( see fig . [ fig1 ] ) .
for the first exited state ( 14.4 kev @xmath26 ) of @xmath25 , the internal conversion coefficient @xmath27 .
this decay is 100% electron capture , and over 99.8% all decays are into the 136 kev level .
captures to the higher lying 366.74 and 706.42 kev levels contribute only marginally . for ground - state
to ground - state electron capture transitions , such as in the decay of @xmath28 , no @xmath0-rays or positrons are emitted .
assuming that the neutrino escapes without interactions , in such transitions the auger electrons and the x - rays constitute the only sources of radioactive heating .
the following four decay chains contribute most to bolometric supernova light curves : @xmath29 here we do not model the radiative transport , we only compare leptonic ( and x - ray ) energy injection rates .
we do not include the energy produced by the pair annihilation and further assume that the kinetic energy of the leptons is completely and locally deposited and thermalized .
the energy generation rates presented here do not include any heating due to @xmath0-rays and therefore are not predictions for bolometric light curves .
this approach , however , does allow for a direct comparison of the relative importance of the different leptonic heating channels ( positrons from the decays of @xmath2 and @xmath6 and electrons from the decays of @xmath4 and @xmath28 . )
the relevant energies of the different decay channels are listed in table [ tab1 ] .
these data are taken from the national nuclear data center .
bolometric light curves have to be reconstructed from multi - band photometry . for reliable reconstruction ,
the contribution of the uv / optical ( @xmath30 ) and near - infrared ( @xmath31 ) bands have to be included ( uvoir light curve ) . however , near - ir observations are rare at very late epochs , and sometimes only @xmath32-through-@xmath33 band observations with a near - ir correction extrapolated from earlier epochs are used @xcite .
below , we discuss some aspects and examples of different supernova light curves .
sne ia are thought to be thermonuclear disruptions of white dwarf stars and as such do not have an extended envelope .
consequently , their ejecta become transparent to @xmath0-rays relatively soon and the positron - dominated phase generally starts @xmath34@xmath35 days after the explosion @xcite . between @xmath35 and @xmath36 days ,
the uvoir light curves fall exponentially with the @xmath2 decay half - life @xcite , a clear sign that a constant fraction of positrons is trapped .
unfortunately , only few sn ia were observed at even later times , but there are indications of a slow - down in the light curves of some optical bands after @xmath37@xmath36 days @xcite .
a particularly interesting case is sn 2003hv , which shows a slow - down in the _ bolometric _ light curve 786 days after b - band maximum light @xcite .
it has been suggested that additional ( possibly radioactive ) heating sources may be needed to explain this observed effect .
we point out that the magnitude and time of occurrence of the observed slow down of the light curve is expected and a natural consequence of the leptonic energy injection from the decay of @xmath4 . to illustrate this , consider the leptonic energy generation rates of important long lived isotopes between 500 and 2000 days for w7 @xcite , a common reference sn ia model ( see fig . [ fig2 ] ) . for this particular choice of yields @xcite ,
the light curve is significantly higher from assuming only heating from @xmath2 positrons after about 750 days .
1000 days after the explosion , the heating from @xmath4 electrons equals the heating from @xmath2 positrons .
the progenitors of core collapse sne are massive stars , which have an extended , massive envelope at the time of explosion , which provides opacity for the emerging @xmath0-rays . compared to sne ia ,
the nuclear burning also takes place at higher entropy , which means that core collapse supernovae synthesize more @xmath5 .
these differences imply that the moment when internal conversion electrons from @xmath4 significantly contribute to the light curves is delayed .
nevertheless , two observed examples of core collapse sn where internal conversion electrons do significantly alter the shape of the light curve are given below .
the well observed sn 1998bw , which was the first case of a @xmath0-ray burst associated with a sn@xcite , was classified as a sn ic .
the explosion was a very energetic , asymmetric disruption of a massive , stripped stellar core ( @xmath38 @xmath39 ; @xcite ) . the high ejecta velocities and the lack of an extended envelope reduce the @xmath0-ray opacity such that by @xmath40 days almost all @xmath0-rays are freely streaming . a nucleosynthetic yield calculation for hypernovae like 1998bw @xcite predicts a @xmath3 to @xmath1 mass ratio @xmath41 , @xmath42 times the solar value for the @xmath25 to @xmath43 ratio ( which is @xmath44 ; @xcite ) .
after modeling the uvoir light curve to @xmath37@xmath40 days after the explosion , @xcite showed that a simple model without contributions from freeze - out effects , circumstellar interaction , accretion onto a central compact object or light echoes requires a @xmath3 to @xmath1 ratio @xmath45 times greater than solar .
this large discrepancy between the value of @xmath46 predicted from explosive nucleosynthesis calculations and the one derived from light curve modeling reduces if the effects of internal conversion and auger electrons are included in the light - curve calculations . in fig .
[ fig3 ] we show the combined leptonic and x - ray luminosity corresponding to the nucleosynthetic yield calculations of @xcite .
the non-@xmath0-ray heating due to @xmath4 is the dominant contribution between @xmath47 and 1600 days .
the observed slow down of the light curve of 1998bw at @xmath48 days ( see fig . [ fig3 ] ) is thereby naturally obtained without the need for strongly super solar @xmath46 .
sn 1987a has demonstrated that light curves can be constructed for nearby supernovae extending for several years after the explosion .
this has led to the consideration of longer lived radioactive species .
the most important of these is @xmath4 , which is expected to be produced in significant amounts as the decay product of the short lived @xmath3 . due to its relatively longer half - life ( 271.79 days ) and the high opacity of the ejecta to the emitted @xmath0-rays , its @xmath0-rays alone may dominate the bolometric light curve , especially of core collapse supernovae , at late times @xcite .
in fact , due to its extended envelope , the opacity to @xmath0-rays in the remnant of 1987a remained high for such a long time that it never entered the @xmath2 positron dominated phase @xcite .
it is argued , that the only other radionuclide which noticeably contributes to the bolometric light curve is @xmath5 @xcite .
@xmath5 has an even longer half - life of 59.8 years , and the short lived daughter @xmath6 has a strong positron channel , which means that the @xmath5 decay chain is the dominant energy source at very late times .
photometric data of sn 1987a lead to the conclusion that the @xmath46 was rather large , about five times the corresponding solar value @xcite . due to the proximity of sn 1987a ,
the presence of @xmath2 and @xmath4 was inferred not only photometrically from their imprint on the light curve , but also through a direct detection of escaping @xmath0-rays from their decays . the observed gamma ray flux from @xmath4 with the oriented scintillation spectrometer experiment ( osse ) on board the _ compton gamma ray observatory _
@xcite favored at most moderate enhancement of @xmath3 ( factor 1.5 ) , in good agreement with nuclear reaction network calculations @xcite .
similarly , based on the light curve , a rather high @xmath5 abundance was claimed in several studies @xcite .
the great difficulty of explaining the light curve of 1987a between 1000 and 1500 days without strong overproduction of @xmath3 can be ameliorated by including `` freeze - out effects '' @xcite or by conveniently re - arranging the spatial distribution of the radionuclides @xcite , but some disagreement of data and models persists . by comparing the various leptonic energy deposition rates for a nucleosynthesis model of 1987a @xcite
, we show that the leptonic channels of @xmath4 decay are significantly contributing between 1000 and 2000 days ( see fig .
[ fig1987a ] ) .
fitting models to observed late time light curves provides a unique and independent method to directly measure the _ isotopic _ yields of prominent radioactive nuclei synthesized in the explosion ( in particular @xmath1 , @xmath3 , and @xmath5 ) .
we have shown that at late times , when the ejecta have become largely transparent to @xmath0-rays , the energy carried by auger and internal conversion electrons may constitute a significant source of heating .
these additional decay channels have to be considered for reliable isotopic abundance determinations from light curves .
in particular , we have shown that a re - analysis of the bolometric light curve of 1987a ( taking the hitherto unconsidered effect of internal conversion electrons into account ) would likely yield significantly different ( smaller ) @xmath5 and @xmath3 masses .
the new derived values for @xmath3 and @xmath5 would allow us to gain more insight into the explosion mechanism of core collapse supernovae .
in particular , observationally driven inferences about the location of the mass cut can be made , which would give us valuable information about the physical processes separating neutron star and black hole formation from massive stars .
last but not least , sne ia are considered as the source of the positrons required to explain the galactic 511 kev annihilation line observed by integral / spi @xcite . the question whether enough positrons can escape the remnant remains unanswered .
an understanding of the physics that underlies the light curves of sne is a crucial step in order to constrain the escape fraction of positrons .
k. iwamoto , f. brachwitz , k. nomoto , n. kishimoto , h. umeda , w. r. hix , and f .- k .
nucleosynthesis in chandrasekhar mass models for type ia supernovae and constraints on progenitor systems and burning - front propagation . , 125:439462 ,
december 1999 .
e. dwek , s. h. moseley , w. glaccum , j. r. graham , r. f. loewenstein , r. f. silverberg , and r. k. smith .
dust and gas contributions to the energy output of sn 1987a on day 1153 . , 389:l21l24 , april 1992 . | radioactive decays contribute significantly to the re - heating of supernova ejecta .
previous works mainly considered the energy deposited by @xmath0-rays and positrons produced by @xmath1 , @xmath2 , @xmath3 , @xmath4 , @xmath5 , and @xmath6 .
we point out that auger and internal conversion electrons constitute an additional heat source . at late times
, these electrons can contribute significantly to supernova light curves for reasonable nucleosynthetic yields .
in particular , the internal conversion electrons emitted in the decay of @xmath4 are an important heating channel for supernovae that have become largely transparent to @xmath0-rays .
we show that when the heating by these electrons is accounted for , the slowing down of the light curves of sn 1998bw and sn 2003hv is naturally obtained for typical nucleosynthetic yields .
additionally , we show that for sn 1987a the effects of internal conversion electrons are likely significant for the derivation of @xmath5 yields from its late time bolometric light curve . |
let @xmath0 be a bounded domain with smooth boundary @xmath1 and let @xmath2 be the unit outward normal vector to @xmath1 .
let us consider @xmath3 and @xmath4 we assume that @xmath5 and @xmath6 . throughout this paper ,
we set @xmath7 , @xmath8 for @xmath9 and @xmath10 , @xmath11 , @xmath12 , and @xmath13 denotes the scalar product of @xmath14 .
+ the main problems in this paper are + + let @xmath15 , @xmath16 , @xmath17 , @xmath18 , @xmath19 be given suitably .
determine @xmath20 , @xmath21 from @xmath22 .
+ moreover we consider @xmath23 and @xmath24 * inverse coefficient problem * + let @xmath25 and @xmath15 be suitably given . determine @xmath26 and/or @xmath27 by data @xmath28
. equations ( 1.1 ) and ( 1.3 ) are transport equations and are models in physical phenomena such as liouville equation and the mass conservation law . moreover the transport equation is related to the integral geometry ( e.g. , amirov [ 1 ] ) . as for other physical backgrounds such as neutron transport and medical tomography , see e.g. , case and zweifel [ 8 ] , ren , bal and hielscher [ 18 ] .
our inverse problem is formulated with a single measurement , and gaitan and ouzzane [ 9 ] , klibanov and pamyatnykh [ 15 ] , machida and yamamoto [ 17 ] discuss the uniqueness and the stability for inverse problems for initial / boundary value problems for transport equations by carleman estimates .
the papers [ 15 ] and [ 17 ] discuss transport equations with integral terms where solutions @xmath29 and @xmath30 depend also on the velocity as well as the location @xmath31 and the time @xmath32 .
the main methodology in [ 9 ] , [ 15 ] , [ 17 ] is based on bukhgeim and klibanov [ 7 ] .
after that , there have been many works .
limited to hyperbolic and parabolic equations , we can refer for example to baudouin , de buhan and ervedoza [ 3 ] , beilina and klibanov [ 4 ] , bellassoued and yamamoto [ 6 ] , imanuvilov and yamamoto [ 11 ] , [ 12 ] , klibanov [ 14 ] , yamamoto [ 21 ] and the references therein .
here we do not intend to give any complete lists of the references . in [ 9 ] and [ 15 ] , the key carleman estimate is the same as the carleman estimate for a second - order hyperbolic equation and in order to apply the carleman estimate one has to extend the solutions @xmath29 and @xmath30 to ( 1.1 ) and ( 1.3 ) to the time interval @xmath33 .
such an extension argument makes the proofs longer , and requires an extra condition to unknown coefficients and initial value as in [ 15 ] . in sections 2 and 4
, we prove carleman estimates ( lemmata 1 and 3 ) , which can directly estimate initial values .
thanks to our carleman estimates , we can simplify the proofs of the stability and relax the constraints of the principal coefficients @xmath15 s .
as for inverse problems for transport equations with many measurements , see surveys bal [ 2 ] , stefanov [ 20 ] and the references therein . klibanov and yamamoto
[ 16 ] established the exact controllability for the transport equations by a carleman estimate .
unlike [ 2 ] and [ 20 ] , we discuss the inverse problems for a single initial / boundary value problem where we need not change initial values or boundary values . more precisely , in the formulation for the inverse problems in [ 2 ] and [
20 ] , we have to change boundary inputs on some subboundary and repeat measurements of the corresponding boundary outputs on other subboundary .
one can apply the method of characteristics to the same kind of inverse problem for the first - order equation and see belinskij [ 5 ] , chapter 5 of romanov [ 19 ] for example .
we set @xmath34 and @xmath35 throughout this paper , we assume that @xmath36 and @xmath37 satisfy @xmath38 we here note by ( 1.5 ) that @xmath39 for @xmath40 .
we give four cases where ( 1.5 ) holds .
+ * case 1*. we assume @xmath41 with some @xmath42 . then (
1.5 ) holds if we choose @xmath43 , @xmath44 . +
* let us assume that @xmath45 is separated from @xmath46 by a hyperplane @xmath47 with some @xmath48 and @xmath49
. then @xmath50 or @xmath51 satisfies ( 1.5 ) . in particular , ( 1.5 ) holds if @xmath27 is a constant vector because @xmath52 is composed of one point . in fact , the separation condition means that @xmath53 for all @xmath54 or @xmath55 for all @xmath56 . +
* let @xmath57 .
we assume that there exists a constant @xmath58 such that @xmath59 then @xmath60 satisfies ( 1.5 ) if @xmath61 is sufficiently small . + * proof .
* by the cauchy - schwarz inequality , we have @xmath62 therefore ( 1.5 ) holds true if @xmath63 * case 4*. we assume that there exists @xmath64 such that @xmath65 for all @xmath66
. then we choose sufficienly small @xmath67 such that @xmath68 . setting @xmath69
, we verify that ( 1.5 ) holds .
in fact , @xmath70 now we state the first main result concerning the stability for the inverse source problem . + * theorem 1 * + let @xmath71 satisfy ( 1.1 ) and ( 1.2 ) , and let ( 1.5 ) be satisfied with some constant @xmath72 .
we assume that @xmath73 and @xmath74 let @xmath75 + ( i ) we assume @xmath76 with fixed constant @xmath77 .
then there exist constants @xmath78 and @xmath79 , which are dependent on @xmath80 , such that @xmath81 for all @xmath82 .
+ ( ii ) without the assumption in ( i ) concerning @xmath83 , there exists a constant @xmath79 , which is dependent on @xmath84 , such that @xmath85 for all @xmath86 .
+ ( iii ) in addition to ( 1.1 ) and ( 1.2 ) , we assume @xmath87 then there exists a constant @xmath79 , which is dependent on @xmath88 , @xmath89 , @xmath90 , such that @xmath91 for all @xmath86 .
+ the conclusion of ( i ) is a stability estimate of hlder type and holds under a priori boundedness @xmath92 , which is called conditional stability . on the other hand ,
the conclusions of ( ii ) and ( iii ) are lipschitz stability and in particular , with ( 1.8 ) we can have both - sided estimate ( 1.9 ) for our inverse problem .
we apply theorem 1 to the inverse coefficient problem of determining @xmath26 .
+ + for @xmath93 , let @xmath94 and let @xmath95 @xmath96 and @xmath97 with suitably given @xmath25 and @xmath98 .
we assume that there exists @xmath99 satisfying ( 1.5 ) for @xmath15 , ( 1.7 ) holds and @xmath100 moreover we assume @xmath101 and @xmath102 where @xmath77 is arbitrarily fixed constant
. then there exists a constant @xmath79 depending on @xmath103 such that @xmath104 + similarly to theorem 2 , we can discuss the determination of @xmath15 , but we have to determine @xmath105 functions as the components of @xmath15 , and so repeats of measurements of boundary data after changing initial values suitably are necessary but arguments can be repeated similarly to the proof of theorem 2 . here we omit detailed discussions for the determination of all the components of @xmath15 , but we consider the determination of the potential in the case of potential flows .
that is , we consider @xmath106 the first - order partial differential equation in ( 1.16 ) describes the mass conservation under stationary potential flow @xmath107 .
then we discuss the inverse problem of determining the stationary potential @xmath108 . for the statement of the main result , we define an admissible set of unknown potentials @xmath108 s .
let constants @xmath77 , @xmath109 and functions @xmath110 , @xmath111 be arbitrarily chosen .
we define the admissible set of @xmath108 s by @xmath112 @xmath113 for the function @xmath114 given in ( 1.17 ) , we set @xmath115 we state our final main result . + * theorem 3 * + for @xmath93 , let @xmath116 and let @xmath117 and @xmath118 with suitable @xmath25 and @xmath98 .
we assume ( 1.14 ) , @xmath119 and @xmath120 then there exists a constant @xmath79 depending on @xmath121 such that @xmath122 for all @xmath123 .
+ the proofs of theorems 1 - 3 are based on an argument by the carleman estimates , which was originated by bukhgeim and klibanov [ 7 ] .
here we used a modified argument by imanuvilov and yamamoto [ 11 ] , [ 12 ] which discussed for inverse problems for second - order hyperbolic equations .
the paper is composed of four sections and an appendix . in section 2 ,
we prove a relevant carleman estimate for the proofs of theorems 1 and 2 , and an energy estimate , and in section 3 , the proofs of theorems 1 and 2 are completed . in section 4 , we prove another carleman estimate for the proof of theorem 3 whose weight function gives a carleman estimate also for the laplacian and complete the proof of theorem 3 . in appendix
, we prove the carleman estimate for the laplacian .
we recall that @xmath34 and we set @xmath124 @xmath125 and @xmath126 with @xmath127 and @xmath128 , and @xmath129 first we prove + * lemma 1 * + ( i ) we have @xmath130 for all @xmath131 and @xmath132 satisfying @xmath133 in @xmath134 .
+ ( ii ) we assume ( 1.5 ) and @xmath135 then @xmath136 @xmath137 for all @xmath138 and @xmath132 satisfying @xmath133 in @xmath134 , where @xmath139 + inequality ( 2.4 ) is an estimate of carleman s type , which holds uniformly for sufficiently large @xmath140 .
we emphasize that carleman estimate ( 2.4 ) can estimate also the initial value @xmath141 , and the weight function is linear in @xmath32 . in [ 17
] such a linear weight function is used for proving a stability estimate for an inverse problem for a transport equation . for the transport equation , the works [ 9 ] and [ 15 ] used the same weight function as the second - order hyperbolic equation , that is , @xmath142 .
our choice ( 2.1 ) of the weight function enables us not to need to take any extensions of @xmath30 to @xmath143 in discussing the inverse problem . in [ 15 ] , the extension argument requires an extra assumption on the initial value and unknown coefficients in addition to ( 1.14 ) . + +
first we assume that @xmath144 .
we set @xmath145 and @xmath146 .
then @xmath147 hence by @xmath133 , we have @xmath148 substituting @xmath149 , we have @xmath150 @xmath151 next let @xmath152 , @xmath153 .
then @xmath154 therefore ( 2.5 ) completes the proof of lemma 1 ( i ) . + * proof of lemma 1 ( ii ) . * + by ( 2.2 ) and ( 2.3 ) , we have @xmath155 on @xmath156 .
therefore we absorb the second term on the right - hand side of the conclusion of lemma 1 into the left - hand side .
more precisely , we have @xmath157 and so we easily see that if @xmath158 , then @xmath159 thus , by noting @xmath160 the proof of the part ( ii ) is completed .
+ next we show the classical energy estimate . +
* lemma 2 . *
+ let @xmath161 and @xmath162 with arbitrarily fixed constant @xmath77 .
let @xmath163 satisfy @xmath164 then there exists a constant @xmath79 , depending on @xmath165 , such that @xmath166 @xmath167 + * proof . * + multiplying @xmath168 by @xmath169 and integrating over @xmath134 , we have @xmath170 setting @xmath171 and integrating by parts , we obtain @xmath172 therefore , noting that @xmath173 and integrating over @xmath174 , we have @xmath175 @xmath176 therefore @xmath177 @xmath178 for @xmath179 . since @xmath180 , @xmath181 , omitting the second term on the left - hand side of ( 2.8 ) , we apply the gronwall inequality , and we obtain @xmath182 @xmath183 substituting ( 2.9 ) into the third term on the right - hand side of ( 2.8 ) , we complete the proof of the lemma .
+ * proof of theorem 1 . * +
henceforth @xmath79 denotes generic constants which are independent of @xmath140 .
we note that @xmath184 and @xmath185 , and we set @xmath186 and @xmath187 . by ( 1.7 ) , we can choose @xmath128 such that @xmath188 in fact , it suffices to choose @xmath128 such that @xmath189 is sufficiently close to @xmath190 . with this @xmath191 , we set @xmath192 then ( 3.1 ) implies @xmath193 therefore by @xmath194 , there exist @xmath195 and @xmath196 such that @xmath197 , @xmath198 for applying lemma 1 , we need a cut - off function @xmath199 such that @xmath200 and @xmath201 we set @xmath202 then @xmath203 , @xmath44 and @xmath204 and @xmath205 applying lemma 1 ( ii ) to @xmath206 , we obtain @xmath207 for all large @xmath208 . here @xmath209 since @xmath210 for @xmath211 or @xmath212 , by ( 3.3 ) and the a priori bound @xmath213 , we have @xmath214 and @xmath215 for all large @xmath140
. moreover @xmath216 for @xmath66 and @xmath217 , @xmath66 , we have @xmath218 therefore ( 3.5 ) yields @xmath219 since @xmath220 for @xmath221 , we have @xmath222 that is , @xmath223 for all large @xmath140 . using @xmath224 by ( 3.3 ) and choosing @xmath140 large , we obtain @xmath225 that is , @xmath226 for all large @xmath227 , where @xmath228 is a sufficiently large constant . here
we set @xmath229 .
we separately consider the two cases : @xmath230 and @xmath231 . + * case 1 @xmath232 : * + estimate ( 3.8 ) implies @xmath233 + * case 2 @xmath231 : * + replacing @xmath234 by @xmath235 , we see that ( 3.8 ) holds for all @xmath140 .
we make the right - hand side of ( 3.8 ) small in @xmath236 .
we choose @xmath237 , that is , @xmath238 therefore ( 3.8 ) is reduced to @xmath239 where @xmath240 .
thus the proof of theorem 1 ( i ) is completed . next we prove theorem 1 ( ii ) .
we have @xmath241 applying lemma 2 to @xmath242 , we obtain @xmath243 @xmath244 for @xmath179 . therefore , omitting the second term on the left - hand side of ( 3.10 ) and applying it to ( 3.6 ) , we obtain @xmath245 and similarly to ( 3.8 ) , from ( 3.5 ) we can obtain @xmath246 choosing @xmath140 large , we can absorb the first term on the right - hand side into the left - hand side , and complete the proof of ( ii ) . finally we prove ( iii ) . by ( 1.8 )
, the conclusion of ( ii ) immediately yields the second inequality in ( 1.9 ) .
next , by ( 1.8 ) and ( 3.10 ) , we have @xmath247 which proves the first inequality in ( 1.9 ) .
thus the proof of theorem 1 ( iii ) is completed .
+ * proof of theorem 2*. + theorem 2 can be derived directly by theorem 1 .
in fact , setting @xmath248 , @xmath249 and @xmath250 , by ( 1.10 ) - ( 1.14 ) we have @xmath251 and @xmath252 , @xmath253 , @xmath254 for @xmath66 .
thus theorem 1 ( iii ) yields the conclusion of theorem 2 , and the proof of theorem 2 is completed .
+ we set @xmath255 for the proof , we further need a carleman estimate for @xmath256 with the same weight function for @xmath257 . unfortunately the weight function defined by ( 2.1 ) does not work as weight for a carleman estimate for @xmath256 .
thus we have to introduce a second large parameter in the weight function .
that is , as the weight function , we set @xmath258 where @xmath259 is chosen later .
the weight function in the form ( 4.1 ) has been known as more flexible weight function producing a carleman estimate ( e.g. , hrmander [ 10 ] , isakov [ 13 ] ) . in the existing works on inverse problems by carleman estimates , given weight functions
have been used , and to the best knowledge of the authors , theorem 3 is the first case where a carleman estimate with unknown coefficient as the weight , is seriously involved .
the use of such a carleman estimate is necessary in order that the admissible set @xmath260 of unknown coefficients @xmath108 s is generously formulated such as ( 1.17 ) .
otherwise the admissible set of unknown coefficients should be more restrictive .
for example , for inverse problems of determining principal parts for second - order hyperbolic equations , there are no works by carleman estimate with weight function given by unknown coefficient , so that choices of admissible sets of unknown coefficients of the principal parts are very limited ( e.g. , bellassoued and yamamoto [ 6 ] , section 6 of chapter 5 ) . for it , we need a carleman estimate where all the constants can be taken uniformly for arbitrary @xmath261 .
we first show a carleman estimate for @xmath257 with weight ( 4.1 ) , which should hold uniformly for all @xmath261 .
we fix @xmath262 such that @xmath263 where the constant @xmath264 characterizes @xmath260 ( see ( 1.17 ) ) .
henceforth we set @xmath265 + * lemma 3 * + for each @xmath259 there exists a constant @xmath266 , which is dependent on @xmath267 , such that we can choose a constant @xmath79 satisfying @xmath268 @xmath269 for all @xmath138 , all @xmath270 and all @xmath132 satisfying @xmath271 , @xmath44 .
+ here we note that the choices of @xmath234 and @xmath272 in ( 4.3 ) are uniformly for @xmath261 . + * proof*. + the proof is similar to lemma 1 by noting the independency of the constant of @xmath261 .
let @xmath261 be chosen arbitrarily .
first we consider the case of @xmath273 .
we set @xmath274 , and @xmath275 we note @xmath276 , @xmath277 and @xmath278 by ( 1.17 ) . then @xmath279 henceforth @xmath79 denotes generic constants which are independent of @xmath261 and @xmath140 , but dependent on @xmath280 . next integration by parts yields @xmath281^t_0 dx + s\int_q ( { \partial}_tj_d)\vert w\vert^2 dxdt \\ \ge & s\int_{\ooo } \la\va_{d}(x,0)(-\beta + \vert\nabla d\vert^2 ) \vert w(x,0)\vert^2 dx - c\int_q s\la^2\va_{d}\vert w\vert^2 dxdt\\ \ge & s(\delta_0 ^ 2-\beta)\int_{\ooo } s\la\va_{d}(x,0)\vert w(x,0)\vert^2 dx - c\int_q s\la^2\va_{d}\vert w\vert^2 dxdt\end{aligned}\ ] ] and @xmath282 hence @xmath283 @xmath284 we choose @xmath285 then , since @xmath286 for @xmath221 and @xmath261 , if @xmath138 , then @xmath287 therefore ( 4.5 ) yields @xmath288 @xmath289 for all @xmath138 .
henceforth @xmath290 denotes generic constants which are dependent on @xmath267 .
since @xmath291 , we rewrite ( 4.6 ) in view of @xmath30 , so that @xmath292 @xmath293 for all @xmath138 . here
we note that @xmath79 and @xmath290 may vary line by line , but dependent only on @xmath267 .
next @xmath294 in @xmath295 for each @xmath261 .
similarly to the proof of lemma 1 ( i ) , we can finish the proof of ( 4.3 ) .
+ next we show a carleman estimate for @xmath256 with the weight function @xmath296 . + * lemma 4 * + there exists a constant @xmath297 such that for any @xmath298 , there exists a constant @xmath299 satisfying : we can choose @xmath79 such that @xmath300 for all @xmath301 , @xmath302 .
+ lemma 4 is a carleman estimate for @xmath256 and the carleman estimate for @xmath256 is well known ( e.g. , hrmander [ 10 ] , isakov [ 13 ] ) .
however we need the uniformity of the constants @xmath303 with respect to @xmath261 in the carleman estimate .
thus in appendix we give a direct proof by integration by parts , which differs from [ 10 ] , [ 13 ] .
+ now we proceed to + * proof of theorem 3*. + * first step .
* we set @xmath304 by assumption ( 1.21 ) : @xmath305 of the theorem , for arbitrary @xmath270 , we can choose @xmath128 and fix such that @xmath306 we choose @xmath307 given in lemma 4 and fix . next we set @xmath308 , where @xmath309 , @xmath310 are proved to exist in lemmata 3 and 4 .
since @xmath311 on @xmath312 for any @xmath313 , we note that the subboundaries @xmath314 and @xmath315 defined by ( 1.18 ) correspond to @xmath316 and @xmath317 respectively .
since @xmath318 is fixed , henceforth by @xmath234 denoting generic constants which are independent of @xmath319 and @xmath270 but dependent on @xmath320 , @xmath25 , @xmath321 , @xmath322 , @xmath323 , @xmath324 , noting ( 4.7 ) and ( 4.4 ) , by lemmata 3 and 4 we obtain @xmath325 @xmath326 and @xmath327 for all @xmath319 , @xmath270 , @xmath71 satisfying @xmath328 in @xmath134 and @xmath329 .
thus we obtained two carleman estimates which hold uniformly for arbitrary @xmath261 . + * second step . *
we set @xmath330 , @xmath331 and @xmath332 in @xmath295 .
then , by ( 1.19 ) and ( 1.20 ) , we obtain @xmath333 we differentiate this partial differential equation with respect to @xmath29 in @xmath32 , noting that @xmath252 , @xmath334 and @xmath335
. then we have @xmath336 applying lemma 2 to ( 4.10 ) and estimating @xmath242 , we readily verify the first inequality in ( 1.23 ) .
thus the rest part of this section is devoted to the proof of the second inequality of ( 1.23 ) . since @xmath337
does not hold , for applying ( 4.8 ) , we need a cut - off function . noting @xmath338 and @xmath339 for all @xmath66 .
henceforth we set @xmath340 by ( 4.7 ) we have @xmath341 .
we note that @xmath342 is independent of @xmath140 and special choices of @xmath343 , but dependent on @xmath344 . therefore , by @xmath261 , we have @xmath349 , and @xmath350 for any @xmath351 , we have @xmath352 @xmath353 at the last inequality , by @xmath351 and the mean value theorem , we have @xmath354 and @xmath355 where @xmath356 is some constant .
let @xmath363 satisfy @xmath200 and ( 3.4 ) with here chosen @xmath364 .
we set @xmath365 then , by ( 4.10 ) , we have @xmath366 applying ( 4.8 ) to ( 4.14 ) , we obtain @xmath367 @xmath368 for all @xmath319 . here and
henceforth we set @xmath369 apply lemma 2 to ( 4.10 ) , by ( 1.21 ) we have @xmath370 @xmath371 consequently , since @xmath372 only if @xmath373 by ( 3.4 ) , it follows from ( 1.14 ) that ( 4.15 ) and ( 4.16 ) yield @xmath374 for all @xmath319 .
for the first integral on the right - hand side , we used @xmath375 for @xmath221 , and so @xmath376 therefore we obtain @xmath377 for all @xmath319 .
the first term on the right - hand side can be absorbed into the left - hand side by choosing @xmath140 large , and by ( 4.13 ) we have @xmath378 @xmath379 for all @xmath319 . since @xmath380 by @xmath381
, we can apply ( 4.9 ) to obtain @xmath382 substituting this into the left - hand side of ( 4.17 ) , we obtain @xmath383 for all @xmath319 , and absorbing the first term on the right - hand side into the left - hand side again by choosing @xmath140 large if necessary , we have @xmath384 for all @xmath319 . noting @xmath385
, we see that @xmath386 for all large @xmath387 . since @xmath341
, for large @xmath140 , we can absorb the first term on the right - hand side into the left -hand side , so that @xmath388 thus the proof of theorem 3 is completed .
+ + * appendix .
proof of lemma 4 * + in order to clarify the dependence in the carleman estimate on @xmath264 and @xmath389 , for arbitrary @xmath261 , we prove the lemma directly and the proof is similar for example to section 3 in yamamoto [ 21 ] , where a carleman estimate is proved for a parabolic equation . by the usual density argument , it suffices to prove for @xmath390 .
we set @xmath391 and @xmath392 we have @xmath393 we set @xmath394 we note that @xmath395 here and henceforth by @xmath396 , etc . , we denote generic constants which are independent of @xmath397 but dependent on @xmath398 , and these constants may change line by line .
moreover we always assume that @xmath399 .
hence we use the following inequality : @xmath400 if @xmath401 . in view of @xmath402
, we have @xmath403 taking into consideration the orders of @xmath404 , we split @xmath405 as follows : @xmath406 since @xmath407 , we obtain @xmath408 now we estimate @xmath409 @xmath410 by the integration by parts and @xmath411 on @xmath312 , we have @xmath412 since @xmath413 we have @xmath414 therefore @xmath415 moreover we have @xmath416 @xmath417 and @xmath418 @xmath419 therefore ( 6 ) - ( 9 ) yield @xmath420 @xmath421 for the proof of the lemma , we have to estimate @xmath422 , but the first and the second terms on the right - hand side have different signs .
thus we need another estimate . that is , multiplying ( 3 ) with @xmath423 and integrating over @xmath134
, we obtain @xmath424 @xmath425 hence , by integration by parts , we obtain @xmath426 here we have @xmath427 consequently @xmath428 next @xmath429 @xmath430 by ( 2 ) , we see @xmath431 since @xmath432 it follows from ( 11 ) - ( 14 ) that @xmath433 therefore @xmath434 @xmath435 thus we consider ( 10 ) @xmath436 @xmath437 ( 15 ) : @xmath438 @xmath439 since @xmath440 for all @xmath66 and @xmath270 , we have @xmath441 @xmath442 and @xmath443 we choose sufficiently large @xmath297 such that @xmath444 in ( 18 ) . for any given @xmath298 , we choose @xmath445 such that @xmath446 in ( 17 ) and ( 19 ) .
then @xmath447 for all @xmath298 and @xmath301 , and noting @xmath448 for all @xmath261 , we can absorb the second and the third terms on the right - hand side of ( 16 ) into the left - hand side to obtain @xmath449 we re - write by means of @xmath450 . replacing @xmath451
, we have @xmath452 and @xmath453 substituting them into ( 20 ) , we complete the proof of lemma 4 . | in this article , we consider inverse problems of determining a source term and a coefficient of a first - order partial differential equation and prove conditional stability estimates with minimum boundary observation data and relaxed condition on the principal part .
inverse problem , transport equation , stability estimates . 35r30 , 35b35 , 82c70 . |
since liners ( low - ionization nuclear emission - line regions ; heckman 1980 ) were first identified as a class the nature of the source of ionization has been debated , with the two main possibilities being a starburst or agn nuclear source .
if an agn is responsible for the ionization , the next question is how similar are these agn to seyferts and qsos .
an interesting development has been the discovery of broad @xmath9 emission in many liners @xcite .
these liners are almost certainly agn powered , particularly when they are observed in early - type galaxies where outflows from massive stars and/or supernovae are not expected to be relevant .
early - type galaxies also tend to harbor very massive central black holes , making broad - line liners observed in early - type galaxies ideal laboratories for studying ( potentially ) low accretion rate physics . of course , this is assuming that the x - ray emission is isotropic and scales with accretion rate , however alternatively the accretion energy may be dissipated anisotropic and/or non - radiatively ( i.e. , in outflows ; di matteo et al .
2000 ) .
ngc 3998 is an so galaxy at a distance of 14.1 mpc @xcite .
its h@xmath0 line contains a broad component with 37% of the h@xmath0 + [ n ii ] flux and a fwhm of @xmath10 @xcite .
asca _ and _ bepposax _ observations find that the 2 - 10 kev luminosity of ngc 3998 is @xmath11 erg s@xmath4@xcite , and the 0.1 - 100 kev luminosity is @xmath12 erg s@xmath4 .
the mass of a putative black hole in the nucleus of ngc 3998 is estimated to be @xmath13 , giving @xmath14 , or well into the radiatively - inefficient accretion flow ( riaf ) regime .
the main goal of this paper is to contrast ngc 3998 with `` normal '' seyfert 1 galaxies and to test the applicability of current riaf models to the observed properties of ngc 3998 .
we present the results of a recent _ xmm - newton _ observation of ngc 3998 , analyzed in conjunction with archival x - ray data .
previous x - ray observations showed that no fe - k line is present in the x - ray spectrum from ngc 3998 , and here we tighten this constraint , which places limits on the accretion geometry in ngc 3998 .
additionally we will use _ xmm - newton _ optical monitor data along with other multi - wavelength data to compare the spectral energy distribution ( sed ) of ngc 3998 with those predicted by various accretion models .
the parameters of x - ray observations are listed in table [ tab1 ] .
the _ asca _
data was reduced as described in @xcite and @xcite .
the _ bepposax _
data were reduced by the xselect and ftools in heasoft 5.2 from the cleaned event files obtained from the _ bepposax _ science data center .
lecs and mecs spectra of ngc 3998 were extracted from a circular region with a 4 arcmin radius .
background spectra were taken from an annular region around ngc 3998 . here
we present ccd spectra from the three epic detectors ( two mos and one pn ) and the optical monitor u - band and uv ( uvw2 filter ) images .
the _ xmm - newton _ data was reduced using xmmsas version 5.4.1 with the use of xassist , a software package that performs the initial steps of x - ray data analysis @xcite .
the xmm pn and mos spectra were extracted from 37 circular regions ( derived from fitting a gaussian surface brightness model to the x - ray images ) .
we compared the radial profile of the mos1 image ( chosen since the target was not close to a ccd boundary in that case ) to a scisim simulation using the spectral model ( see [ specsec ] ) and found that the x - ray emission from ngc 3998 is dominated by an unresolved source .
the only other significant source in the vicinity of ngc 3998 is ngc 3998 x-1 @xcite which is discussed in [ ulxsec ] .
the om data was reprocessed with xmmsas and the images and source lists produced by the tool `` omichain '' were used in this paper .
the om spatial resolution depends on photon energy and the fwhm of the point - spread function ( psf ) is @xmath15 and @xmath16 in the u and uvw2 bands ( _ xmm - newton _
user s guide ) .
all spectra were binned to 20 counts / bin to allow the use of the @xmath17 statistic .
llll & 0.4 - 10.0 & 05/10/94 & 40 + _ bepposax _ & 0.1 - 100.0 & 06/29/99 & 24 ( lecs ) , 77 ( mecs ) , 38 ( pds ) + _ xmm - newton _ & 0.2 - 10.0 & 05/09/01 & 12 ( mos ) , 9 ( pn )
we fit the _ xmm - newton _ ( ccd ) spectra of ngc 3998 with a simple absorbed power - law model , allowing the overall normalization of the model fit to each detector to be independent .
this resulted in a good fit with @xmath18 , consistent with the previous _ asca _ and _ bepposax _ observations as shown in table [ specfittab ] .
note that while the three observations resulted in very close agreement for the power - law slope , the column density measurements varied from @xmath19 ( or a factor of @xmath20 greater than the galactic value ; murphy et al .
1996 ) to @xmath21 .
given the calibration uncertainties below 0.5 - 1.0 kev for these detectors it is not clear to what extent these variations are significant .
we show the power - law fit to the _ xmm - newton _ data in figure [ xmmplfig ] ( the _ bepposax _ and _ asca _ data are plotted in pellegrini et al .
2000 and terashima et al .
the _ asca _ data also resulted in a flux @xmath22 lower than the _ bepposax _ and _ xmm - newton _ measurements ( see [ varsec ] ) . note that the pds flux was @xmath23 higher the mean of the other flux measurements , however the bandpass of the pds data was @xmath24 kev so the 2 - 10 kev flux is based on an extrapolation .
we also tried restricting the _ bepposax _ data to only include data below 10 kev ( i.e. , excluding the pds data ) and found no significant impact on the fit parameters . power - law fit parameters from a fit to the _ xmm - newton _ and _ bepposax _ data simultaneously , with the normalization of each data set allowed to vary , is shown in table [ specfittab ] , and @xmath17 only increased by 3 when the normalization of the pds component was fixed result in the same 2 - 10 kev flux as the mecs flux ( i.e. , the pds data systematically deviate from the power - law fit at a confidence of only @xmath25 ) .
there are no systematic residuals suggesting additional spectral features might be present .
we do not find evidence for any edges suggesting that an ionized absorber is not present ( note that the edges discussed in pelligrini et al .
were detected at less than the 3@xmath26 level ) .
the rgs data do not have high enough signal - to - noise to constrain spectral features .
there is a slight excess of counts in the 0.65 - 0.7 kev region but this feature is not statistically significant .
we also tried a plasma plus power - law fit @xcite to the _ xmm - newton _ spectra and found a significant improvement in @xmath17 ( @xmath27 = 1406/1334 compared with 1423/1338 for the power - law fit ) .
an absorption component applied to the whole spectrum resulted in @xmath28 and a separate absorbed applied to just the power - law component resulted in an upper limit of @xmath29 .
the temperature of the plasma component was 0.24 ( 0.18 - 0.43 ) kev , and the abundance was 0.016 ( @xmath30 ) .
note that the bulge velocity dispersion in ngc 3998 is 304 km s@xmath4 @xcite , implying a virial temperature of @xmath31 kev . since ngc 3998 is an early - type galaxy and the inferred abundance from this fit is low , it is likely that this component is due to virial heating of the ism ( rather than starburst emission as is often observed in llagn ) .
the power - law slope in this case was 1.84 ( 1.84 - 1.89 ) , unchanged within the errors from the simple power - law fit .
the thermal component obviously contributes to the spectrum most significantly at the lowest energies , with a 0.3 - 0.7 kev flux of @xmath32 erg @xmath3 s@xmath4 , or @xmath33 of the total 0.3 - 0.7 kev flux .
however note that in this band the flux estimates of the 3 epic detectors differ by @xmath34 and therefore this component should be treated with caution . in the _ asca _ gis spectrum
there is a small excess of counts near 6.4 kev which results in a `` detection '' of an fe - k line .
however the addition of a narrow gaussian line ( @xmath35 kev ) to the _ asca _ power - law fit only reduces @xmath17 by 3.7 @xcite , which is only significant at a confidence of @xmath36 based on the f - test . in table
[ ewtab ] we list the upper - limits obtained for narrow line emission with the energy fixed at 6.4 kev .
not surprisingly , the limit obtained when _ _ asca _ _ data is included in the fit is _ higher _ than when the _ xmm - newton _ and _ bepposax _ data are fit alone . in this table
we list 90% errors for one interesting parameter ( @xmath37 ) , which are appropriate since here there is only one parameter of interest ( the normalization of the line ) , however we also give @xmath38 errors for comparison with 2-parameter errors in the literature ( or alternatively 97% 1-parameter errors ) .
we also attempted to fit a disk line model ( with the energy fixed at 6.4 kev , inclination fixed at 0@xmath39 , power - law emissivity slope fixed at -2 and the inner and outer radii fixed at 5 and 500 in schwarzschild units ) , and obtained upper - limits ( for @xmath40 = 2.7 ) of 33 ev and 42 ev , without and with the _ asca _ data included , respectively .
llllll _ asca _ & 8.0 ( 6.9 - 9.2 ) & 1.89 ( 1.86 - 1.92 ) & 0.80 & 310.9/281 & this work + _ bepposax _ & 5.4 ( 3.5 - 7.8 ) & 1.87 ( 1.83 - 1.90 ) & 1.2 & 222.8/227 & this work + _ bepposax _ , @xmath4110 kev & 5.1 ( 3.4 - 7.5 ) & 1.86 ( 1.82 - 1.90 ) & 1.1 & 207.9/210 & this work + _ xmm - newton _ & 3.3 ( 3.0 - 3.6 ) & 1.88 ( 1.87 - 1.90 ) & 1.1 & 1422.6/1338 & this work + _ xmm - newton _ + _ bepposax _ & 3.3 ( 3.0 - 3.6 ) & 1.88 ( 1.87 - 1.89 ) & 1.1 & 1652.5/1567 & this work + lll & @xmath41 32.7 & @xmath4141.4 + _ bepposax _ & @xmath4138.2 & @xmath4156.0 + _ asca _ & 85.5 ( 11.7 - 159 ) & @xmath41186 + _ xmm - newton _ + _ bepposax _ & @xmath4125 ev & @xmath41 33 ev + _ xmm - newton _ + _ bepposax _ + _ asca _ & @xmath4133 & @xmath4140 + the om detector exposed five 1000 s images in the u and uvw2 filters , for a net exposure time of 5000 s in each band ( additional exposures were taken of flanking fields which will not be discussed here ) .
the central regions of the u and uvw2 images are shown in figure [ omfig ] . in order to properly model the surface brightness distributions for the two filters we also required model psfs
( there were no stars sufficient for this purpose in our observations ) .
we initially produced model psfs using scisim .
however , the current version of scisim ( 3.0.0 ) does not specifically model the psf - dependence of the individual filters , and also only models the psf as a single gaussian .
nevertheless , the resultant psf fwhm was @xmath42 which is similar to that expected for the uvw2 filter and therefore we proceeded with this psf model for that filter . in the case of the u band filter we instead produced a model psf by generating double gaussian images and adjusting standard deviations of the two gaussians and their relative normalization until the encircled energy function ( i.e. , the integrated radial profile ) matched the values given in the om calibration file om_psf1drb_0006.ccf . in the surface brightness
fitting described below we model the nuclear point - source component with a gaussian whose standard deviation ( @xmath26 ) is permitted to be a free parameter . the fitted value of @xmath26 would then represent some combination of telescope jitter during the observation and also any adjustment to the psf due to the inadequacies of our psf model . for uvw2 data , the xmmsas tool `` omichain '' produces om point source lists based on a 6 aperture and background taken from an aperture with radii 7 - 12 .
the nuclear 2120 magnitude was 15.1 which corresponds to a flux density of @xmath43 erg @xmath3 s@xmath4@xmath44 ( based on a vega zero point and flux of 0.025 and @xmath45 erg @xmath3 s@xmath4@xmath44 ) .
the uvw2 images were fit with a double gaussian model ( with one gaussian intended to represent any diffuse flux ) from which it was determined that @xmath5 of the uv flux from ngc 3998 is extended over a @xmath46 ( fwhm ) scale ( the individual fits typically resulted in gaussian @xmath26 values of @xmath47 for the nuclear component and @xmath48 for the diffuse component ) .
our spatial analysis implies that the flux density from within a 2 aperture was @xmath49 erg @xmath3 s@xmath4@xmath44 .
similarly , integrating the gaussian fit for the nuclear component results in a count rate of @xmath50 counts s@xmath4 , which implies a flux density of @xmath51 erg @xmath3 s@xmath4@xmath44 ( using a count rate to flux conversion factor of @xmath52 erg @xmath3 s@xmath4@xmath44 counts@xmath4 s ) .
this result is in fairly good agreement with the values determined spectrally from iue data in @xcite , @xmath53 erg @xmath3 s@xmath4@xmath44 and @xmath54 erg @xmath3 s@xmath4@xmath44 for the unresolved and extended ( fwhm @xmath55 ) continuum components in the 1950 - 2155 region . given the uncertainties ( particularly in the psf model ) , we note that the actual nuclear flux density most likely lies in the range of @xmath56 erg @xmath3 s@xmath4@xmath44 .
we therefore assume a flux of @xmath57 erg @xmath3 s@xmath4@xmath44 with a 50% uncertainty . however , this value is a factor of @xmath58 lower than the foc lower limit ( @xmath59 erg @xmath3 s@xmath4@xmath44 ) determined by @xcite .
since the foc uv data was overexposed in that observation our analysis will be based on the om flux , and we note that it is possible that hst observed a flare in the uv bandpass .
the u band magnitude was determined from a 3 aperture ( using the xmmsas tool `` omsource '' which allowed for smaller source radii in the case of u band images ) .
the u band magnitude was 14.3 , which corresponds to a flux of @xmath60 erg @xmath3 s@xmath4@xmath44 ( with the vega zero - point and flux being 0.025 and @xmath61 erg @xmath3 s@xmath4@xmath44 flux ) .
the u band flux contains a contribution from the extra - nuclear emission but can of course nevertheless be considered to be an upper - limit to the nuclear u - band flux of ngc 3998 .
we fit the central images with a model that consisted of a gaussian ( again to represent a point source including jitter ) and a de vaucouleurs profile ( @xmath62 $ ] , where @xmath63 is the scale length of the profile .
these fits resulted in @xmath63 values of @xmath64 , although the precise values are not particularly meaningful given the uncertainty in the psf and the fact that we only fit the central 30 of the images since we are mainly interested in the determining the nuclear flux .
the gaussian @xmath26 values were @xmath65 , with the gaussian component contributing 40 - 60% of the flux within a radius of 3 . here
the count rate integrated from the nuclear gaussian component is @xmath66 counts s@xmath4 , which results in a flux density of @xmath67 erg @xmath3 s@xmath4@xmath44 .
the u band hst foc image was also overexposed , and @xcite report a lower - limit of @xmath68erg @xmath3 s@xmath4@xmath44 which is consistent with our flux .
the current flux calibration uncertainty for om u and uv magnitudes is @xmath69 ( u.s .
_ xmm - newton _ guest observer facility , private communication ) . to check the calibration of the u band magnitudes
, we computed the the ( total ) count rate of ngc 3998 and another galaxy in the u band fov , ngc 3990 , and derived u band magnitudes of 11.92 and 13.94 .
these values can be compared to the rc3 u band magnitudes of 12.13 and 13.81 , respectively , or differences of 0.21 and 0.13 .
this implies that the flux calibration is indeed accurate to at least @xmath69 . as with the uv flux , the nuclear u band flux
the uncertainty is dominated by the contribution of the extra - nuclear component , and therefore we assume a value of @xmath70 erg @xmath3 s@xmath4@xmath44 ( 50% of the aperture photometry value ) with a @xmath22 uncertainty ( i.e. , with 1@xmath26 being the offset to the model nuclear flux value discussed above ) . in order to supplement the om data with additional optical photometric points , we derived nuclear fluxes from the hst wfpc2 f547 m ( u2uh0505 t , 200 second exposure ) and f791w ( u2uh0507 t , 80 second exposure ) observations of ngc 3998 .
for both images the net count rate was computed within a 0.1 circular aperture ( with the background estimated from an annulus at 2.5 - 2.6 ) .
the count rates were converted to fluxes using the photflam fits keywords ( @xmath71 erg @xmath3 s@xmath4@xmath44 and @xmath72 erg @xmath3 s@xmath4@xmath44 for f791w and f547 m , respectively ) .
this resulted in flux densities of @xmath73 erg @xmath3 s@xmath4@xmath44 and @xmath74 erg @xmath3 s@xmath4@xmath44 at 7872 and 5484 , respectively .
we produced model psfs for these images using tinytim and fitted the central 3
( i.e. , a @xmath75 pixel image ) with a gaussian plus de vaucouleurs model .
this resulted in nuclear flux density estimates of @xmath76 erg @xmath3 s@xmath4@xmath44 and @xmath77 erg @xmath3 s@xmath4@xmath44 , which do not differ by more than @xmath69 from values obtained above from aperture photometry , and we assume an error of @xmath78 on these fluxes .
no statistically - significant short - term variability has been observed in the x - ray data , except for a marginally - significant , @xmath79 scale variation observed in the mecs data @xcite . in figure [ lcfig ]
we show the long - term 0.5 - 2.0 kev light curve of ngc 3998 .
these fluxes were derived from the _ asca _ , _ xmm - newton _ , and _ bepposax _ data discussed in this paper as well as archival rosat pspc and hri data ( in which we converted the count rate to the 0.5 - 2.0 kev flux using a power - law with a slope of 1.88 ) . the einstein data point from 1979
is from @xcite converted to the 0.5 - 2.0 kev bandpass .
we chose 0.5 - 2.0 kev bandpass since it is within the energy ranges of these detectors , and , as shown in
[ specsec ] , there is a negligible contribution from hot gas to the spectrum ( as would be expected since ngc 3998 is not a star - forming galaxy ) . from this plot , it appears that the flux of ngc 3998 was fairly constant during the measurements but may have been a factor of @xmath5 higher during the _ bepposax _ and _ xmm - newton _ observations .
the gina 2 - 10 kev flux for ngc 3998 was also high , by a factor of @xmath80 ( @xmath81 ) relative to the _ asca _ ( _ bepposax _ ) fluxes , although it should be kept in mind that ginga was not an imaging telescope and had a fov of @xmath82 @xcite .
note that x - ray binaries are not expected to be a significant contributor to the x - ray luminosity of ngc 3998 since typical normal galaxy luminosities tend to be @xmath83 erg s@xmath4@xcite , in contrast to @xmath11 erg s@xmath4for ngc 3998 .
more specifically , the b - band luminosity of ngc 3998 is @xmath84erg s@xmath4(based on @xmath85 from rc3 ) , which implies a 0.5 - 4.5 kev luminosity of @xmath86 erg s@xmath4using the correlation in @xcite .
ngc 3998 x-1 @xcite is an extra - nuclear x - ray source just outside of the rc3 @xmath87 ellipse of ngc 3998 . in figure [ pn_x-1_fig ]
we show the pn image of the central @xmath88 of ngc 3998 with the position of x-1 marked .
x-1 was detected with 441 , 73 , and 84 net counts by the pn , mos1 and mos2 detectors , respectively , in 27.5 regions .
background spectra were extracted from circular regions of the same size in a source - free region away from the ulx and ngc 3998 .
a power - law fit resulted in @xmath89 , @xmath90 and @xmath17/dof = 25.9/32 ( see figure [ ulx_pl ] ) .
the 0.5 - 2.0 kev and 2 - 10 kev fluxes ( averaged from the three detectors which differed from the mean by less than @xmath91 ) were @xmath92 erg @xmath3 s@xmath4and @xmath93 erg @xmath3 s@xmath4 .
if this source is associated with ngc 3998 then the implied luminosity is @xmath94erg s@xmath4 , making it an `` ultra - luminous x - ray source '' ( ulx ) . a multi - color disk ( mcd ) black body model resulted in an unacceptable fit , with @xmath95 , @xmath96 kev , and @xmath17/dof = 53.1/32 .
the observed x - ray spectrum is consistent with ulxs , often observed with power - law spectra ( e.g. , * ? ? ?
* ; * ? ? ?
however of course a power - law spectrum with @xmath18 is also consistent with a background agn .
the _ xmm - newton _ 2 - 10 kev logn - logs shown in @xcite implies that @xmath97 background sources @xmath98 would be expected at this flux level or higher .
x-1 is @xmath99 from the nucleus ngc 3998 , and so only @xmath100 background sources are expected at this flux or higher within this area , implying that it is unlikely that this is a background source . since we used a source - free region for the background spectra to avoid inaccurately subtracting flux due to ngc 3998 ( as would result from using a `` local '' background )
, however the spectra of the ulx spectra then contain some contamination from ngc 3998 . to quantify this
, we used our scisim simulation of the epic data and estimate that 0.19% of the flux of ngc 3998 is contaminating the x-1 source regions , or @xmath101erg @xmath3
s@xmath4= 24% of the observed flux of x-1 .
we also show in figure [ pn_x-1_fig ] the u band image . after adjusting the astrometry by applying a @xmath102 shift ( derived by aligning the first position of the nucleus of ngc 3998 and two usno - b stars ) , an additional usno - b source , 1454 - 0216445 , is coincident with a u band source and ngc 3998 x-1 .
the u band source is detected with a signal - to - noise ratio of 9.3 and a magnitude @xmath103 ( derived using the tool `` omsource '' ; note that here the statistical error on the u magnitude is larger than the calibration uncertainties ) .
the usno - b 2nd epoch ( 1974.9 ) r and b magnitudes are 20.0 and 21.6 .
the usno - b 1st epoch values are brighter by 0.5 and 0.7 magnitudes , respectively .
however the uncertainty of usno - b magnitudes is typically @xmath104 @xcite , and since these magnitudes are close to the flux limits of the survey ( v @xmath105 ) , their uncertainties are probably somewhat larger , particularly in the case of the b band magnitude .
therefore the difference between the two usno - b epochs is probably just due to statistical fluctuation . similarly
the unusual `` concave '' spectrum implied by the usno - b second epoch r and b magnitudes and the om u band magnitude is largely due to the questionable b band magnitude .
as discussed in @xcite , the x - ray spectrum of ngc 3998 is fit well in the 0.1 - 100 kev bandpass with a simple power - law model that is only moderately absorbed ( @xmath106 in contrast to the galactic value of @xmath107 ) .
we do not find any evidence for large - scale ( i.e. , @xmath108 ) variability on short ( @xmath109 day ) time scales but ngc 3998 may have varied in the 0.5 - 2.0 kev bandpass on long time scales .
our main goal is to assess the physics of the accretion in ngc 3998 , and below we compare these observed x - ray properties of ngc 3998 with those observed in other agn , and compare the sed of ngc 3998 with seyfert 1 and riaf models .
we also speculate on the constraints that our results can place on the geometry of the nucleus of ngc 3998 .
the presence of broad @xmath9 in ngc 3998 suggests that the nuclear region might be similar in structure to `` typical '' seyfert 1 galaxies .
the x - ray slope of ngc 3998 ( @xmath110 ) is consistent with typical seyfert 1 galaxies , and the 2 - 10 kev luminosity of ngc 3998 ( @xmath111 erg s@xmath4for @xmath112erg @xmath3 s@xmath4 ) is only marginally low for typical seyfert galaxies @xcite .
however , ngc 3998 differs in two important respects from typical seyfert 1 galaxies .
first , the lack of strong x - ray variability is in contrast with typical seyferts , particularly on short time scales , since the trend there is for variability to increase with decreasing luminosity @xcite .
@xcite suggested that this is typical of liners and llagn ( including ngc 3998 ) and that the x - ray source regions are larger in liners and llagn than in typical seyferts .
one possibility is that the x - ray production is occurring in optically - thin flows ( see below ) rather than flares in optically - thick accretion disks .
the x - ray spectrum is also remarkably featureless in contrast with typical seyfert 1 galaxies , particularly with the lack of any fe - k emission at 6.4 kev .
the upper - limit to the equivalent width ( ew ) in the _ xmm - newton _ and _ bepposax _ joint fit is 25 ev .
figure [ ewfig ] shows the distribution of narrow - line ews measured in a sample of seyfert 1 galaxies @xcite , where we also show the upper - limit for ngc 3998 .
the trend in seyfert 1s is to have higher ew fe - k lines with decreasing luminosity , and accordingly the lack of a line in ngc 3998 is conspicuous .
a cautionary point is that figure [ ewfig ] is based on seyfert 1s with fe - k lines and does not account for seyfert 1s lacking fe - k lines .
however , the vast majority of ( x - ray bright ) seyfert 1 galaxies appear to exhibit some level of fe - k emission @xcite , with very few seyfert 1s being known that lack a fe - k line when the spectra are of sufficient quality for fe - k emission to be detected ( t. yaqoob , priv . comm . ) .
thus , it is likely that the accretion physics and/or geometry in ngc 3998 differs somewhat from typical seyfert 1 galaxies . by integrating the best - fitting power - law to the _ xmm - newton _ and _ bepposax _ data ( @xmath113 with normalization @xmath114 ) we calculate the unabsorbed 0.1 - 100 flux to be @xmath115erg @xmath3 s@xmath4 , or a luminosity of @xmath116 erg s@xmath4 . using the bulge velocity dispersion of 304 km s@xmath4 from @xcite and
the bulge dispersion / black hole mass relation given in @xcite we estimate the black hole mass of ngc 3998 to be @xmath117 .
this implies an eddington luminosity ( @xmath118 ) of @xmath119 erg s@xmath4 , so @xmath120 ( note that we get a similar result if we use the typical llagn ratio of @xmath121 found by ho [ 1999 ] ) .
one possibility is that the low luminosities in ngc 3998 are reflecting low accretion rates .
recent accretion modeling suggests that at low accretion rates ( @xmath122 a few percent of the eddington rate ; see narayan , mahadevan , & quataert 1998 ) the flow will have low particle densities , and in turn may be optically thin and radiatively inefficient . in general these models
predict that the x - ray spectrum from an riaf is dominated by inverse - compton scattering ( of synchrotron emission ) and results in somewhat steep x - ray slopes ( as opposed to the flat slopes expected from bremsstrahlung emission , e.g. , see mahadevan 1997 ) . for example
, in the case of convection - dominated accretion flows ( cdafs ) , x - ray slopes of @xmath123 are expected when @xmath124 @xcite .
the x - ray temporal and spectral properties of ngc 3998 described above are similar to other liners thought to contain riafs ( e.g. , * ? ? ?
* ; * ? ? ?
* ) ( although again the x - ray spectral slope is consistent with typical seyfert 1s also ) .
we turn now to comparison of the sed of ngc 3998 to the predictions of various models .
the data used to derived the sed of ngc 3998 is listed in table [ sedtab ] and is shown in figure [ sedplot ] .
the x - ray data shown has been `` unfolded '' to remove the spectral response of the detectors and have been binned more so than in figure [ allplfig ] for display purposes .
the points marked with open circles can be considered to be upper - limits since they are sampling radii @xmath125 . in the near - ir
the large aperture points are clearly dominated by galactic emission ( and accordingly the small aperture points contain a large contribution from galactic light ) .
however in the radio to mid - ir region the radiation appears to be fairly compact ( i.e. , unresolved to aperture sizes less than @xmath126 , see also knapp et al .
also variability is evident among several of the radio and ( possibly also ) mid - ir measurements listed in table [ sedtab ] , which also implies that the emission is compact .
note that seds for ngc 3998 were also presented in @xcite and @xcite , however those papers assume the hst foc uv flux , which is a a factor of @xmath58 higher than our om measurement .
this higher uv flux results in a sed that resembles a typical agn , while our lower uv flux results in a sed more consistent with the sample of low - luminosity agn discussed in @xcite .
it is also evident that in the optical band , only the hst data with @xmath127 apertures are sufficient to exclude large contributions of extra - nuclear flux .
given the large low - mass star contribution to near - ir emission of early - type galaxies , this is also true for the near - ir band .
the relative amounts of flux in the om u band and hst foc u band suggests that this contribution is at least 90% , and likely to be much larger at longer wavelengths .
we assessed the impact of extinction using the extinction equations in @xcite .
@xcite found that the upper - limit for hi absorption to the nucleus of ngc 3998 was @xmath128 . including the galactic contribution of @xmath129 results in a conservative estimate of @xmath130 .
this value is consistent with the x - ray observations here with the exception of the _ asca _ data , however again we note that the calibration of the low energy response of the _ asca _ ccd detectors is somewhat uncertain .
this corresponds to @xmath131 @xcite , and from @xcite we derive that @xmath132 50% , 30% and 25% of the flux at uvw2 , u and b , respectively , would be absorbed .
this level of extinction does not impact our conclusions .
llll @xmath133 & @xmath134 & 60 & @xcite + @xmath135 & @xmath136 & 30 & @xcite + @xmath137 & @xmath138 & 24 & @xcite + @xmath137 & @xmath139 & 45 & @xcite + @xmath137 & @xmath140 & 5 & first + @xmath141 & @xmath142 & 6 & @xcite + @xmath143 & @xmath144 & 0.5 & @xcite + @xmath143 & @xmath145 & 0.008 & @xcite + @xmath143 & @xmath146 & 0.007 & @xcite + @xmath147 & @xmath148 & 6 & @xcite + @xmath149 & @xmath150 & 0.1 & @xcite + @xmath151 & @xmath152 & 180 & iras + @xmath153 & @xmath154 & 90 & iras + @xmath155 & @xmath156 & 45 & iras + @xmath157 & @xmath158 & 8 & @xcite + @xmath159 & @xmath160 & 45 & iras + @xmath161 & @xmath162 & 8 & @xcite + @xmath161 & @xmath163 & 6 & @xcite + @xmath161 & @xmath156 & 6 & @xcite + @xmath164 & @xmath165 & 5 & @xcite + @xmath166 & @xmath167 & 3 & @xcite + @xmath166 & @xmath168 & 5 & @xcite + @xmath166 & @xmath169 & 11 & @xcite + @xmath166 & @xmath170 & 15 & @xcite + @xmath166 & @xmath168 & 4 & 2mass point source catalog + @xmath171 & @xmath172 & 5 & @xcite + @xmath171 & @xmath173 & 11 & @xcite + @xmath171 & @xmath174 & 15 & @xcite + @xmath171 & @xmath175 & 4 & 2mass point source catalog + @xmath176 & @xmath173 & 3 & @xcite + @xmath176 & @xmath177 & 5 & @xcite + @xmath176 & @xmath178 & 11 & @xcite + @xmath176 & @xmath179 & 15 & @xcite + @xmath176 & @xmath180 & 4 & 2mass point source catalog + @xmath181 & @xmath182 & 0.1 & this work + @xmath183 & @xmath184 & 0.1 & this work + @xmath185 & @xmath186 & 0.1 & @xcite + @xmath187 & @xmath188 & 3 & this work + @xmath189 & @xmath190@xmath191 & 0.1 & @xcite + @xmath192 & @xmath193 & 2 & this work + @xmath194 & @xmath190@xmath195 & 0.1 & @xcite + for accretion onto a black hole , the gravitational in - fall timescale can be shorter than the radiative timescale ; the accretion energy is then stored as thermal energy instead of being radiated ( e.g. , rees et al .
1982 ; narayan & yi 1994 ) . in the past few years
, analytical and numerical work has shown that the physics of such riaf models is far more complex than originally anticipated .
in particular , very little mass available at large radii actually accretes into the black hole ; most of it is lost to an outflow ( i.e. , in an adiabatic inflow / outflow solution , or `` adios '' ) or circulates in convective motions ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the accretion of slightly rotating gas also may result in a substantially reduced accretion rate relative to the bondi rate@xcite .
following a proposal due to blandford & begelman ( 1999 ) , we can parameterize the radial variation of the gas density in the flow with a parameter @xmath196 , where @xmath197 ( equivalently , the accoretion rate varies with radius as @xmath198 ) ; @xmath199 , rather than @xmath200 ( as in the original analytical advection - dominated accretion flow [ adaf ] models of narayan & yi 1994 ) , is favored by the simulations , implying much lower gas densities close to the black hole . figure [ sedplot]a shows several riaf models for the sed of ngc 3998 .
@xcite showed that there are significant degeneracies in comparing theoretical riaf spectra to observed data so these results should be interpreted as representative models , not unique `` fits . '' the solid line in [ sedplot]a is an adaf with an accretion rate of @xmath201 ( hereafter accretion rates are given in eddington units ) and @xmath202 (= fraction of the accretion energy transfered to electrons ) .
the dotted line is the adaf model with an outer thin disk starting at a transition radius @xmath203 ( @xmath204 = the schwarzschild radius ) .
the dot - dashed line is an adios model with @xmath205 and @xmath206 and a similar accretion rate at the outer edge of the flow , also with an outer thin disk ( the so - called multi - color disk ( mcd ) model ) with @xmath207 .
note that adios models result in similar spectral models to cdaf models .
the thin disk ( for both models ) has an outer radius of @xmath208 , an inclination of 60 degrees , and @xmath209 ( hereafter all @xmath210 values are quoted in eddington units ) .
given theoretical uncertainties about electron heating , values of @xmath211 to @xmath212 are plausible ( e.g. , quataert & gruzinov 1999 ) , and the two models chosen in fact span this range .
these models fit the ir - x - ray emission reasonably well , except of course for the large contribution of stellar flux at near - ir wavelengths .
the ( nuclear ) ir emission is dominated by the ( putative ) thin disk component while the uv - x - rays are dominated by the riaf , and in fact @xmath213 and the accretion rate at @xmath213 were chosen to approximately match the ir flux .
the smaller @xmath213 value could have been chosen which would have resulted in a larger thin - disk flux , and correspondingly a smaller @xmath210 would have been chosen .
therefore the values of @xmath210 and @xmath213 are only constrained to within @xmath132 an order of magnitude . the x - ray emission in these riaf models
is dominated by inverse - compton scattering which results in approximately in a power - law continuum .
the adios model has a lower @xmath210 and density near the black hole , but a larger temperature .
in that situation , the compton peaks are more separated ( since the energy gain per scattering is proportional to temperature ) and the peaks are more distinct .
this is a generic difference between high @xmath210 adaf type models and lower @xmath210 models .
since the x - ray spectrum of ngc 3998 is fairly featureless lower temperature models are more consistent with the x - ray data .
note that in general different thermal line emission characteristics are predicted for these different types of models , but these lines are not detectable at this @xmath210 @xcite .
neither model accounts for the radio emission , in spite of the fairly large outer disk radius of @xmath208 .
the most restrictive constraint in this regard is the vlbi 5 ghz point which has a beam size of only 7 mas , or @xmath214 pc , and is clearly nuclear in origin . in figure
[ sedplot]b we also plot the median seds for radio - load and radio - quiet agn from @xcite , rescaled to the flux of ngc 3998 at @xmath215 hz .
the qso seds grossly overestimate the blue - uv emission , i.e. , ngc 3998 is lacking the `` big blue bump '' that is considered to be due to an optically - thick accretion disk .
however , this is not unexpected since the accretion rate is much lower than that of a typical qso , and mcd temperatures scale as @xmath216 @xcite .
we also show an approximate comptonized thin - disk model that fits the x - ray data from ngc 3998 without over - predicting optical or ir flux . this thermal comptonization model ( implemented in xspec as `` comptt '' ; see titarchuk 1994 , hua & titarchuk 1995 ) uses the wien approximation to black body flux for computational efficiency , however this should be a reasonable approximation . in this model
there are 5 parameters : the overall normalization , the temperature of the wien component , the temperature and optical depth ( @xmath217 ) of the hot plasma , and a geometry switch to select a spherical or disk geometry . in general inverse - compton spectra
tend to be power - laws with a slope of @xmath218 , where @xmath219 is the mean amplification of the photon energy per scattering @xcite .
therefore it is not surprising that for a given wien and plasma temperature combination , @xmath217 could be adjusted to result in our observed power - law slope and hence give a good fit .
the fits generally resulted in a power - law extrapolated from the peak of the wien component , and we therefore initially fixed the temperature of the wien spectrum at @xmath220 kev ( @xmath221 ) to see if the power - law could be extended to the optical bandpass .
we tried hot plasma temperatures of 100 and 200 kev ( see * ? ? ?
* ) , and fitted for the model normalization and @xmath217 as free parameters .
the best - fit optical depths were 0.3 and 0.1 for a disk geometry and plasma temperatures of 100 and 200 kev , respectively . with a spherical geometry
the best - fit optical depths were 0.5 and 1.0 for plasma temperatures of 100 and 200 kev .
clearly the value of @xmath217 is not meaningful since our data do not constrain the geometry or temperature of the hot plasma .
these fits ( simultaneously to the _ xmm - newton _ and _ bepposax _ data ) resulted in @xmath17/dof = 1653 - 1654/1567 , or basically equivalent to the pure power - law fit .
the fit with a plasma temperature of 200 kev and a disk geometry is shown in figure [ sedplot ] .
we computed a mcd spectrum ( with inner radius = @xmath222 and outer radius = @xmath208 ) consistent with the wien component by assuming an inclination angle of 60 degrees and adjusting @xmath210 , which resulted in @xmath223 .
such a model accounts for the spectrum of ngc 3998 from the optical to @xmath224 kev , i.e. , the optical flux is consistent with the extrapolation of the power - law observed in the x - ray band , particularly if we allow for variability in the foc f480lp flux ( the offset between the pds point in figure [ sedplot ] and the comptonized spectrum model is due in part to the fact that the x - ray data were unfolded assuming the simple power - law spectrum extending to 100 kev ) .
however this model requires that a _ negligible _ fraction of the emission longward of the optical is due to the agn in ngc 3998 .
this may be unlikely given the compactness of the radio and mid - ir emission .
of course there is some degeneracy between the inclination angle and accretion rate of the disk in this scenario , with an angle of 0@xmath39 and @xmath225 also being consistent with the wien component .
however the luminosity of the mcd disk component is not arbitrary .
the mcd spectrum is given by the sum of blackbody spectra with temperatures varying as @xmath226 where @xmath227 is radius of each point on the surface of the disk @xcite .
this results in a spectrum similar to a single blackbody emitting at the peak temperature of the disk , and for given black hole mass the peak temperature scales as @xmath216 .
of course the luminosity of a blackbody emitting over a given surface area is proportional to @xmath228 , where h@xmath229kt .
therefore we could not match the luminosity of the mcd component for other disk temperatures by simply adjusting the accretion rate since the expected luminosity of the wien component would move in the @xmath230 plane along a slope of 4 rather than 0.9 .
this makes the comptonized mcd model inconsistent with the observed x - ray power - law if the peak disk temperature is below @xmath231 kev ( i.e. , the wien component normalization would be larger than the mcd expectation ) , and also at higher temperatures this model is only consistent with the data if the mcd component is attenuated ( e.g. , by inclination effects ) . also note that the optical ( uv ) flux would not be due to the accretion disk for disk temperatures in excess of @xmath232 kev .
conversely , assuming that the optical and uv data are due to a thin disk requires that @xmath233 .
finally , note that a standard @xmath0 disk is both stable and optically thick when @xmath210 is in the range discussed here ( @xmath234 ) .
in addition to a very low accretion rate , another possibility is that both the radio and x - ray flux from ngc 3998 are due to a jet , as has been suggested recently for llagn such as ngc 4258 @xcite , ic 1459 @xcite and ngc 4594 @xcite . along these lines
we show a simple broken power - law model in figure [ sedplot ] , where we found that the radio data were consistent with a slope of @xmath235 ( i.e. , with @xmath236 ) and break frequency of @xmath237 hz .
this is obviously an overly simplistic model but here the radio slope is due to optically - thick synchrotron emission and the x - ray emission is due to optically - thin synchrotron emission , modeled by @xcite as being due to the outer , `` post - shock '' portion of the jet .
the markoff - falcke model also predicts a mid - ir `` bump '' due to synchrotron emission from the inner , `` pre - shock '' jet which may be consistent with the mid - ir fluxes observed in ngc 3998 .
alternatively , the synchrotron emission may break to a flatter slope and the x - ray is again also due to inverse - compton emission , and the mid - ir and/or optical - uv flux is due to a standard @xmath0 disk @xcite .
clearly these models would also be consistent with the sed of ngc 3998 and may imply accretion rates possibly orders - of - magnitude higher than the riaf and comptonized @xmath0-disk models since a large fraction of the accretion energy would be released kinetically in the jet .
however , as was the case in the sed modeling discussed above , any @xmath0 disk in this scenario would likely still need to be truncated to avoid over - predicting the optical and uv fluxes . also ,
the lack of any short - term , large - amplitude x - ray variability may argue against the x - ray emission being dominated by a ( relativistic ) jet .
the fe - k lines in seyfert 1 galaxies are assumed to originate from reflection from a thin disk and/or matter further out such as the putative `` torus '' in the standard model @xcite .
if the fe - k emission is dominated by a thin disk , then the maximal solid angle as seen from the x - ray source would be @xmath238 , and the solid angle of a toroidal geometry of matter most likely would not exceed this . if we assume that the ew @xmath224 ev fe - k line typically observed in seyfert 1 galaxies is produced by a reflection from optically - thick material with @xmath239 , then our upper - limit of @xmath240 ev implies that @xmath241 in ngc 3998 .
interestingly , @xcite report a similar result of @xmath242 for either a neutral or ionized compton reflection hump based on reflection model fits to the _ bepposax _ spectrum of ngc 3998 .
we also computed an upper - limit for a disk line model at 6.4 kev and obtained @xmath243 ev .
a @xmath244 erg s@xmath4 seyfert 1 would typically exhibit a ew = 300 ev disk line , which implies @xmath245 .
this suggests that the solid angle of _ any _ optically - thick material in ngc 3998 is less than at most 25% of that typically observed in seyfert 1 galaxies . we can roughly constrain the size of an outer thin disk using the line information . assuming a disk geometry with the x - ray source at a height @xmath246 above the center of the disk with inner and outer radii @xmath247 and @xmath248 , the solid angle subtended by the disk is given by @xmath249 $ ] .
the x - rays are produced in the inner part of the riaf , implying @xmath250 .
it is therefore likely that @xmath251 .
this implies that @xmath252 and therefore @xmath253 , or @xmath254 for @xmath255 .
this is consistent with figure [ sedplot ] where we assumed @xmath256 .
a possible caveat to this analysis is that ngc 3998 may contain an optically - thick disk with @xmath257 but whose surface layers are fully ionized .
this could result in little or no fe - k emission @xcite .
if this is source is associated with ngc 3998 , then the optical luminosity of the source ( @xmath258 -11 ) implies several to tens of supergiant stars or a cluster of hundreds to thousands of low - mass stars , if the optical flux is stellar .
if the optical variability observed between the two usno - b epochs ( separated by of order 25 years ) were real it would argue against this , although again the variability is probably not statistically significant . in addition , the high u - band absolute magnitude ( @xmath259 ) and very blue color imply that any such cluster would be composed of early - type stars .
this rules out a globular cluster with late - type stars which would be expected in a early type galaxy . using @xmath260 = log(@xmath261 ) + r/2.5 + 5.7 @xcite
, we derive @xmath262 for the ulx which is typical of agn found in the cdf - s @xcite and is much larger than values typical of stars ( @xmath263 ) and somewhat smaller than values typical of low - mass x - ray binaries ( @xmath264 ; bradt & mcclintock 1983 ; although realistically we would be observing an x - ray binary in a cluster ) .
the most likely identification for this source is therefore a background agn , although as stated in
[ ulxsec ] , the detection of a background source at this flux was unlikely ..
we have determined the tightest constraint to date on fe - k emission in ngc 3998 , specifically ew @xmath265 ev for a narrow line at 6.4 kev .
our analysis of _ xmm - newton _ data combined with archival _ asca _ and _ bepposax _ data shows that the spectrum is fit well with a simple power - law with @xmath1 and a small amount of intrinsic absorption .
the _ xmm - newton _ om uv flux measurement is consistent with an extrapolation of the x - ray power - law , suggesting that the uv to x - ray spectrum of ngc 3998 is consistent with a power - law over 5 decades in energy .
the sed of ngc 3998 is consistent with both riaf models and a comptonized thin - disk model ( i.e. , as expected to be found in seyfert galaxies ) , both of which can account for the uv to x - ray spectrum of ngc 3998 .
none of the accretion models account for the compact radio emission in ngc 3998 which is probably dominated by a jet .
the `` pure '' thin - disk model also does not account the mid - ir emission , although a truncated thin - disk surrounding an inner riaf fits the mid - ir to x - ray spectrum well ( with the near - ir to optical region dominated by extra - nuclear flux ) . if the accretion is dominated by an riaf then the transition radius is @xmath266 and the @xmath210 is @xmath267 ( at @xmath268 in the models with p @xmath190 0 ) . in either case
the inner radius of the thin disk is probably @xmath269 in order to account for the lack of x - ray reflection features .
if the accretion is dominated by a comptonized thin disk that extends to a radius of 3@xmath204 , then the accretion rate in the disk is in the range @xmath270 in eddington units and the lack of fe - k emission may be due to an ionized accretion disk .
finally , a jet model may be consistent with the radio and x - ray flux , and jet emission may also be responsible for the mid - ir and optical / uv flux .
this work was supported by nasa grants nag5 - 11378 and nag5 - 11572 .
y. t. is supported by the japan society for the promotion of science .
q. received support by nasa atp grant nag5 - 12043 and a sloan foundation fellowship .
we made use of the high energy archive and research center at nasa / gsfc and the nasa extragalactic database .
we thank the anonymous referee for useful comments that improved this paper .
alonso - herrero , a. , rieke , m. , rieke , g. , & shields , j. 2000 , , 530 , 688 awaki , h. , koyama , k. , kunieda , h. , takano , s. , tawara , y. , & ohashi , t. 1991 , , 366 , 88 ball , g. , narayan , r. & quataert , e. 2001 , , 552 , 221 ballantyne , d. & ross , r. 2002 , , 332 , 777 becker , r. , white , r. , & edward , a. 1991 , , 75 , 1 blandford , r. & begelman , m. 1999 , , 303 , l1 bradt , h. & mcclintock , j. 1983 , 21 , 13 burstein , d. , jones , c. , forman , w. , marston , a. , & marzke , r. 1997 , , 111 , 163 cardelli , j. , clayton , g. , & mathis , j. 1989 , , 345 , 245 condon , j. et al .
1998 , , 115 , 1693 condon , j. & dressel , l. 1978 , , 221 , 456 david , l. , jones , c. , & forman , w. 1992 , , 388 , 82 di matteo , t. , quatart , e. , allen , s. , narayan , r. & fabian , a. 2000 , , 311 , 507 done , c. & nayakshin , s. 2001 , , 328 , 616 elvis , m. et al .
1994 , , 95 , 1 fabbiano , g. , fassnacht , c. , & trinchieri g. 1994 , , 434 , 67 fabbiano , g. et al . 2003 , 588 , 175 filho , m. , barthel , p. & ho , l. c. 2002 , , 385 , 425 foschini , l. , et al .
2002 , , 392 , 817 frank , j. , king , a. , & raine , d. 2002 , accretion power in astrophysics , cambridge university press , cambridge frogel , j. , persson , s. , aaronson , m. & matthews , k. 1978 , , 220 , 75 hasinger , g. et al .
2001 , , 365 , l45 hawley , j. f. & balbus , s. a. , 2002 , apj , 573 , 738 heckman , t. 1980 , , 87 , 152 ho , l. c. , filippenko , a. , & sargent , w. 1997 , , 112 , 391 ho , l.c .
1999 , , 516 , 672 hornschemeier , a. et al .
2003 , , 126 , 575 hua , x .-
& titarchuk , l. 1995 , , 449 , 188 hummel , e. 1980 , , 41 , 151 hummel , e. , fanti , c. , parma , p. , & schilizzi , r. 1982 , , 114 , 400 hummel , e. , van der hurst , j. m. , & dickey , j. m. 1984 , , 134 , 207 igumenshchev , i. v. , & abramowicz , m. a. 1999 , mnras , 303 , 309 igumenshchev , i. v. , & abramowicz , m. a. 2000 , apjs , 130 , 463 jenkins , e. & savage , b. 1974 , , 187 , 243 knapp , g. , van driel , w. , & van woerden , h. , 1985 , , 142 , 1 longmore , a. & sharples , r. 1982 , , 201 , 111 mahadevan , r. 1997 , , 477 , 585 markoff , s. , falcke , h. & fender , r. 2001 , , 372 , l25 markoff , s. , nowak , m. , corbel , s. , fender , r. , & falcke , h. 2003 , , 397 , 645 mcelroy , d. 1995 , , 100 , 105 monet , d. et al .
2003 , , 125 , 984 mushotzky , r. , done , c. , & pounds , k. 1993 , , 31 , 717 murphy , e. , lockman , f. , laor , a. , & elvis , m. 1996 , , 105 , 369 nandra , k. , george , i. , mushotzky , r. , turner , t. & yaqoob , t. 1997a , , 476 , 70 nandra , k. , george , i. , mushotzky , r. , turner , t. & yaqoob , t. 1997b , , 477 , 602 nandra , k. , george , i. , mushotzky , r. , turner , t. & yaqoob , t. 1997c , , 488 , l91 narayan , r. , mahadevan , r. & qautaert , e. 1998 , narayan , r. & yi , i. 1994 , , 428 , l13 norman , c. et al .
2004 , , submitted pellegrini , s. , cappi , m. , bassani , l. , della ceca , r. , & palumbo , g. 2000 , , 360 , 878 pellegrini , s. et al .
2003 , , 585 , 677 perna , r. , raymond , j. & narayan , r. 2000 , , 541 , 898 pietrini , p. & krolik , j. 1995 , , 447 , 526 proga , d. & begelman , m. 2003 , , 592 , 767 ptak , a. , yaqoob , t. , mushotzky , r. , serlemitsos , p. , & griffiths , r. 1998 , , 501 , l37 ptak , a. , serlemitsos , p. , yaqoob , t. , & mushotzky , r. 1999 , , 120 , 179 ptak , a. & griffiths , r. 2002 , adass 2002 proceedings , in press , astro - ph/0303104 quataert , e. & narayan , r. 1999 , , 520 , 298 quataert , e. , di matteo t. , narayan , r. & ho , l. 1999 , , 525 , l89 quataert , e. & gruzinov , a. , 2000 , apj , 539 , 809 roberts , t. & warwick , r. 2000 , , 315 , 98 reichert , g. , branduardi - raymont , g. , filippenko , a. , mason , k. , puchnarewicz , e. , & wu , c .- c .
1992 , 387 , 536 rees , m. j. , begelman , m. c. , blandford , r. d. , & phinney , e. s. , 1982 , nature , 295 , 17 rengelink , r. et al .
1997 , astron .
astrophys . suppl . , 124 , 25 rybicki , g. & lightman , a. 1979 , radiative processes in astrophysics ( new york : john wiley & sons ) snowden , s. 2002 , from estec symposium new visions of the x - ray universe in the _ xmm - newton _ and _ chandra _ era , astro - ph/0203311 stone , j. m. , pringle , j. e. , & begelman , m. c. 1999 , mnras , 310 , 1002 strickland , d. et al .
2001 , , 560 , 707 szokoly , p. et al .
2004 , , submitted , astro - ph/0312324 terashima , y. , iyomoto , n. , ho , l. c. & ptak , a. 2002 , , 139 , 1 titarchuk , l. 1994 , , 434 , 570 tonry , j. et al .
2001 , , 546 , 681 tremaine , s. , et al .
2002 , , 574 , 740 white , r. & becker , r. 1992 , , 79 , 331 willner , s. , elvis , m. , fabbiano , g. , lawrence , a. & ward , m. 1985 , , 299 , 443 yaqoob , t. & padmanabhan , u. 2002 , from the proceedings of `` active galactic nuclei : from central engine to host galaxy '' , eds .
s. collin , f. combes , & i. shlosman , in press ( astro - ph/0211349 ) yuan , f. , markoff , s. , falcke , h. , & biermann , p. 2002
, , 391 , 139 | we present the results of a 10 ks _ xmm - newton _ observation of ngc 3998 , a `` type - i '' liner galaxy ( i.e. , with significant broad h@xmath0 emission ) .
our goal is to test the extent to which radiatively - inefficient accretion flow ( riaf ) models and/or scaled - down agn models are consistent with the observed properties of ngc 3998 .
a power - law fit to the _ xmm - newton _ spectra results in a power - law slope of @xmath1 and 2 - 10 kev flux of @xmath2 erg @xmath3 s@xmath4 , in excellent agreement with previous hard x - ray observations .
the om uv flux at 2120 appears to be marginally resolved , with @xmath5 of the flux extended beyond 2 .
the nuclear component of the 2120 flux is consistent with an extrapolation of the x - ray power - law , although @xmath5 of the flux may be absorbed .
the om u flux lies significantly above the x - ray power - law extrapolation and contains a significant contribution from extragalactic emission .
the upper - limit for narrow fe - k emission derived from the _ xmm - newton _ spectra is 33 ev ( for @xmath6 ) . the upper - limit for narrow fe - k emission derived from a combined fit of the _ xmm - newton _ and _ bepposax _ spectra is 25 ev , which is one of the strictest limits to date for any agn .
this significantly rules out fe - k emission as is expected to be observed in typical seyfert 1 galaxies .
the x - ray flux of ngc 3998 has not been observed to vary significantly ( at @xmath7 level ) within the x - ray observations , and only between observations at a level of @xmath5 , which is also in contrast to typical seyfert 1 galaxies .
the lack of _ any _ reflection features suggests that any optically - thick , geometrically - thin accretion disk must be truncated , probably at a radius of order 100 - 300 ( in schwarzschild units ) .
riaf models fit the uv to x - ray spectral energy distribution of ngc 3998 reasonably well . in these models
the mid - ir flux also constrains the emission from any outer thin disk component that might be present .
the uv to x - ray sed is also consistent with a comptonized thin disk with a very low accretion rate ( @xmath8 ) , in which case the lack of fe - k emission may be due to an ionized accretion disk .
accretion models in general do not account for the observed radio flux of ngc 3998 , and the radio flux may be due to a jet .
recent jet models may also be consistent with the nuclear fluxes of ngc 3998 in general , including the x - ray , optical / uv and mid - ir bands .
the ( ground - based ) near - ir to optical photometric data for the nuclear region of ngc 3998 contains large contributions from extra - nuclear emission .
we also derive nuclear fluxes using archival hst wfpc2 data , resulting in meaningful constraints to the nuclear sed of ngc 3998 in the optical band .
we discuss a possible om u band and usno - b detection of the ngc 3998 ulx . |
a _ density _ on the plane is a function weighting both perimeter and area , or , on a general manifold , a function weighting surface area and volume .
manifolds with density occur broadly in mathematics : from probability , where the classical example of gauss space ( @xmath3 with density @xmath5 ) plays an important role , to riemannian geometry , where they appear as quotient spaces of riemannian manifolds , and even to physics , where they describe spaces with different mediums .
they have recently been in the spotlight for their role in perelman s proof of the poincar conjecture .
( for general references on these topics , see @xcite or better ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
18 ) . )
we study the isoperimetric problem in planar sectors with certain densities . in this context
, the isoperimetric problem seeks to enclose prescribed ( weighted ) area with least ( weighted ) perimeter ( not counting the boundary of the sector ) .
solutions are known for very few surfaces with densities ( see sect . [
sec : isoperimetric - problems - on - manifolds - with - density ] below ) .
our major theorem after dahlberg _ et al .
_ characterizes isoperimetric curves in a @xmath6-sector with density @xmath0 , @xmath7 : ( [ thm : major theorem on mwd ] ) given @xmath7 , there exist @xmath8 such that in the @xmath6-sector with density @xmath0 , isoperimetric curves are ( see fig .
[ fig : possible minimizers for sector w / density r^p ] ) : \1 . for @xmath9 ,
circular arcs about the origin , \2 . for @xmath10 , undularies , \3 . for @xmath11 , semicircles through the origin .
we give bounds on @xmath12 and @xmath13 in terms of @xmath14 , and a conjecture on their exact values .
section [ sec : cgcc-1 ] gives further results on constant generalized curvature curves .
sectors with density @xmath0 are related to @xmath15 spaces ( see _ e.g. _ cor .
[ cor : l^p norm inequality ] ) , have vanishing generalized gauss curvature ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
5.1 ) , and have an interesting singularity at the origin where density vanishes . adams _
@xcite previously studied sectors with gaussian density . in theorems [ thm :
thetalessthanpi ] , [ thm : pi < thetanot < api ] , and [ thm : api < thetanot ] , we provide a relatively minor result after ca@xmath16ete _ et al . _
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
3.20 ) characterizing isoperimetric curves in a @xmath6-sector with density @xmath1 inside the unit disk and density @xmath2 outside the unit disk .
an interesting property of this problem is that it deals with a noncontinuous density .
there are five different kinds of isoperimetric regions depending on @xmath6 , @xmath17 , and the prescribed area , shown in figure [ fig : isoperimetric - sets - for - thetanot - sector ] . in section [ sec : r^n w/ radial ]
we discuss basic results in @xmath3 with radial density .
we use a simple averaging technique of carroll _ et al .
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
4.3 ) _ to give a short proof of a result of betta _ et al . _
( ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
4.3 ) and our thm .
[ thm : bettaetal - r^n ] ) and to solve the isoperimetric problem in @xmath3 with density @xmath4 ( thm .
[ pro : hypersphere - minimize - p_lt_-n ] ) . due to length concerns , there are several points where we have chosen to omit or shorten arguments that are either standard in the field or similar to arguments given in earlier works . among these
are : the existence results of theorems [ the : existence of minimizers in manifolds with area density satisfying certain properties ] , the ball - density theorems of section [ sec : ball density ] , the results on constant generalized curvature curves of section [ sec : cgcc-1 ] , and the numeric results referred to in subsection [ sub : the - sector w / density r^p ] and seen in figure [ fig : transition - from - circular arc to semi - circle-1 ] . for a more detailed account of our work , including full proofs and expanded discussions of these topics ,
please see the 2009 small geometry group report @xcite . in the plane with density @xmath0 , carroll
_ et al_. ( * ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
? * ; * ? ? ?
4 ) prove that for @xmath18 , isoperimetric curves are circles about the origin bounding area on the outside and prove that for @xmath19 , isoperimetric regions do not exist .
both of these results generalize easily to sectors of arbitrary angle ( prop . [
pro : non - existence - pin-2,0 ] and the remark after theorem [ thm : major theorem on mwd ] ) .
et al_. prove that for @xmath7 , isoperimetric curves are circles through the origin . by a simple symmetry argument ( prop .
[ pro : cone - sector equivalence for iso . regions ] ) , isoperimetric circles about the origin and circles through the origin in the plane correspond to isoperimetric circular arcs about the origin and semicircles through the origin in a @xmath20-sector . in this paper
, we consider @xmath6-sectors for general @xmath21
. for @xmath22 , existence in the @xmath6-sector follows from standard compactness arguments ( prop .
[ prop : in the circular cone , isoperimetric exist for ( inf,-2)u(o , inf ) ] ) .
lemma [ lem : minimizers are circles , semicircles , or unduloids ] limits the possibilities to circular arcs about the origin , semicircles through the origin , and undularies ( nonconstant positive polar graphs with constant generalized curvature ) .
proposition [ pro : if circle minimizes , it minimizes for all theta less ] shows that if the circle is not uniquely isoperimetric for some angle @xmath6 , it is not isoperimetric for all @xmath23 .
proposition [ pro : the - isoperimetric - ratio is increasing for theta_0
< theta_2 , constant after , semicircle uniquely ] shows that if the semicircle is isoperimetric for some angle @xmath6 , it is uniquely isoperimetric for all @xmath23 .
therefore , transitional angles @xmath24 exist .
isoperimetric regions that depend on sector angle have been seen before , as in the characterization by lopez and baker ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
6.1 , fig .
10 ) of perimeter - minimizing double bubbles in the euclidean cone of varying angles , which is equivalent to the euclidean sector .
theorem [ thm : major theorem on mwd ] provides estimates on the values of @xmath12 and @xmath13 .
we conjecture ( conj .
[ con : conjecture on major thm for mwd ] ) that @xmath25 and @xmath26 .
proposition [ pro : circles are stable until pi / root p+1 ] proves that the circle about the origin has positive second variation for all @xmath27 , and proposition [ pro : semi - circles do nt minimize before pi/2*(p+2)/(p+1 ) ] proves the semicircle through the origin is not isoperimetric for all @xmath28 .
using the characterization of constant generalized curvature curves in section [ sec : cgcc-1 ] , we wrote a computer program to predict isoperimetric curves in sectors with density @xmath0 , and the results of this program also support our conjecture as can be seen for @xmath29 in figure [ fig : transition - from - circular arc to semi - circle-1 ] . an easy symmetry argument ( prop .
[ pro : cone - sector equivalence for iso .
regions ] ) shows that the isoperimetric problem in the @xmath6-sector is equivalent to the isoperimetric problem in the @xmath30-cone , a cone over @xmath31 .
we further note that the isoperimetric problem in the cone over @xmath31 with density @xmath0 is equivalent to the isoperimetric problem in the cone over the product of @xmath31 with a @xmath14-dimensional manifold @xmath32 among regions symmetric under a group of isometries acting transitively on @xmath32 .
this provides a classical interpretation of the problem , which we use in proposition [ pro : when p=00003d1 , circles minimize until theta=00003d2 ] to obtain an improved bound for @xmath12 in the @xmath29 case by taking @xmath32 to be a rectangular two - torus .
theorem [ thm : polygon - r^p ] classifies isoperimetric regions of small area in planar polygons with density @xmath0 .
proposition [ pro : maintheoremplaneperimeterdensity ] through corollary [ cor : l^p norm inequality ] reinterpret theorem [ thm : major theorem on mwd ] and conjecture [ con : conjecture on major thm for mwd ] in terms of the plane with differing area and perimeter densities and in terms of analytic inequalities .
ccc [ cols="^ " , ] + [ thm : classification of cgcc]in the plane with density @xmath0 , a curve with constant generalized curvature @xmath33 normal at @xmath34 is ( see fig . [
fig : possible - types - of cgcc ] ) : 1 . for @xmath35 , a periodic undulary , 2 . for @xmath36 $ ] , a periodic nodoid , 3 . for @xmath37 , a circle through the origin , 4 . for @xmath38 , a circle about the origin , 5 . for @xmath39 ,
the geodesic @xmath40 which asymptotically approaches the radial line @xmath41 .
these results follow from the differential equations for constant generalized curvature curves .
two points deserve special mention .
first , reflection and scaling provides a correspondence between curves outside the circle and curves inside the circle .
second , to prove a curve @xmath42 which is not the circle , semicircle or geodesic is periodic , it suffices to show @xmath42 has a critical point after @xmath34 .
it follows directly from the differential equation that there is a radius at which @xmath42 would attain a critical point . to prove @xmath42 actually attains this radius , we note that otherwise @xmath42 would have to asymptotically spiral to this radius .
this implies @xmath42 has the same generalized curvature as a circle of this radius , which the equations confirm can not happen .
isoperimetric curves in the @xmath6-sector are polar graphs with constant generalized curvature that intersect the boundary perpendicularly . given these restrictions and the classification of theorem [ thm : classification of cgcc ] , we can give a sufficient condition for conjecture [ con : conjecture on major thm for mwd ] on the value of the transitional angles .
[ pro : if period of cgcc increasing , then ...
-1]in the plane with density @xmath0 , assume the half period of constant generalized curvature curves normal at @xmath34 with generalized curvature @xmath43 is bounded below by @xmath44 and above by @xmath45
. then the conclusions of conjecture [ con : conjecture on major thm for mwd ] hold .
given the correspondence mentioned in the proof of theorem [ thm : classification of cgcc ] , it suffices to only check curves with @xmath43 . by proposition [ lem : minimizers are circles , semicircles , or unduloids ] , isoperimetric curves are either circles about the origin , semicircles through the origin , or undularies .
an undulary is only in equilibrium when the sector angle is equal to its half period . if there are no undularies with half period less than @xmath44 or greater than @xmath45 , the only possible isoperimetric curves outside of that range are the circle and the semicircle .
the proposition follows by the bounds given in theorem [ thm : major theorem on mwd ] .
[ rem : cgcc remark]we now discuss some results on the periods of constant generalized curvature curves , and provide numeric evidence . using the second order constant generalized curvature equation
, one can prove that curves with generalized curvature near that of the circle have periods near @xmath44 .
similarly , using the first integral of the constant generalized curvature equation for @xmath46 , we see that the half period of a curve with constant generalized curvature is given by@xmath47 where @xmath48 is the curve s maximum radius . using this formula
, we generated numeric evidence for the bounds given in proposition [ pro : if period of cgcc increasing , then ...
-1 ] , as seen in figure [ fig : numerical - evidence - for period of cgcc-1 ] . this plot is given for @xmath49 , and is representative of all @xmath14 .
moreover , the integral seems to be monotonic in @xmath48 which would imply that every undulary with @xmath50 is isoperimetric for exactly one @xmath6-sector .
francisco l@xmath51pez has suggested studying the integral above with the techniques of complex analysis .
this mathematica plot of undulary half period @xmath52 as a function of maximum radius @xmath48 gives strong evidence for our main conjecture [ con : conjecture on major thm for mwd ] ( see prop .
[ pro : if period of cgcc increasing , then ...
in this section we classify the isoperimetric curves in the @xmath6-sector with density 1 outside the unit disk @xmath53 centered at the origin and @xmath1 inside @xmath53 .
ca@xmath16ete _ et al . _
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
3.3 ) consider this problem in the plane , which is equivalent to the @xmath20-sector .
proposition [ pro : anyarea - isosets - are ] gives the potential isoperimetric candidates .
theorems [ thm : thetalessthanpi ] , [ thm : pi < thetanot < api ] , [ thm : api < thetanot ] classify the isoperimetric curves for every area and sector angle . a _ bite _ is an arc of @xmath54 and another internal arc ( inside @xmath53 ) , the angle between them equal to @xmath55 ( see fig . [
fig : isoperimetric - sets - for - thetanot - sector](c ) ) .
@xmath56(a)isoperimetric sets in sectors with disk density : ( a ) an arc about the origin inside or outside @xmath53 , ( b ) an annulus inside @xmath53 , ( c ) a bite , ( d ) a semicircle on the edge disjoint from the interior of @xmath53 , ( e ) a semicircle centered on the x - axis enclosing @xmath53 for @xmath57.,title="fig : " ] isoperimetric sets in sectors with disk density : ( a ) an arc about the origin inside or outside @xmath53 , ( b ) an annulus inside @xmath53 , ( c ) a bite , ( d ) a semicircle on the edge disjoint from the interior of @xmath53 , ( e ) a semicircle centered on the x - axis enclosing @xmath53 for @xmath57.,title="fig : " ] @xmath56(b)isoperimetric sets in sectors with disk density : ( a ) an arc about the origin inside or outside @xmath53 , ( b ) an annulus inside @xmath53 , ( c ) a bite , ( d ) a semicircle on the edge disjoint from the interior of @xmath53 , ( e ) a semicircle centered on the x - axis enclosing @xmath53 for @xmath57.,title="fig : " ] ( c)isoperimetric sets in sectors with disk density : ( a ) an arc about the origin inside or outside @xmath53 , ( b ) an annulus inside @xmath53 , ( c ) a bite , ( d ) a semicircle on the edge disjoint from the interior of @xmath53 , ( e ) a semicircle centered on the x - axis enclosing @xmath53 for @xmath57.,title="fig : " ] @xmath56(d)isoperimetric sets in sectors with disk density : ( a ) an arc about the origin inside or outside @xmath53 , ( b ) an annulus inside @xmath53 , ( c ) a bite , ( d ) a semicircle on the edge disjoint from the interior of @xmath53 , ( e ) a semicircle centered on the x - axis enclosing @xmath53 for @xmath57.,title="fig : " ] ( e ) isoperimetric sets in sectors with disk density : ( a ) an arc about the origin inside or outside @xmath53 , ( b ) an annulus inside @xmath53 , ( c ) a bite , ( d ) a semicircle on the edge disjoint from the interior of @xmath53 , ( e ) a semicircle centered on the x - axis enclosing @xmath53 for @xmath57.,title="fig : " ] [ pro : anyarea - isosets - are]in the @xmath6-sector with density @xmath1 inside the unit disk @xmath53 and @xmath2 outside , for area @xmath58 , an isoperimetric set is one of the following ( see fig . [
fig : isoperimetric - sets - for - thetanot - sector ] ) : \(a ) an arc about the origin inside or outside @xmath53 ; \(b ) an annulus inside @xmath53 with @xmath54 as a boundary ; \(c ) a bite ; \(d ) a semicircle on the edge of the sector disjoint from the interior of @xmath53 ; \(e ) a semicircle centered on the x - axis enclosing @xmath53 for @xmath57 .
any component of an isoperimetric curve has to meet the sector edge perpendicularly since any component that meets the boundary at a different angle contradicts proposition [ des : proposition - isoperimetric curves meet boundary perpendicular-1 ] , and any component that does not meet the sector edge can be rotated about the origin ( which will still give an isoperimetric region ) until it hits the sector edge at which point , since by proposition [ pro : isoperimetric exist in unit ball problem ] the curve is smooth , the intersection will be tangent so it will then contradict proposition [ des : proposition - isoperimetric curves meet boundary perpendicular-1 ] .
since each part of the boundary has to have constant generalized curvature it must be made up of circular arcs ( here we mean any circular arcs , not necessarily arcs of circles about the origin as earlier in the paper ) .
we can discard the possibility of combinations of circular arcs within the same density since one circular arc is better than @xmath59 circular arcs ( just as in the proof of lemma [ lem : minimizers are circles , semicircles , or unduloids ] ) . therefore , there are five possibles cases : @xmath60 1 .
a circular arc from one boundary edge to itself ( including possibly the origin ) .
+ there are three possibilities according to whether the semicircle has 0 , 1 , or 2 endpoints inside the interior of @xmath53 .
a semicircle with two endpoints inside @xmath53 has an isoperimetric ratio of @xmath61 .
for a semicircle with one endpoint inside @xmath53 ( fig .
[ fig : semicircle - meeting - db - perpendicular ] ) , by proposition [ pro : isoperimetric exist in unit ball problem ] snell s law holds . then the only possible curve that intersects the boundary
normally would also have to intersect @xmath53 normally .
its perimeter and area satisfy : + a semicircle meeting @xmath54 perpendicularly has more perimeter than a semicircle disjoint from the interior of @xmath53 ( fig .
[ fig : isoperimetric - sets - for - thetanot - sector]d ) . ] + @xmath62 @xmath63 @xmath64 + therefore a semicircle ( d ) outside @xmath53 with isoperimetric ratio @xmath65 is the only possibility .
+ @xmath60 2 .
a circular arc from one boundary edge to another .
+ an arc ( a ) or annulus ( b ) about the origin are the only possibilities for any @xmath66 or @xmath67 . at @xmath57 and
area @xmath68 a semicircle ( e ) centered on the x - axis enclosing @xmath53 is equivalent to a semicircle centered at the origin . for @xmath69 and @xmath68 , a semicircle tangent to @xmath54 together with the rest of @xmath54 ( fig .
[ fig : combinations of arcs ] ) is in equilibrium , but we will show that it is not isoperimetric .
its perimeter and area satisfy : + @xmath70 + a semicircle tangent to @xmath54 together with the rest of @xmath54 is never isoperimetric . ] + comparing it with an arc ( a ) about the origin we see that it is not isoperimetric .
+ @xmath60 3 . two circular arcs ( c ) meeting along @xmath54 according to snell s law ( prop .
[ pro : isoperimetric exist in unit ball problem ] ) .
+ @xmath60 4 .
three or more circular arcs meeting along @xmath54 .
+ by caete _
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
3.19 ) this is never isoperimetric .
+ @xmath60 5 .
infinitely many circular arcs meeting along @xmath54 .
+ by caete _
_ ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
3.19 ) this is never isoperimetric .
[ thm : thetalessthanpi]for some @xmath71 , in the @xmath6-sector with density @xmath1 inside the unit disk @xmath53 and 1 outside , for @xmath72 , there exists @xmath73 , such that an isoperimetric curve for area @xmath74 is ( see fig . [
fig : isoperimetric - sets - for - thetanot - sector ] ) : \(1 ) if @xmath75 , an arc about the origin if @xmath76 , semicircles on the edge disjoint from the interior of @xmath53 if @xmath77 , and both if @xmath78 ; \(2 ) if @xmath79 , both type ( 1 ) and ( 3 ) ; \(3 ) if @xmath80 a bite , if @xmath81 an annulus inside @xmath53 , and if @xmath82 both ; if @xmath83 , @xmath84 and the annulus is never isoperimetric ; \(4 ) if @xmath85 , an arc about the origin ; at @xmath57 any semicircle centered on the x - axis enclosing @xmath53 .
[ thm : pi < thetanot < api]in the @xmath6-sector with density @xmath1 inside the unit disk @xmath53 and 1 outside , for @xmath86 , there exists @xmath87 , @xmath88 , such that an isoperimetric curve for area @xmath74 is ( see fig . [
fig : isoperimetric - sets - for - thetanot - sector ] ) : \(1 ) if @xmath75 , a semicircle on the edge disjoint from the interior of @xmath53 ; \(2 ) if @xmath79 , both type ( 1 ) and ( 3 ) ; \(3 ) if @xmath89 , a bite ; \(4 ) if @xmath90 , an arc about the origin ; \(5 ) if @xmath91 , both type ( 4 ) and ( 6 ) ; \(6 ) if @xmath92 , a semicircle on the edge disjoint from the interior of @xmath53 .
[ thm : api <
thetanot]in the @xmath6-sector with density @xmath1 inside the unit disk @xmath53 and 1 outside , for @xmath93 , the isoperimetric curves for area @xmath74 are semicircles on the edge disjoint from the interior of @xmath53 ( fig .
[ fig : isoperimetric - sets - for - thetanot - sector]e ) .
for @xmath94 an arc about the origin is isoperimetric , while for @xmath95 a semicircle in the edge is isoperimetric . for area infinitesimally less than @xmath96 , for @xmath97 an annulus inside @xmath53 is isoperimetric , while for @xmath98 a bite is isoperimetric . ]
given proposition [ pro : anyarea - isosets - are ] , most of the proofs of theorems [ thm : thetalessthanpi ] , [ thm : pi < thetanot < api ] , [ thm : api < thetanot ] are by simple perimeter comparison , which we omit here .
figure [ fig : transition angle in function of density-1 ] shows numerically the transition between different isoperimetric curves .
the hardest comparison is handled by the following sample lemma .
[ lem : once bite wins , it always does]in the @xmath6-sector with density @xmath1 inside the unit disk @xmath53 and 1 outside , for a fixed angle and area @xmath99 , once a bite encloses more area with less perimeter than an annulus inside @xmath53 , it always does .
increasing the difference @xmath100 , it will be sufficient to prove that a bite is better than a scaling of a smaller bite because an annulus changes by scaling .
therefore , there will be at most one transition .
a bite is better than a scaling of a smaller bite , and hence once better than an annulus , it is always better . ] to prove that a bite does better than a scaling of a smaller bite we show that it has less perimeter and more area .
take a smaller bite and scale it by @xmath101 ( see fig .
[ fig : bite better than scaling bite ] ) and eliminate the perimeter outside @xmath53 .
another bite is created with perimeter @xmath102 and area @xmath103 : @xmath104 thus it is better than scaling a smaller bite .
in this section we look at the isoperimetric problem in @xmath3 with a radial density .
we use symmetrization to show that if an isoperimetric region exists , then an isoperimetric region of revolution exists ( lem . [ lem : isoperimetric regions bounded by surface of revolution ] ) , thus reducing the problem to a planar problem ( lem .
[ lem : r^n w / density f(r ) equivalent to 1/2 plane with density y^(n-2)f(r ) ] ) .
we also consider the specific case of @xmath3 with density @xmath0 .
we first give a nonexistence result ( prop .
[ pro : minimizers do nt exist for -n < p<0 in r^n ] ) for @xmath105 .
the methods used by carroll _ et al .
_ to study planar densities @xmath106 ( * ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
4.3 ) can be used to recover an earlier result of betta _ et al . _ on perimeter densities in @xmath3 ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
2.1 , 4.2 , 4.3 ) , and we provide a proof along these lines in theorem [ thm : bettaetal - r^n ] .
we then adapt this proof in proposition [ pro : hypersphere - minimize - p_lt_-n ] to show that hyperspheres about the origin bounding volume away from the origin are uniquely isoperimetric in @xmath3 with density @xmath107 .
we conclude by conjecturing that hyperspheres through the origin are isoperimetric in @xmath3 with density @xmath108 .
[ con : r^n with density r^p conjecture ] ) .
we d like to thank bruno volzone and alexander kolesnikov for bringing the paper by betta
_ to our attention , as well as robin walters and frank morgan for providing initial versions of our averaging proof of theorem [ thm : bettaetal - r^n ]
. using spherical symmetrization ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
9.2 ) generalized with density @xcite , we can replace the intersection of the isoperimetric region with each sphere about the origin by a polar cap of the same volume .
[ lem : r^n w / density f(r ) equivalent to 1/2 plane with density y^(n-2)f(r)]the isoperimetric problem in @xmath3 with density @xmath109 is equivalent to the isoperimetric problem in the half plane @xmath110 with density @xmath111 . up to a constant
, this is the quotient space of @xmath3 with density @xmath109 modulo rotations about an axis . by lemma [ lem : isoperimetric regions bounded by surface of revolution ]
, there exists an isoperimetric region bounded by surfaces of revolution , so a symmetric isoperimetric region in @xmath3 corresponds to an isoperimetric region in the quotient space and vice - versa .
isoperimetric curves in the plane with density @xmath0 are circles about the origin when @xmath18 , do not exist when @xmath19 , and are circles through the origin when @xmath7 ( ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
4.2 , 4.3 ) , ) .
we now look at the isoperimetric problem in @xmath3 with density @xmath0 . consider a sphere @xmath112 of radius @xmath113 not containing the origin .
let @xmath114 @xmath115 be the minimum and maximum values of @xmath116 attained on @xmath112 , note that @xmath117 .
let @xmath118 @xmath119 be the volume and perimeter respectively of @xmath112 .
we have that:@xmath120 working with a fixed @xmath121 , we get that for any sphere of radius @xmath122 with volume @xmath121 , @xmath123 where @xmath124 and @xmath125 .
so , if we fix @xmath121 large enough so that @xmath126 , we obtain @xmath127 for any @xmath128 . looking at @xmath119 we have @xmath129 which can be made as small as we want by taking @xmath122 large , and so since there are spheres of arbitrarily large radius not containing the origin and having volume @xmath121 , we see that there are regions of arbitrarily small perimeter enclosing this fixed volume @xmath121 .
then , by scaling these regions , we obtain the result for all volumes . a proof that applies for @xmath130
was given by ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
their next proposition , ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* prop 4.3 ) , which extends immediately to sectors , is by proposition [ pro : power change of coordinates ] equivalent to the @xmath0 case of a theorem of betta _ et al . _
* ; * ? ? ?
* ; * ? ? ?
4.3 ) . here
we use the averaging technique of carroll _ et al . _ to provide a short proof of the whole theorem of betta _ et al . _ [ thm : bettaetal - r^n]suppose @xmath17 is a non - negative non - decreasing smooth function on the positive real axis with @xmath131 for @xmath132 and with convex
. then hyperspheres about the origin are uniquely isoperimetric in @xmath3 with euclidean volume and surface area density @xmath17 . as in betta _
_ we can assume @xmath133=0 , since if we let @xmath134 then @xmath17-weighted surface area is just the sum of @xmath135-weighted surface area and a constant times euclidean surface area , and so if hyperspheres are isoperimetric for both these densities and uniquely isoperimetric for one then they will also be uniquely isoperimetric for @xmath17 .
let @xmath136 be a region bounded by rectifiable hypersurfaces .
we will show that a ball about the origin @xmath137 of the same volume has less surface area than @xmath136 with equality only when @xmath136 is also a ball .
in fact , we will prove a stronger statement : note that the surface area of @xmath136 is greater than or equal to the tangential surface area of @xmath136 , i.e. the component of surface area perpendicular to radial lines , with equality only when the boundary of @xmath136 consists only of hypersurfaces .
we will show that @xmath137 has less tangential surface area than @xmath136 .
@xmath138 and the tangential surface area element is @xmath139 where @xmath140 is the element of surface area on the unit sphere . under the change of coordinates @xmath141 , the volume element becomes @xmath142 and the tangential surface area element becomes@xmath143 where @xmath144 .
now , in the @xmath145-variable we obtain @xmath146 where @xmath145 is signed according to the radial orientation of the boundary ( positive if pointing away from the origin , negative if pointing towards the origin ) .
let @xmath147 be the sign - weighted sum of all the @xmath145 values of points in the intersection of @xmath148 and a radial ray in the direction of @xmath149 .
note that this is not defined in directions where @xmath148 is radial , but this is a set of @xmath150-dimensional lebesgue measure @xmath151 in the @xmath149-space so it will not effect our computations .
we then have that@xmath152 where the latter integral is taken over all possible directions .
now , let @xmath153 be the average value of @xmath154 over all directions .
then in these coordinates we see that @xmath137 , the ball about the origin with the same volume as @xmath136 , is actually the ball about the origin of radius @xmath153 : @xmath155 now , we get the first inequality coming from @xmath157 increasing , and the second inequality from @xmath157 convex ( note that the middle integral should in general be taken only over the directions where the intersection with @xmath148 is non - empty , but in this case since @xmath157 is zero at the origin we can in fact take it over all directions ) .
so , we see that @xmath137 has tangential surface area less than or equal to that of @xmath136 , and since all of the surface area of @xmath137 is counted in the tangential surface area ( @xmath137 is a ball about the origin ) , we conclude that @xmath137 has surface area less than or equal to that of @xmath136 with equality possible only if all of the surface area of @xmath136 is also tangential , that is if @xmath148 consists only of hyperspheres about the origin .
but if @xmath148 consists of multiple hyperspheres about the origin then the first inequality in [ eq : ineqchainbettaproof ] must be strict , and thus we conclude that equality holds if and only if @xmath158 .
the proof of theorem [ thm : bettaetal - r^n ] also gives this result in the @xmath6-sector and even more generally in any cone over a subset of @xmath163 , but we may have to change the convexity condition : if hyperspheres about the origin are not isoperimetric in such a space with euclidean surface area , then we can not perform the trick of subtracting off @xmath133 , and thus we will want to impose the stronger condition instead that @xmath164 is convex .
we note that if we take the cone over a convex subset of @xmath163 then in the euclidean case hyperspheres about the origin will be isoperimetric ( ( * ? ? ?
* thm 1.1 ) , see also ( * ? ? ?
after thm . 10.6 ) ) and therefore the original convexity condition will suffice .
the proof of proposition [ pro : hypersphere - minimize - p_lt_-n ] goes through in any of these spaces with no new conditions .
pedrosa , renato ; ritor , manuel .
isoperimetric domains in the riemannian product of a circle with a simply connected space form and applications to free boundary problems , indiana u. math .
j. 48 ( 1999 ) , 13571394 . | we consider the isoperimetric problem in planar sectors with density @xmath0 , and with density @xmath1 inside the unit disk and @xmath2 outside .
we characterize solutions as a function of sector angle .
we also solve the isoperimetric problem in @xmath3 with density @xmath4 .
alexander daz , nate harman , sean howe , david thompson |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.